No Max Pain, No Max Gain: A Case of

Predictable Reversal

ILIAS FILIPPOU PEDRO A. GARCIA-ARES FERNANDO ZAPATERO

September 30, 2024

Abstract

The Max Pain Theory, popular among retail investors, postulates that at options ex-

piration the price of the underlying stock converges to the strike price at which the

aggregate total payment obligations of the short positions is the lowest. Based on this, it

is possible to construct long-short portfolios that offer excess risk-adjusted performance.

We further show that this price predictability is mostly the result of a reversal affecting

stock prices that had fallen substantially during the recent period. We cannot rule out

that stock price manipulation is a factor that explains why the timing of the start of this

reversal regularly overlaps with the options expiration. Consistent with this conjecture,

the effect is stronger among small, illiquid stocks.

Keywords: Max Pain, options expiration, return predictability, reversal, price manipulation.

JEL Classification: G12, G13, G14, G23.

We would like to thank Julio Crego, Benjamin Golez (discussant), Chris Jones, Dmitriy Muravyev,
Neil D. Pearson (discussant), Miguel de Jesus (discussant), Paola Pederzoli (discussant), Francisco Perez,
Aurelio Vasquez and seminar participants at the 2024 MFA Annual Meeting, 2023 EFA Annual Meeting, 30th
Finance Forum, ITAM finance conference, Boston University and ITAM for useful comments. The paper
was a recipient of the best paper award at the 30th Finance Forum. Filippou: Florida State University,
ifilippou@[Link]; Garcia-Ares: Mendoza School of Business, University of Notre Dame, pgarciaa@[Link]
and Instituto Tecnológico Autónomo de México (ITAM), [Link]@[Link]; Zapatero: Questrom School
of Business, Boston University, fzapa@[Link].
1 Introduction

We test the Max Pain theory and show that it is possible to design long-short strategies

based on an intuitive measure that generates large and statistically significant returns and

alphas. The Max Pain theory postulates that as the expiration date of options approaches,

the prices of the underlying stocks converge to the maximum pain price, the strike price

at which the total payoff of all the options written on that stock and with that expiration

is the lowest. The Max Pain hypothesis is based on the investors’ assumption that there

is a price point at option expiration which will “hurt" the aggregate of options buyers the

most and, therefore, the aggregate of the options writers the least. It is not clear how or

when the Max Pain theory originates, but it is a topic of discussion among retail investors

in social media. In this paper we first want to establish whether actual price dynamics are

in any way consistent with this “theory", and then whether the predictability implied by it

permits profitable trading strategies.

As a first step, we first identify the Max Pain strike price corresponding to each stock for

a particular expiration date. We proceed as follows. Suppose that all the short positions

in both calls and puts written on a stock that expire on a particular date were held by a

single market participant. For each strike price we compute the cost for that hypothetical

single holder of all short positions if the stock price finished at that strike price at maturity.

Some options (puts and/or calls) would be out of the money, some would be in the money.

The strike price that yields the lowest cost for this hypothetical agent is the Max Pain strike

price. This Max Pain price will change as new positions in options on that stock are created,

or as the open interest in existing positions changes. Next, we introduce our measure of

Max Pain for each stock and day. In particular, we define it as the Max Pain strike price on

a particular day minus the stock price on that day, divided by the stock price. Obviously, it

can have positive or negative sign.

1
 To examine whether there is any truth in this conjecture, we design a Max Pain strategy

in the equities market. In particular, on the second Friday of each month -typically one

week before the options expiration- we allocate stocks into deciles according to the Max

Pain measure, so that decile 1 includes the stocks with the lowest (possibly negative) value,

and decile 10 stocks with the highest (likely a positive number). We construct a zero-cost

portfolio that buys high Max Pain stocks (decile 10) as, according to the Max Pain theory,

their prices are supposed to go up, and sells low Max Pain stocks (decile 1) as they are

expected to go down. We hold our portfolio until the expiration of the attached options, in

general, the third Friday of the month. We find a positive and statistically significant return

of the spread for both equally-weighted and value-weighted portfolios. Very importantly for

our analysis of possible explanations, we observe that the abnormal return comes mostly

from the stocks in decile 10, that is, those stocks that according to the Max Pain theory are

expected to increase in price as the date of the options expiration approaches, and indeed

we verify that is the case. The stocks of decile 1 also behave as predicted by the Max Pain

theory, that is, they drop in price, but the effect is economically not very large.

We confirm our results in Fama and MacBeth (1973) regressions: We regress the cross-

section of stock returns on the Max Pain measure as well as on other drivers of stock

returns such as price, size, stock volume, institutional ownership, book-to-market ratio,

debt-to-assets, stock reversals, momentum, stock illiquidity, and idiosyncratic volatility. We

find that the coefficients of Max Pain remain highly positive and statistically significant. We

find similar results if we use abnormal returns as in Daniel, Grinblatt, Titman, and Wermers

(1997) (DGTW).

When we restrict the regression analysis to a subset of stocks, particularly those charac-

terized as small, illiquid, growth-oriented, or listed on the NASDAQ exchange –all these

categories overlap extensively– our results are even more compelling. That is, the relation-

ship between the Max Pain measure and stock returns is stronger. As it is obvious –and has

2
been documented in the literature– these stocks are easier to manipulate, which could be a

factor in the price dynamics, as we explain later.

We also analyze strategies with different formation and holding periods within the

option’s life. Our baseline is one week prior to expiration. We also form decile portfolios

two weeks, three weeks, and four weeks prior to expiration, and we hold them until the

expiration of the options. We find that the strategies formed three and four weeks to

expiration offer economically positive but not statistically significant returns and CAPM or

FF3 alphas. In sum, the abnormal returns of the Max Pain strategy mostly accrue in the

week leading up to the expiration of the options.

In our quest to explain what is driving this pattern, we study the price history of the

stocks in the deciles of our long-short strategy, decile 1, stocks whose prices are expected

to drop –and actually drop, although not by a large quantity–, and decile 10, stocks that

the Max Pain theory predicts will go up in price before the expiration of the options, and

the data corroborates that is the case, and in an amount substantial enough to justify the

abnormal returns of our long-short strategy. The results are striking: stocks in decile 1 have

experienced on average a large increase in prices in the weeks previous to our portfolio

formation. Then they experience a slight drop in price right before the options expire, and

further increases –but of a very limited extent– thereafter. Regarding the stocks in decile 10,

in the weeks prior to our portfolio formation, they have experienced a large drop in prices,

then the week of the options expiration their prices go up and recover some of the lost

value, a pattern that continues during the subsequent weeks. In sum, what the stock prices

experience during the week leading up to the options expiration, especially the stocks of

decile 10, is a price reversal. In principle, this is not surprising, price reversals, especially

after big drops in price, are common in financial markets and have been amply documented

and analyzed in the literature for decades –for example, Jegadeesh and Titman (1995).

What is striking is that the reversal takes place, precisely and consistently, during the week

before the expiration of the options.

3
 To shed further light on the security dynamics around the options expiration, we

compute abnormal volume and order imbalances for all the stocks, sorted on Max Pain,

one week before expiration. We find significantly higher abnormal volume for low and

high Max Pain stocks -deciles 1 and 10, respectively. Regarding abnormal imbalances, for

which we show that investors are net buyers of high Max Pain stocks –expected to go up in

price, according to the Max Pain theory– and net sellers of low Max Pain stocks –expected

to go down– during the third week of the month.

In sum, it is possible that the timing of the price reversal –that starts the week right

before the expiration of the options– might be due, at least in part, to price manipulation,

that will benefit the holders of the short positions. Market participants are expecting the

reversal, therefore it is not surprising that investors will buy -especially– the stocks that

are expected to increase in price. The evidence in support of this conjecture –or at least

consistent with it– is that the effect is much stronger in small and illiquid stocks, which are

easier to manipulate, and the fact that the effect is stronger in decile 10, in which prices

are expected to go up –it is easier to force the price of a stock to go up than to go down.

Given the persistence of the price patterns we just described, a natural question is

whether sophisticated market investors are aware of these opportunities. To respond to

that question, we study the pattern of option trading right the days before expiration of the

options, excluding the options that are about to expire. We confirm that conjecture and find

that investors are net buyers of call options and net sellers of put options written on the

stocks in decile 10 of our spread portfolio –stocks that are expected to go up. As we have

explained, stocks in decile 1 behave in the opposite way, but in weaker fashion. We observe

that investors are net sellers of both call and put options, but with statistically significant

higher activity in the sale of call options, indicating a negative sentiment, consistent with

the expected drop in price of these stocks. In intermediate deciles we observe that investors

tend to sell options, but at a much lower intensity than in decile 10 for puts and decile 1

4
for calls. Overall, sophisticated investors are aware of this trading opportunity and take

advantage of it.

Throughout our analysis, we are cognizant of the fact that trading activity increases in

the days approaching option expiration, often related to the rebalancing and unwinding of

the hedges of market makers. A substantial amount of literature documents these patterns

and tries to explain them. We review them and study whether they are possible reasons

behind the Max Pain effect we find. We first show that Max Pain differs from the clustering

effect discussed by Ni, Pearson, and Poteshman (2005). As an initial examination, we

calculate the absolute distance between Friday’s closing price and the strike price closest

to the Friday’s closing stock price for all stocks in our sample. We find that the largest

distances correspond to stocks belonging to the extreme Max Pain portfolios, suggesting

that clustering does not drive our results. Additionally, we find that the profitability of

our strategy is not primarily derived from stock returns on the expiration Friday, where

the clustering effect is more pronounced. We also run a contemporaneous cross-sectional

regression of stock returns on Max Pain and on different measures of clustering, like in Ni

et al. (2005). We find that the Max pain strategy is significant after controlling for these

variables. Therefore, our results appear to be unrelated to pinning effects on expiration

dates. Finally, an alternative explanation is that stock price changes during expiration week

are at least in part the result of the rebalancing and unwinding of option market makers’

delta hedges away from expiration. We check this effect and regress stock returns of our

spread portfolio on the gamma exposure of the market makers. We find that aggregate

gamma imbalances do not provide an explanatory factor of our findings.

In the last part of our paper, we perform a large number of robustness tests and they all

support unambiguously the results we just described: The Max Pain theory is validated by

stock price dynamics. These dynamics are not the result of the price patterns documented

in the literature and seem to be the result of a reversal anticipated by sophisticated market

participants. The timing of the reversal, that starts the week before options expiration is

5
consistent with price manipulation that does not affect final prices, just the timing of the

reversal’s term structure.

The rest of the paper is organized as follows: in section 2, we lay out the fundamentals

of the Max Pain theory. In section 3 we describe the data and portfolio construction. Section

4 describes the empirical results. Section 5 considers potential explanations of the stock

predictability we document. Section 6 provides a broad range of robustness and other

specification tests. Section 7 concludes.

2 Fundamentals of Max Pain Theory and its Analysis

Given a stock and a maturity date on options written on this stock, Max Pain price is the

price of the stock at which the total payoff of all the options written on that stock with

that maturity date will be the lowest on said maturity date. In other words, the aggregate

liability of all the short positions will be the lowest possible, so that the holders of the

long positions will experience on aggregate the “maximum pain." The Max Pain Theory

conjectures that the market price of the stock converges toward the Max Pain price as the

expiration day approaches.

The origin of the Max Pain Theory is not clear, but is a frequent topic of discussion in

online chat rooms focused on finance –as WallStreetBets in Reddit. If the conjecture were

correct it would provide a straightforward investment strategy based on the stock price

predictability –we would explore later in the paper possible explanations of this empirical

pattern.

The objective of this paper is first to study whether there is any truth to this popular

conjecture among retail investors and, if that is the case, analyze the results of the associated

investment strategy. Lastly, we will explore possible explanations behind this pattern. With

this agenda in mind, we first define the Max Pain measure for a given stock as the difference

between the Max Pain price of the stock and its current price.

6
 From the fact that standard options expire on the third Friday of the month, that is, at

the end of the third week of the month, we design our first test of the Max Pain Theory and

compute whether the Max Pain measure explains abnormal returns during the third week

of the month –in other tests we consider other time periods.

As stock prices of large companies tend to be more efficient, we also explore whether

the Max Pain theory is more apparent for stocks of smaller companies –that tend to be more

illiquid, and often are growth stocks. Furthermore, we check if the stock price predictability

increases as the options approach maturity.

In the same spirit, and in our quest to explore the reasons for the predictability of

prices, we study order imbalances, both their timing and their sign, which can offer further

evidence of the validity of the theory, as well as shed some light on its foundations.

Finally, if the Max Pain theory is correct, we should only observe the stock price pre-

dictability it postulates in stocks with options, and it should not show in similar stocks but

without options.

In sum, our hypotheses first postulate the price convergence predicted by the Max Pain

theory. As we shall see, our empirical tests support very strongly the predictions of Max Pain

theory. Next we study the trading strategy implied by the price predictability implicit in

the Max Pain theory. Finally, we explore possible explanations behind this price dynamics,

especially the possibility that the holders of short positions in options have, on average,

superior information.

7
3 Data and Portfolio Construction

3.1 Stock and Option Data

Stock Data. Our daily and monthly stock returns and daily trading volume are from the

Center for Research in Security Prices (CRSP), including the New York Stock Exchange

(NYSE), American Stock Exchange (AMEX), and NASDAQ markets, as well as common

stocks with share codes 10 and 11, financials and non-financials. The data spans the period

from January 1996 to December 2021. We use accounting information for these stocks

from Compustat.

Option Data. We collect options data from the OptionMetrics IvyDB database and for all

exchange-listed options on U.S equities. We focus on the end-of-day bid and ask quotes,

trading volume, open interest, strike prices, deltas, gammas, and implied volatility. The

data spans the period from January 1996 to December 2021.

As it is standard in the empirical literature, we eliminate observations that do not satisfy

several criteria to guard against tradability concerns. First, to avoid illiquidity concerns,

we remove options with no option interest, options for which the ask price is lower than

the bid price, the bid price is equal to zero, or the average of the bid and ask quote is less

than 0.125 dollars. Second, to remove the effect of the early exercise premium in American

options, we exclude options whose underlying stock pays a dividend during the remaining

life of the option. Finally, we exclude all options that violate arbitrage restrictions. We

impose these filters only when we compute our Max Pain measure and create our portfolios

in order to avoid any forward-looking bias.

Stock Imbalances. We collect buy and sell trades from TAQ. The trades are signed

following the Lee and Ready (1991) procedure. The data spans the period from January

1996 to December 2021. We define the order imbalance of stock i at time t as OI Bi,t =

8
(Bu yi,t − Sell i,t )/(Bu yi,t + Sell i,t ) where Bu yi,t (Sell i,t ) denotes the dollar value of buy

(sell) orders of stock i at time t.

Signed Option Volume. We also collect daily option buy and sell volume from four

exchanges: NASDAQ GEMX (GEMX), NASDAQ International Security Exchange (ISE),

NASDAQ Options Market (NOTO), and NASDAQ PHLX (PHOTO). Specifically, the dataset

includes option volume of open buy, open sell, close buy and close sell orders. Each

category includes different types of participants: broker/dealer, proprietary firms, and

public customers. Then, we sum the buy and sell trades to compute the long and short open

interest at the participant level, and the order imbalances. The four exchanges account for

a significant portion of the options market. We do not consider other exchanges and OTC

markets. The datasets cover different time periods. Specifically, ISE covers the period of

May 2005 to December 2021, GEMINI is from August 2013 to December 2021, NOTO is

from November 2011 to December 2021, and PHOTO is from January 2009 to December

2021.

Max Pain Strike Price. The key element of the Max Pain theory is the strike price associ-

ated with the minimum aggregate liability of the holders of short positions. To retrieve it,

for a particular stock we collect all the strike prices of the attached options with a given

maturity date. Then, for each available strike price, we compute the total payment due by

the aggregate of all call and put writers assuming that the stock price at expiration is equal

to that strike price.1 The Max Pain price is the minimum such aggregate liability across all

available strike prices for that stock and maturity date.

Max Pain measure. We define the Max Pain measure as the price return implied by the

Max Pain strategy. Specifically, the Max Pain measure (our sorting variable) is the difference
1
We take into account that open interest from Optionmetrics is lagged by one day after November 28,
2000. Prior to this date, the open interest is not lagged.

9
between the Max Pain strike price computed as explained before for options expiring the

third Friday of the month, and the stock price as of the second Friday of the month, divided

by this stock price (e.g., M ax Pain = (X M ax Pain −S 2nd F r ida y )/S 2nd F r ida y ). The second Friday

of the month is the day at which we form our portfolios. For simplicity, we will refer to

this measure just as Max Pain. We provide a detailed example of the computation of the

measure in Section 2 of the Internet Appendix.

Figure 1 offers two examples of Max Pains. The top left panel shows the potential

payoffs (in millions) of put and call options of GameStop Corp. (GME) for different strike

prices on March 12, 2021. The top right panel shows the total potential liability. We find

that the strike price, associated with the minimum liability (e.g., Max Pain strike price), is

$150, and the stock price on the second Friday of the month (formation period) is $264.5.

Thus, the Max pain return for GME is (150 − 264.5)/264.5 = −43.29%. GME stock price

is supposed to go down by more than 43% in order to minimize liabilities for the option

sellers.

We also show an example of positive Max Pain in the bottom panel for Moderna. The

Max Pain return is (69.5 − 59.5)/59.5 = 17.75%. According to the Max Pain theory,

Moderna stock price is expected to go up by 18% to minimize the losses of the option

sellers.

[Figure 1 about here.]

Institutional Ownership. We collect end-of-quarter institutional stock holdings from the

13F form filed with the SEC (Thomson Reuters Institutional, included in WRDS). The data

spans the period from January 1996 to September 2021.

10
3.2 Max Pain Portfolios

We form decile portfolios of stocks sorted on the Max Pain measure, which we remind, is

computed one week before the expiration of the options, typically the third Friday of the

month. We order them from low Max Pain (decile 1) to high Max Pain (decile 10). In

particular, we form portfolios on the second Friday of each month and hold them for one

week, until the expiration date of the attached options. We offer results for equally-weighted

and value-weighted portfolios. We construct a zero-cost portfolio that buys stocks with

high Max Pain and sells stocks with low Max Pain.

In Figure 2, we present an example of the construction of our portfolios in May 2012.

Specifically, we build our portfolio on Friday 11th of May, 2012, and hold it until the 18th

of May, which refers to the option’s last trading day.

[Figure 2 about here.]

4 Empirical Results

4.1 Summary Statistics

We first present time-series averages of median characteristics of stocks sorted into deciles

based on Max Pain. Panel A of Table 1 shows time-series averages of the median Max Pain

of decile portfolios of call options sorted based on Max Pain every month, as well as the

spread in Max Pain between the extreme portfolios. We find that the median Max Pain is

-0.081 in the lowest decile and 0.076 in the highest decile rendering a difference of 0.157,

which is statically significant.

Panel B of table 1 reports additional characteristics of these portfolios. In particular,

we find that high Max Pain portfolios tend to have lower institutional ownership (IOR),

debt-to-assets (D/A), reversals (previous month return), momentum, idiosyncratic volatility,

11
stock price, and size, in comparison to low Max Pain portfolios. At the same time, we find

that the high Max Pain portfolios exhibit higher book-to-market ratios and illiquidity. We

do not observe statistical differences between low and high Max Pain portfolios in NASDAQ

membership, trading volume, idiosyncratic volatility and debt-to-assets ratios.

Finally, we note that the absolute difference between closing prices (the closing price

the third Friday of the month) and the nearest strike prices is the lowest for the middle

portfolios (P5, P6 and P7) compared to stocks with high and low Max Pain. This implies that

the pinning effect discussed in the literature is more pronounced for the stocks that belong

to these portfolios than to our high and low Max Pain portfolios. We examine pinning (or

clustering) effects in more detail later on, when we explore a possible explanation of our

findings.

[Table 1 about here.]

4.2 Univariate Sorts

We allocate stocks into portfolios on the second Friday of each month according to their

Max Pain measure and hold the portfolios until the expiration of the underlying options,

typically the third Friday of the month. We repeat this exercise every month. Stock returns

are calculated using the midpoint of the bid and ask prices at market close adjusted for

stock splits using the CRSP Cumulative Factor to Adjust Price (cfacpr). Table 2 presents our

main empirical result, that we will analyze further in additional tests.

Panel A of table 2 reports returns of equally-weighted and value-weighted decile port-

folios. We find that the return of the spread equally-weighted portfolio is positive and

statistically significant (0.4% in a week), consistent with the spirit of the Max Pain theory.

Notably, the return comes mostly from the highest Max Pain portfolio, that is, the portfolio

of securities whose price on the second Friday of the month is the furthest below the Max

Pain strike price (strike price at which the aggregate of the short positions will receive the

12
lowest payoff). Results for value-weighted portfolios are similar. Panel B reports alphas of

the spread portfolios risk-adjusted by the factors of the CAPM and FF3 asset pricing models.

In all cases, the strategy offers positive and statistically significant alphas. This is the first

piece of evidence that validates the Max Pain theory. Furthermore, it suggests a trading

strategy that delivers a positive alpha.

[Table 2 about here.]

Next, we analyze the performance of the max pain strategy throughout the holding

period. Figure 3 represents the daily cumulative returns of the spread max pain portfolios.

In this graph, the cumulative returns start at time 0, the point of portfolio formation, and

go until day 20. For instance, day 1 corresponds to the Monday of the expiration week.

The strategy reaches its highest cumulative performance Thursday of the expiration week,

after which it experiences a reversal that is complete 20 days after portfolio formation.

[Figure 3 about here.]

4.3 Cross-Sectional Regressions

In the previous subsection, we have studied the cross-sectional predictive ability of Max Pain

without taking into consideration other factors that are known to have explanatory power.

To rule out that we are not capturing any of those factors with our Max Pain measure, we

perform a cross-sectional regression that projects stock returns at time t + 1 on several

predictor variables at time t, including Max Pain. Table 3 reports the average coefficient of

this regression and the corresponding average adjusted R-squared. The set of predictor

variables (e.g., Cont r ols) includes stock illiquidity, price, size, book to market, debt to

assets, stock volume (SVOL), institutional ownership (IOR), idiosyncratic volatility (IVOL),

the stock return the second week of the month (REV Weekl y ), the stock return the previous

month (REV M onthl y ), and momentum. We show results for the entire sample in the first

13
column, but also for small, illiquid, and Nasdaq stocks in other columns. Specifically, we

focus on the model below:

Ret i,t+1 = α + β M ax Paini,t + γC ont r olsi,t + "i,t+1 (1)

where Ret i,t+1 represents the return of stock i at time t + 1. Table 3 shows that Max Pain is

a strong positive predictor of the cross-section of stock returns even after controlling for

other drivers of stock returns.

Previous literature demonstrates that stock return predictability is not equally likely

across all stocks and it is more likely among NASDAQ stocks, growth stocks, illiquid

stocks, and small stocks. In line with this, we find that our Max Pain coefficients are more

economically and statistically significant for small stocks, illiquid stocks, NASDAQ stocks,

and growth stocks.2

[Table 3 about here.]

4.4 Different Holding and Formation Periods

Next, we consider different formation and holding periods. In our main analysis, we

buy/sell stocks a week before expiration and hold the positions until expiration. Table 4

shows average returns for different holding and formation periods: We form portfolios two,

three, or four weeks before expiration and hold them until expiration in all cases. We do

this for equally-weighted and value-weighted portfolios. Panel A of Table 4 shows their

average returns. In any case, we find that strategies that consider alternative formation and

holding periods within the life of the options offer economically positive but not statistically

significant Max Pain spread portfolios. For example, the return of the spread portfolio
2
We sort all stocks in our sample in quintiles and we run the cross-sectional regressions for all the stocks
that belong to the smallest quintile in terms of size and book-to-market and the highest quintile for stock
illiquidity.

14
that we hold for two weeks is 40 bps for the full two weeks, which is the return our

primary strategy of Table 2 achieves. Therefore, the profitability of the Max Pain strategy is

generated in the expiration week.

[Table 4 about here.]

4.5 Trading Activity during the Holding Period

To further understand the dynamics of the previous results, we turn our attention to trading

of the stocks involved in the Max Pain strategy. We explore to different dimensions of

trading.

Abnormal Volume. Panel A of Table 5 reports time-series averages of abnormal trading

volume of stocks sorted into deciles based on Max Pain. We form portfolios on the first,

second, and third Friday of the month. We compute abnormal volume as:

Abnor malVolumew,m = Volumew,m − Volumem−1→m−n , (2)

where Volumew represents the average daily trading volume (expressed in millions) over

the week w of month m, prior to the expiration of the option and Volumem−1→m−n is the

average daily volume over the months m − n to m − 1 where n = 1, 3, 6, 12. We find that

the abnormal stock trading volume is concentrated in low and high Max Pain portfolios.

These portfolios contain stocks that are supposed to go down or up, respectively, according

to the Max Pain theory. In particular, the difference between the time-series average of P10

and P1 is positive and not statistically significant, but the differences between portfolios

P10 and P5 and P1 and P5 are large, positive and also significant. The results are similar if

we compute the abnormal volume compared to the previous three-, six-, or twelve-month

average stock trading volume.

15
Abnormal Order Imbalances. Panel B of Table 5 reports time-series averages of abnormal

order imbalances of stocks sorted into deciles according to Max Pain. We form portfolios on

the first, second, and third Friday of the month. We compute abnormal order imbalance as:

Abnor malOI Bw,m = OI Bw,m − OI Bm−1→m−n , (3)

where OI Bw represents the average daily order imbalance over the week w of month m

prior to the expiration of the option and OI Bm−1→m−n is the average daily order imbalance

over the months m − n to m − 1 where n = 1, 3, 6, 12. We find that investors are net buyers

of high Max Pain portfolios –expected to go up, according to the Max Pain theory– and net

sellers of low Max Pain portfolios –expected to go down. The difference in order imbalances

between high and low Max Pain portfolios is statistically significant. We find similar results

for abnormal order imbalances compared with average imbalances of the previous three,

six, and twelve months.

[Table 5 about here.]

4.6 Placebo Test

Our previous results show that the returns of the Max Pain strategy are more significant

the closer we are to the expiration of the options. Here we consider the following placebo

test. We form a portfolio with two weeks to expiration (instead of one week to expiration)

and hold the portfolio for one week instead of waiting until expiration of the options.

Panel A of Table 6 reports average stock returns of portfolios sorted on Max Pain two

weeks before the expiration of the attached options. We show results for equally-weighted

and value-weighted portfolios. W e find that a strategy lagged by one week (in comparison

to the baseline Max Pain strategy) offers spread portfolios that are neither economically

nor statistically significant. In Panel B we show that the corresponding alphas are not

statistically significant. This finding justifies our next test.

16
 [Table 6 about here.]

4.7 Weekly Performance before and after the Formation Period

The strong evidence that the Max Pain dynamics take place primarily the week of options

expiration leads us to ask whether these stocks display any specific pattern preceding that

week and/or subsequently. For that reason, we next study the history of the stocks in the

decile portfolios before and after the critical week before options expiration, during which

the Max Pain strategy yields abnormal risk-adjusted returns.

That history is represented in Figure 4 for stocks with low and high Max Pain (deciles

1 and 10). The time point 0 is the second Friday of the month, when we form the decile

portfolios of the Max Pain strategy. Then we compute the returns of these decile portfolios

in weeks -3 to -1 and in weeks 1 to 4, and include them in the plot. We observe that in the

weeks before the formation period, low Max Pain portfolios exhibit very positive returns

that reverse the third week of the month, when the Max Pain strategy delivers abnormal

results. Symmetrically, high Max Pain portfolios record negative returns that become even

more negative as we get closer to the formation period, then reverse and become positive

during the Max Pain strategy period, which leads to a return during that week of 40 bp.

Consistent with Figure 3, we observe that this return of 40 bp disappears over the following

weeks.

The pattern of returns before the third week of the month provides a clear characteriza-

tion of the type of stocks that constitute deciles 1 and 10 in our sorting scheme. In the case

of high Max Pain (decile 10) for example, the stocks had experienced very negative returns

for several weeks and, in typical momentum pattern, were oversold and ready for some

reversal, as we observe in the data. However, this begs the following question: Why the

reversal consistently starts during the week options expire –as we have established with

the tests we just presented?

17
 In the next section we explore possible explanations of the Max Pain strategy. In the

process, we will rule out some of the patterns documented in the literature around options

expiration as possible reasons.

[Figure 4 about here.]

5 Discussion of Possible Explanations

The available data does not permit the type of granular analysis that would unveil a

conclusive explanation of why the Max Pain strategy pays very high risk-adjusted returns.

However, the tests we have presented and a few more we include in this section allow us

to establish a narrative that is both sound from an economic point of view, and consistent

with the available data.

5.1 Short-Term Reversal (With a Little Help)

Figure 4 is very revealing as it shows that stocks in the extreme Max Pain deciles have

experienced dramatic price trends over the weeks prior to the implementation of the

strategy (i.e. the weeks that precede the week of options expiration). A generally accepted

regularity in stock price dynamics is the tendency of prices to overreact to news, which

leads to contrarian strategies that will benefit from the short-term reversal –for example,

Jegadeesh and Titman (1995). Therefore, the reversal pattern captured in Figure 4 is

not a new finding by itself. What is new is the timing, which in this case is completely

predictable: The first stage of the reversal takes place during the week of options expiration.

We emphasize that, although there is a reversal during that week affecting both overbought

and oversold stocks (deciles 1 and 10, respectively, of the Max Pain strategy), it is in the

latter case where the reversal becomes more obvious and consistent, as Figure 4 illustrates.

It is also not new that the more significant type of reversal (or possibly the only type of

18
reversal) tends to occur in stocks that have experienced a substantial price drop. This was

discussed in the literature a long time ago, for example in Brown, Harlow, and Tinic (1988)

and Bremer and Sweeney (1991). However, there is no case in the literature in which the

timing of the reversal is predictable. This is a special feature of the reversal we contemplate

in this paper that helps in our quest for an explanation of the profitability of the Max Pain

strategy.

First, we need to point out that the drop in prices of the stocks in decile 10 in the Max

Pain strategy should also trigger trading in options. In particular, investors that expect

a drop in stock prices and want to benefit from it, in addition to trading in the stocks

(selling or short selling them) can buy puts and sell calls. As the price of the stock drops,

the Max Pain measure increases, as the puts get further in the money. We also note that,

simultaneously, the more sophisticated participants in the market probably expect a reversal

of this heavily sold stocks at some point, and in anticipation they buy them. From the data

discussed earlier, we know that the reversal starts in the week prior to the expiration of

the options on which we base our Max Pain measure: In particular, in Panel A of Table 5

we show that there is a large abnormal trading activity on the stocks in deciles 1 and 10

of the Max Pain strategy. Why precisely in that week? A possible, at least partial reason

is that it is in the best interest of the holders of short positions in the puts that were sold

during the price drop and are now in the money. They would be accompanied by other

market participants that over time have learned to predict the timing of the reversal and

also buy stocks. Panel B of Table 5 reports that investors buy stocks in decile 10 and sell

stocks in decile 1, predicting the dynamics that take place before expiration of the options.

In sum, we are suggesting possible price manipulation –the incentives are obvious–, in

line with other evidence of price manipulation by sophisticated investors, for example,

Ben-David, Franzoni, Landier, and Moussawi (2013). Arguably, as this dynamics have

become a pattern, other investors with no positions in options anticipate the price increases

and buy the stocks, further increasing the price pressure.

19
 In the next subsection we show further evidence that some sophisticated investors

predict the reversal of prices in decile 10 precisely on the expiration week of the options

that is the basis of the Max Pain strategy.

5.2 Evidence From Informed Investors

A growing literature studies the connection between options trading and information. For

example, investors with superior information are motivated to engage in options trading due

to the implicit leverage they provide (Black, 1975) and the absence of constraints on short

selling (Figlewski and Webb, 1993). Biais and Hillion (1994), as well as Easley, O’Hara,

and Srinivas (1998), indicate that in sufficiently liquid options markets, it is advantageous

for informed individuals to trade both stocks and options. Consequently, trading volumes in

options should reflect the existence of higher quality information about the future price of

the underlying. Numerous empirical studies support this assertion through the analysis of

options order flow (for instance Easley et al., 1998; Pan and Poteshman, 2006; Hu, 2014).

In the context of the dynamics of the Max Pain strategy we analyze in this paper, we would

expect investors who anticipate the price reversal to trade options accordingly. To this end,

we examine whether the trading activity in options predicts the price reversal at the core of

the Max Pain strategy.3

With that context in mind, we look at the order imbalance (IMB), that measures the

net quantity of options purchased by end-users. A positive (negative) IMB is indicative

of net buying (selling) from end-users. We show results separately for calls and puts and

for positive and negative exposure to the underlying –that we define later. All our volume

measures are net of delta-hedges.

3
The Max Pain strategy is based on options that expire the week after portfolio formation. However, the
informed investors who can predict the trend of prices will engage in trading of options that expire later. The
options that expire within that week are subject to unpredictable activity motivated by the strategies of the
holders who might want to avoid the delivery process or are rolling over their positions. For this reason, it is
standard in the empirical literature to exclude options that are about to expire.

20
 We report our results in Table 7. Panel A displays time-series averages for option order

imbalances of calls and puts in low and high Max Pain portfolios (deciles 1 and 10), the

week before options expiration. For comparison purposes, we also show results for decile 5.

Similarly to Golez and Goyenko (2021), we compute the positive, negative, and net view of

the investors on the underlying stock. We define positive view as the sum of trades buy-call

and sell-put options. The negative view of investors is measured as the total number of

trades of sell-call and buy-put options. The net effect is the difference between positive

and negative views.

As expected, in the high Max Pain portfolio (decile 10), we find that investors are net

buyers of call options and net sellers of put options. The put trading activity is marginally

more economically significant. We also find that the difference between the positive and

negative views is positive and statistically significant because investors tend to buy more

calls and sell more put options. In the middle portfolio, we observe that investors are net

sellers of call and put options and the negative view dominates but it is much smaller than

in the case of low max pain portfolios. In Panel B we find similar results when we focus on

the trading activity on the 2nd Friday of the month which is the day in which we form our

portfolios.

Regarding the low Max Pain portfolio (decile 1), we find that investors are net sellers

of call and put options, with more trading activity for call options. The difference between

positive and negative options trading is negative and statistically significant.

[Table 7 about here.]

Next, similarly to Hu (2014), we examine directly –as opposed to assuming it, as in

the previous analysis– whether there is information in the trading activity of call and put

options about the returns that the Max Pain strategy benefits from. Table 8 shows daily

predictive regressions of stock returns for call and put order imbalances during our holding

period (i.e. the expiration week). Our cross-sectional regression takes the following form:

21
 5
X 5
X 5
X
Ret i,t = α + β j I M Bi,t−
C all
j
+ γ j I M Bi,t−
Put
j
+ δ j Ret i,t− j + "i,t ,
j=1 j=1 j=1

where Ret i,t+1 represents the stock return at time t + 1, I M B C all is the order imbalance of

the call options and I M B Put is the order imbalance of the put options. We also control for

stock returns during the previous five days. Our analysis incorporates both opening and

closing option positions, consistent with existing literature.

In the low Max Pain portfolio (decile 1), we find that the coefficient of the option order

imbalance of calls with one lag is positive and statistically significant, but the trading activity

of puts does not contain relevant information about future stock returns. We recall from

Table 7 that investors are net sellers of calls and puts, but primarily the former, anticipating,

therefore, the upcoming reversal in the aggregate. More strongly, as it is the case throughout

the analysis of the strategy, in the high Max Pain portfolio (decile 10), the option imbalance

of call options with one lag is positive and statistically significant –therefore predicting the

upcoming increase in stock prices. We also find that the coefficient of option imbalances of

puts with two lags is negative and statistically significant –consistent with the purchases of

calls. To confirm the predictive ability of option traders, we also find statistically significant

coefficients for the calls (positive) and puts (negative) for the mid portfolio (decile 5), but

the coefficients are economically much smaller than in the extreme deciles.

In sum, in this subsection we have shown that some investors trade options in a way

consistent with predictability of upcoming price changes. This ability manifests itself

especially (in terms of the trading strategy as well as in the size and statistical significance

of the coefficients that link volume and returns) in the high Max Pain portfolio, from where

the biggest share of the returns of the spread portfolio strategy are coming.

Overall our results are consistent with an expected reversal in prices (especially after

large price drops) that some investors are able to predict. As we have already stated,

one possible explanation for the timing of the reversal, which consistently overlaps with

22
the expiration of options, is that some holders of short positions in the expiring options

manipulate the prices of the underlying so as to minimize their exposure. In fact, a

kinder interpretation is that this investors just "push forward" an increase in prices that,

nonetheless, was going to take place. Furthermore, these investors could be deliberately

forcing prices upward to minimize the payoff of expiring put options, or just might be

investors experienced enough to know how the prices are going to evolve whose trading

activity would actually help the reversal.

In the rest of this section we consider pricing patterns around option expiration that

have been documented in the literature and could explain the dynamics we have described.

However, our analysis rules out that such is the case.

[Table 8 about here.]

5.3 Clustering

Previous literature shows that, for mechanical reasons, on expiration dates the closing prices

of stocks with options tend to cluster at the strike prices. We have already shown that the

price dynamics that lead to the profits based on the Max Pain strategy happen throughout the

expiration week, not just the expiration day, so we conjecture that “clustering" is a different

type of event. In order to rule out any connection, we consider three of the mechanical

reasons the literature has studied on this topic, and evaluate whether they can explain the

stock dynamics predicted by the Max Pain theory and subsequent abnormal returns. We

first describe how these price pressures work, and how we are going to estimate them in

our data. At the end of the subsection we present our empirical results and conclusions.

Rebalancing of delta-hedges on existing option positions. Avellaneda and Lipkin

(2003) propose a model in which option markets participants delta-hedge their positions

in options dynamically, and they show that as a result of the unwinding of delta-hedged

23
positions, stock prices can cluster or “pin" at strike prices on expiration days. Ni et al.

(2005) and Golez and Jackwerth (2012) show evidence consistent with this prediction.

Therefore, we need to account for delta-hedging trading in our analysis of the Max Pain

strategy.

To account for this delta-hedging activity, we assume that market-makers are the primary

delta-hedgers, as Ni et al. (2005), and we compute their net purchased open interest to get

our first measure of clustering pressure. We obtain this variable from the open interest data

at the close of trading on Thursday in the expiring puts and calls whose strike prices are

nearest to the Thursday closing stock price. In particular, we assume that the market-maker

net open interest is the reciprocal of the non-market-makers net purchased open interest.

We estimate the buy and sell open interest by aggregating the open buy, close sell, open

sell, and close buy from four exchanges.

Additional delta-hedging from changes in positions. Changes in positions in options

leads to further hedge adjustments of market makers. It is especially important to consider

trades of the many investors with positions in options who want to avoid the possibility of

taking or making delivery of the underlying if their long or short positions end up in the

money. For this purpose, they sell their long positions or close their short positions on the

expiration Friday. As the counterpart is typically a market-maker, they need to trade in the

underlying to stay delta-neutral.

Following also Ni et al. (2005), we compute another measure to capture the clustering

pressure from this source as follows: New Delta Hedging = sign(S−K)×(DeltaAdjChgOICMall

M
+

DeltaAdjChgOI M M
Put
), where sign(S-K) takes the values +1, 0 and −1 when the Thursday

closing stock price is, respectively, greater than, equal to, or less than the nearest strike

price. In sum, DeltaAdjChgOICMalMl (DeltaAdjChgOI M M

Put
) denotes the delta-adjusted Thursday

to Friday change in net market-maker open interest aggregated across all calls (puts) on

the underlying security.

24
Option-related stock trading by non-delta hedging investors. Investors who do not

delta-hedge sometimes combine a position in the option and the underlying simultaneously.

When the options expire, they often unwind the whole position, which leads to trading in

the stock. The most popular among these are covered calls –long stock, short call– and

protective puts –long stock, long put. Investors are prone to fully unwind them. Both cases

involve the sale of the stock, and can contribute to clustering when the stock closing price

at expiration is above the strike price, as suggested by the model of Avellaneda and Lipkin

(2003).

In line with Ni et al. (2005), our third measure attempts to capture this clustering pres-

sure: C C P PU = I(S−K)×[OT M Pur chasedOI Put

F ir m+Cust omers
+OT M W r i t t enOI CF all
ir m+Cust omers
],

where I(S−K) = 1 if the Thursday closing price of the underlying stock exceeds the exercise
F ir m+Cust omers
price nearest the Thursday closing stock price, and 0 otherwise. OT M Pur chasedOI Put

denotes the purchased open interest of expiring OTM puts at the close on Thursday by

firms and customers, and OT M W r i t t enOI CF all

ir m+Cust omers
is the call written open interest of

expiring OTM calls at the close on Thursday by firm and customers.

In Table 9 we present the relation between Max Pain and stock returns after controlling

for the previous indicators of clustering pressure. We include in the regression a variable

that measures the absolute distance between Thursday’s closing stock price and the strike

price that is nearest to Friday’s closing stock price. This variable captures the possibility

that the stock price might be more likely to close near an option exercise price on the

expiration day if it closed near that price on the preceding day. Similar to Ni et al. (2005),

we also control for the option volume that opens new written positions on Tuesday through

Thursday leading up to expiration for both firm proprietary traders and public customers,

and the written open interest for firm proprietary traders and public customers at the close

of trading on Thursday. Both of these variables are constructed from the expiring calls and

puts with a strike price nearest to Thursday’s closing stock price. Finally, we include in one

of our specifications the same controls as in Table 3.

25
 We find that, even after accounting for these factors, the Max Pain coefficient is highly

significant. Therefore, clustering does not appear to drive our results.

[Table 9 about here.]

5.4 Net Gamma

In the previous subsection we ruled out that clustering, the price pressure that stocks

experience mechanically at expiration of the options, is behind the Max Pain returns.

However, the literature has also considered the price effects of hedging away from expiration.

In particular, changes in deltas (so called gammas) lead to rebalancing of the hedges. For

example, Ni, Pearson, Poteshman, and White (2021) show that this activity significantly

affects volatility of stock prices.

To complement our previous analysis, we examine whether hedge rebalancing during

the expiration week correlates with the profitability of the max pain strategy. In particular,

a long portfolio of options implies a positive gamma, which implies that the size of the

delta-hedging position is positively related to the underlying price. On the other hand, a

short portfolio of options indicates a negative gamma, i.e. the size of the delta-hedging

position is negatively related to the underlying price.

Table 10 shows cross-sectional regressions of daily stock returns on lagged values of

the aggregate hedgers’ gamma exposure on stock i on day t. We follow Ni et al. (2021) to

compute the net gamma position of the marker makers. Gamma exposure on stock i on

day t is computed as the gamma-weighted sum of inventories across the options written on

that stock.

Our cross-sectional regression takes the following form:

5
X 5
X
Ret i,t = α + β j Gammai,t− j + δ j Ret i,t− j + "i,t
j=1 j=1

26
where Ret i,t+1 represents the stock return at time t + 1, Gamma is the net gamma. We also

control for stock returns for the previous five days.

From Table 10 it is clear that hedge rebalancing activities during the expiration week do

not account for the profitability observed in our low and high max pain portfolios. Despite

this, our analysis reveals predictive patterns for the middle portfolios, which are particularly

influenced by clustering pressure around expiration dates.

[Table 10 about here.]

6 Robustness Tests

6.1 Put to Call Open Interest Ratios

Stocks in the high Max Pain bin (number 10) of our strategy are expected to go up, which

would be consistent, if the Max Pain Theory is correct, with a larger number of puts than of

calls on these stocks. In Table 11 we report the time-series average of the median put to

call open interest portfolios. The ratio increases monotonically with the Max Pain value.

The difference between high and low Max Pain portfolios is highly positive and significant,

indicating higher open interest for put (call) options for high (low) Max Pain portfolios,

consistent with the rational of the Max Pain Theory.

[Table 11 about here.]

27
6.2 Stocks with and without Options

The Max Pain strategy should only work for optionable stocks, and we should not observe

abnormal returns in similar stocks without options. Of course, stocks with options are, in

general, very different from stocks without options, but following standard methodology,

we look for stocks without options that look as close as possible to the stocks whose prices

display the dynamics predicted by Max Pain theory. In particular, we match stocks with

and without options according to size and volume, using propensity score matching. We

apply a one-to-one matching without replacement. Then we assign the Max Pain value of

each optionable stock to a non-optionable stock with similar characteristics. We sort stocks

with and without options based on Max Pain for the matched samples. We show in Panel A

of Table 12 that only stocks with options render positive and statistically significant returns.

We obtain similar results, in Panel B, for the CAPM and FF3 alphas of the spread portfolios.

[Table 12 about here.]

6.3 Max Pain with Different Number of Strike Prices per Stock

In our baseline analysis, we have considered all stocks, regardless of the number of attached

options. Here we examine if our results are affected by the number of strike prices of

attached options for a particular stock.

Panel A of Table 13 reports average returns of stocks that have at least 2, 3 or 4 strike

prices, sorted on Max Pain. The results previously reported hold regardless of the minimum

number of available strike prices. All portfolios are value-weighted. Panel B shows the

alphas of the portfolios, that are positive and statistically significant.

[Table 13 about here.]

28
6.4 Index Options

In all our tests so far we have used single stock options. Yet, some of the most broadly

traded options are options on indices. It is reasonable then to ask whether the Max Pain

strategy will yield similar profits when applied to index options. However, we conjecture

that the strategy will fail here because two key factors that drive the price predictability

when the underlying is a stock are missing here. First, we have shown evidence that the

predictability is the result of a price reversal for stocks whose prices have trended up or

(and this is the strongest case) down over recent weeks. However, by constructions, indices

are less volatile than individual stocks –at least than some individual stocks–, therefore,

steep price trends and the subsequent reversals are less likely to happen. Additionally, the

price manipulation that our analysis shows is consistent with the timing of the reversal is

practically impossible to execute on an index. We test the Max Pain strategy using indices

and present the results in Table 14. There are not enough optionable indices to use deciles,

therefore we allocate them into terciles. We verify that there is dispersion in the Max Pain

measure across the different portfolios and report their corresponding average Max Pain

measure. Panel A shows that the average index return decreases with the Max Pain measure

and yields a spread portfolio that is negative and statistically insignificant. We find similar

results, in Panel B, for the alphas of the spread portfolio.

[Table 14 about here.]

29
7 Conclusions

We test the empirical validity of the Maximum Pain theory. We construct a measure of Max

Pain price and form decile portfolios on the second Friday of each month. We hold the

portfolios for one week. We find that a spread portfolio that buys high Max Pain stocks and

sells low Max Pain stocks offers very positive and statistically significant returns. We find

similar results for alphas. We document that this is the result of price reversals for stocks

whose prices had increased or decreased substantially –especially the latter– in the weeks

before the options expiration. Price reversals are a common occurrence, what is surprising

in this case is the timing of the reversals, which is predictable, as they start exactly the

week before options expiration. A possible factor is that if the reversal takes place before

the options expire, it benefits investors with short positions, as it minimizes their payoffs.

Therefore, they have clear incentives to push forward the reversal by manipulating the stock

prices. The price manipulation does not represent a deviation of the stock prices from their

fundamental values, just the opposite, as it speeds up the convergence of the prices to their

fundamental value. Further evidence of the possible existence of manipulation is that the

profits resulting from the Max Pain strategy are highest for small and illiquid stocks, which

are easier to sway. The Max Pain strategy yields profits beyond the factors that arguably

explain the clustering of stock prices at maturity of the options and the gamma imbalances

before expiration, both types of price dynamics discussed in the literature. The results are

robust to a number of additional tests and the inclusion of controls.

30
References

Avellaneda, M. and M. D. Lipkin (2003). A market-induced mechanism for stock pinning.

Quantitative Finance 3(6), 417.

Ben-David, I., F. Franzoni, A. Landier, and R. Moussawi (2013). Do hedge funds manipulate

stock prices? The Journal of Finance 68(6), 2383–2434.

Biais, B. and P. Hillion (1994). Insider and liquidity trading in stock and options markets.

The Review of Financial Studies 7(4), 743–780.

Black, F. (1975). Fact and fantasy in the use of options. Financial Analysts Journal 31(4),

36–41.

Bremer, M. and R. J. Sweeney (1991). The reversal of large stock-price decreases. The

Journal of Finance 46(2), 747–754.

Brown, K. C., W. V. Harlow, and S. M. Tinic (1988). Risk aversion, uncertain information,

and market efficiency. Journal of financial economics 22(2), 355–385.

Daniel, K., M. Grinblatt, S. Titman, and R. Wermers (1997). Measuring mutual fund

performance with characteristic-based benchmarks. The Journal of finance 52(3), 1035–

1058.

Easley, D., M. O’Hara, and P. S. Srinivas (1998). Option volume and stock prices: Evidence

on where informed traders trade. The Journal of Finance 53(2), 431–465.

Fama, E. F. and K. R. French (1993). Common risk factors in the returns on stocks and

bonds. Journal of Financial Economics 33(1), 3–56.

Fama, E. F. and J. D. MacBeth (1973). Risk, return, and equilibrium: Empirical tests. The

Journal of Political Economy, 607–636.

31
Figlewski, S. and G. P. Webb (1993). Options, short sales, and market completeness. Journal

of Finance, 761–777.

Golez, B. and R. Goyenko (2021). Disagreement in the equity options market and stock

returns. The Review of Financial Studies 35(3), 1443–1479.

Golez, B. and J. Jackwerth (2012). Pinning in the s&p 500 futures. Journal of Financial

Economics 106(1), 566–588.

Hu, J. (2014). Does option trading convey stock price information? Journal of Financial

Economics 111(3), 625–645.

Jegadeesh, N. and S. Titman (1995). Short-horizon return reversals and the bid-ask spread.

Journal of Financial Intermediation 4(2), 116–132.

Lee, C. and M. J. Ready (1991). Inferring trade direction from intraday data. The Journal

of Finance 46(2), 733–746.

Newey, W. K. and K. D. West (1987). A simple, positive semi-definite, heteroskedasticity

and autocorrelation consistent covariance matrix. Econometrica 55(3), 703–708.

Ni, S. X., N. D. Pearson, and A. M. Poteshman (2005). Stock price clustering on option

expiration dates. Journal of Financial Economics 78(1), 49–87.

Ni, S. X., N. D. Pearson, A. M. Poteshman, and J. White (2021). Does option trading have

a pervasive impact on underlying stock prices? The Review of Financial Studies 34(4),

1952–1986.

Pan, J. and A. M. Poteshman (2006). The information in option volume for future stock

prices. Review of Financial Studies 19(3), 871–908.

32
 Figure 1. Example of the Construction of Max Pain

GME Stock - Call and Put Max Pain GME Stock - Total Max Pain

5000 5000
MaxPain
Current
Loss
Max Pain (in million $)

Max Pain (in million $)

Stock
4000 4000 -43.29%
Price
MaxPain $264.5
3000 3000 Strike
Price
$150
2000 2000

1000 1000

0 0
100 200 300 400 500 600 700 800 100 200 300 400 500 600 700 800
Strike Prices Strike Prices
Call Max Pain Put Max Pain Max Pain

Moderna Stock - Call and Put Max Pain Moderna Stock - Total Max Pain
80 80
33

70 70
MaxPain
60 60
Max Pain (in million $)

Max Pain (in million $)

Current Gain
Stock 17.75%
50 50 Price
MaxPain
$59.5
40 40 Strike
Price
30 30 $69.5

20 20

10 10

0 0
35 40 45 50 55 60 65 70 75 80 35 40 45 50 55 60 65 70 75 80
Strike Prices Strike Prices

Call Max Pain Put Max Pain Max Pain

The figure displays two example of the construction of the Max Pain gain loss measure. The top figure show an example of Max Pain loss using data from the
GME stock on March 12, 2021. The bottom panels use data of Moderna for a gain on September 11, 2020. The data are collected from OptionMetrics and
CRSP.
 Figure 2. Example of Portfolio Formation and Holding Periods

May 2012
Monday Tuesday Wednesday Thursday Friday Saturday Sunday

1 2 3 4
Forming Portfolios
5 6

Holding Period
7 Holding Period
8 Holding Period
9 10
Holding Period
11
Holding Period
12
Options Expire
13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30 31

The figure displays an example of the formation and holding period of the strategy for May 2012.

34
 Figure 3. Daily Cumulative Returns

Stock Returns
0.45

0.4

0.35

0.3
Cumulative Stock Returns

0.25

0.2

0.15

0.1

0.05

0 2 4 6 8 10 12 14 16 18 20
Days
HML

The figure displays daily cumulative stock returns for the Max Pain portfolio. The period starts on the 2nd Friday of the month
and includes 20 trading days after the formation period. The data are collected from OptionMetrics and CRSP and contain daily
series from January 1996 to December 2021.

35
 Figure 4. Weekly Event Study

Event Study - Stock Returns

0.03

0.02

0.01
Stock Returns

-0.01

-0.02

-3 -2 -1 0 1 2 3 4
Weeks
Low Max Pain High Max Pain

The figure displays an event study of weekly stock returns around the week of the holding period for low and high Max Pain
portfolios. The data are collected from OptionMetrics and CRSP and contain daily series from January 1996 to December 2021.

36
 Table 1. Characteristics of Max Pain Portfolios
This table displays time-series averages of median characteristics of stocks sorted into deciles on the last week before the
expiration Friday of the month based on Max Pain. Panel A shows time-series averages of median Max Pain (in millions) of
portfolios sorted based on the Max Pain measure. We report in Panel B the stock price, stock volume (SVOL), momentum
(MOM), stock reversals (REV), illiquidity (ILLIQ) (in percent), institutional ownership (IOR), book to market (B/M), Size
(in billion), Debt to assets (D/A), idiosyncratic volatility (IVOL), a Nasdaq variable that takes a value of one if the stock is
a Nasdaq stock and the absolute difference between closing prices and nearest strike price (AD prices). We report Newey
and West (1987) t-statistics with 6 lags in parenthesis. The data is from CRSP, Compustat and Optionmetrics and contains
monthly series from January 1996 to December 2021.

Panel A: Portfolios sorted based on Max Pain

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P10-P1 t-stat

M a x Pain -0.081 -0.049 -0.033 -0.021 -0.010 -0.000 0.011 0.023 0.040 0.076 0.157 (17.67)

Panel B: Stock Characteristics of Portfolios sorted based on Max Pain

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P10-P1 t-stat

SV O L 0.619 0.596 0.631 0.657 0.672 0.700 0.659 0.633 0.574 0.595 -0.024 (-1.17)
IOR 0.778 0.784 0.780 0.772 0.764 0.762 0.762 0.770 0.777 0.759 -0.019 (-5.87)
B/M 0.300 0.334 0.360 0.377 0.384 0.397 0.398 0.393 0.380 0.348 0.048 (6.30)
D/A 0.154 0.184 0.204 0.211 0.217 0.217 0.217 0.209 0.192 0.151 -0.003 (-0.44)
REV 0.044 0.030 0.022 0.018 0.013 0.008 0.005 -0.001 -0.008 -0.028 -0.072 (-10.08)
M OM 0.235 0.175 0.153 0.134 0.128 0.120 0.118 0.123 0.121 0.110 -0.126 (-5.51)
IVOL 0.024 0.019 0.017 0.016 0.016 0.016 0.016 0.017 0.019 0.025 0.001 (1.69)
S t ockP r ice 33.360 37.146 38.466 37.816 36.864 35.742 34.909 33.817 30.949 23.631 -9.729 (-11.90)
Size 1.587 2.292 2.788 3.095 3.193 3.169 2.889 2.514 1.928 1.114 -0.473 (-11.46)
I L L IQS t ocks 0.002 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.002 0.002 0.001 (5.51)
N asd aq 0.590 0.486 0.426 0.398 0.388 0.392 0.395 0.415 0.475 0.596 0.006 (0.62)
AD Prices 0.968 0.950 0.893 0.813 0.761 0.711 0.722 0.775 0.852 0.932 -0.036 (-1.91)

37
 Table 2. Stocks sorted on the Max Pain measure
This table displays average returns of stocks sorted into deciles on the last week before the expiration Friday of the month based
on the Max Pain measure. In particular, we form portfolios on the second Friday of each month and hold them for one week,
which refers to the expiration date of the attached options. Panel A shows average stock returns. Panel B shows alphas based on
CAPM and the Fama and French (1993) 3-factor model. We use weekly factors and we use the third weekly return of every
month. Stock returns are calculated using the midpoint of the bid and ask prices at market close adjusted for stock splits. We
report Newey and West (1987) t-statistics with 6 lags in parenthesis. The data is from CRSP, Compustat and Optionmetrics and
contains monthly series from January 1996 to December 2021.

Panel A: Stock returns of portfolios of stocks sorted based on the Max Pain measure

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P10-P1 t-stat

EW -0.001 0.001 0.002 0.003 0.002 0.002 0.003 0.003 0.003 0.004 0.004 (3.66)
VW 0.000 0.002 0.003 0.003 0.004 0.002 0.004 0.004 0.003 0.004 0.004 (2.65)

Panel B: Alphas of portfolios of stocks sorted based on the Max Pain measure

CAPM FF3

EW 0.004 0.004
(3.40) (3.68)
VW 0.003 0.004
(2.19) (2.83)

38
 Table 3. Cross-Sectional Regressions
This table displays cross-sectional regressions of stock returns on the Max Pain measure and a number of controls. Specifically, our cross-sectional
regression takes the following form:

Ret i,t+1 = α + β M ax Paini,t + γC ont r olsi,t + "i,t+1

where Ret i,t+1 represents the stock return at time t + 1 and the set of controls include the stock illiquidity, a dummy variable that takes a value of one if
the stock is traded in NASDAQ, price, size, book to market, debt to assets, stock volume (SVOL), institutional ownership (IOR), idiosyncratic volatility
(IVOL), weekly stock reversal, monthly stock reversal and momentum. We show results for the full sample, small stocks, illiquid stocks, Nasdaq stocks,
and growth stocks. We report Newey and West (1987) t-statistics with 6 lags in parenthesis. The data is from CRSP, Compustat and Optionmetrics and
contains monthly series from January 1996 to December 2021.

Stock Returns

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Full Sample Small stocks Illiquid Stocks Nasdaq Stocks Growth Stocks
M a x Pain 0.017*** 0.014*** 0.027*** 0.027*** 0.028*** 0.027*** 0.019*** 0.018*** 0.026*** 0.021**
(2.94) (3.05) (3.41) (3.45) (3.12) (3.00) (2.69) (2.77) (2.85) (2.52)
P r ice -0.000 -0.000 -0.000 -0.000 0.000
39

(-0.77) (-0.97) (-1.33) (-0.64) (0.50)

Size 0.000 -0.000 -0.000 0.000** 0.000
(0.87) (-0.27) (-0.10) (2.01) (0.89)
SV O L -0.000 0.000 0.000** -0.000 -0.000
(-0.63) (1.24) (2.25) (-1.01) (-0.22)
REV Weekl y -0.005 -0.004 -0.005 -0.009 -0.012**
(-1.04) (-0.56) (-0.64) (-1.50) (-2.03)
IOR 0.000 0.001 0.001 0.001 0.001
(0.44) (0.46) (0.28) (0.39) (0.77)
B/M 0.000 0.000 0.001 0.000 0.001
(0.13) (0.28) (0.58) (0.05) (0.05)
D/A -0.002 -0.004 -0.004* -0.002 -0.001
(-1.29) (-1.57) (-1.96) (-1.09) (-0.60)
REV M onthl y -0.004 -0.002 0.001 -0.001 -0.001
(-1.36) (-0.52) (0.22) (-0.21) (-0.46)
M OM 0.000 -0.001 0.000 0.000 0.000
(0.32) (-0.65) (0.54) (0.16) (0.30)
I L LIQS t ocks -0.015 -0.024 -0.026 -0.012 0.076
(-0.70) (-0.74) (-0.81) (-0.46) (0.63)
IVOL -0.019 -0.007 -0.044 0.002 -0.011
(-0.55) (-0.19) (-1.13) (0.06) (-0.26)
cons 0.002 0.003*** 0.001 0.004 0.002 0.005** 0.002 0.003** 0.003* 0.002
(1.47) (3.09) (0.84) (1.35) (0.92) (2.16) (1.58) (2.36) (1.66) (1.22)

R-squared 0.006 0.079 0.011 0.131 0.012 0.128 0.007 0.084 0.015 0.140
 Table 4. Stocks sorted on Max Pain: Different Weeks
This table displays average stock returns that are sorted based on the Max Pain measure. We form portfolios on the first,
second, or third Friday of the month. We hold the portfolio until the expiration of the option. Panel A shows stock returns that
correspond to the holding period of 2, 3 and 4 weeks to expiration. Panel B shows alphas based on CAPM and the Fama and
French (1993) 3-factor model for the different holding periods. Stock returns are calculated using the midpoint of the bid and
ask prices at market close adjusted for stock splits. We report Newey and West (1987) t-statistics with 6 lags in parenthesis. The
data is from CRSP, Compustat and Optionmetrics and contains monthly series from January 1996 to December 2021.

Panel A: Stock returns of portfolios of stocks sorted based on Max Pain

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P10-P1 t-stat

Two Weeks to Expiration

EW -0.001 0.001 0.002 0.003 0.003 0.003 0.003 0.003 0.003 0.003 0.004 (1.88)
VW 0.001 0.002 0.003 0.004 0.002 0.003 0.003 0.005 0.005 0.005 0.004 (1.44)

Three Weeks to Expiration

EW 0.003 0.005 0.006 0.006 0.007 0.007 0.007 0.007 0.007 0.007 0.004 (1.85)
VW 0.006 0.006 0.005 0.007 0.007 0.007 0.008 0.009 0.007 0.009 0.004 (1.23)

Four Weeks to Expiration

EW 0.008 0.009 0.008 0.010 0.009 0.009 0.010 0.009 0.010 0.010 0.002 (0.67)
VW 0.007 0.008 0.007 0.008 0.006 0.008 0.009 0.010 0.012 0.011 0.004 (1.45)

Panel B: Alphas of portfolios of stocks sorted based on Max Pain

CAPM FF3 CAPM FF3 CAPM FF3

Two Weeks to Expiration Three Weeks to Expiration Four Weeks to Expiration

EW 0.003 0.003 0.005 0.005 0.003 0.002

(1.54) (1.63) (2.01) (1.98)) (1.00) (0.92)
VW 0.003 0.003 0.005 0.005 0.005 0.005
(1.02) (1.17) (1.55) (1.54) (1.82) (1.65)

40
 Table 5. Stocks sorted on Max Pain: Abnormal Stock Volume
This table reports the average abnormal volume and order imbalances of stocks that are sorted based on the Max Pain measure.
We form portfolios on the first, second, or third Friday of the month. In Panel A, we show results for abnormal volume. We
compute abnormal volume as:
Abnor mal Volumew,m = Volumew,m − Volumem−1→m−n ,
where Volumew represents the average daily volume (expressed in millions) over the last week w of month m before the
expiration of the option and Volumem−1→m−n is the average daily volume over the months m − n to m − 1 where n = 1, 3, 6, 12.
In Panel B, we present results for order imbalances. We form portfolios on the first, second, or third Friday of the month. We
compute abnormal order imbalances as:

Abnor malOI Bw,m = OI Bw,m − OI Bm−1→m−n ,

where OI Bw represents the average daily order imbalance over the last week w of month m before the expiration of the option
and OI Bm−1→m−n is the average daily volume over the months m − n to m − 1 where n = 1, 3, 6, 12.
We report Newey and West (1987) t-statistics with 6 lags in parenthesis. The data is from CRSP, TAQ and Optionmetrics and
contains monthly series from January 1996 to December 2021.

Panel A: Abnormal Volume of Portfolios of stocks sorted based on Max Pain

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P10-P1 P10-P5 P1-P5

One Month

EW 0.191 0.064 0.040 0.027 0.016 0.030 0.007 0.052 0.087 0.234 0.043 0.219 0.175
(7.69) (3.61) (1.93) (1.48) (0.74) (1.25) (0.34) (2.05) (3.98) (3.54) (0.72) (3.30) (6.59)

Three Months

EW 0.245 0.075 0.045 0.022 0.034 0.033 0.014 0.082 0.114 0.289 0.045 0.255 0.211
(8.60) (3.91) (1.97) (0.92) (1.37) (1.06) (0.51) (2.92) (4.39) (3.94) (0.70) (3.82) (6.84)

Six Months

EW 0.313 0.095 0.057 0.023 0.036 0.045 0.019 0.094 0.135 0.353 0.040 0.317 0.277
(8.00) (4.13) (1.85) (0.68) (1.06) (1.13) (0.54) (2.76) (4.23) (4.39) (0.60) (4.47) (6.22)

Twelve Months

EW 0.396 0.133 0.084 0.039 0.045 0.053 0.036 0.114 0.170 0.457 0.061 0.412 0.351
(7.25) (4.60) (2.09) (0.94) (1.04) (1.04) (0.75) (2.56) (4.37) (4.75) (0.78) (5.05) (6.03)

Panel B: Abnormal Order Imbalance of Portfolios of Stocks sorted based on Max Pain

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P10-P1 P10-P5 P1-P5

One Month

EW -0.004 -0.002 -0.001 -0.000 0.001 0.000 0.002 0.002 0.001 0.005 0.009 0.004 -0.005
(-3.32) (-1.97) (-1.17) (-0.11) (0.78) (0.26) (1.84) (1.83) (1.40) (3.46) (5.10) (3.07) (-5.62)

Three Months

EW -0.003 -0.001 -0.001 -0.000 0.001 0.000 0.001 0.001 0.001 0.003 0.007 0.003 -0.004
(-2.53) (-1.13) (-0.64) (-0.01) (0.71) (0.02) (1.37) (0.83) (0.63) (2.46) (4.08) (2.16) (-4.86)

Six Months

EW -0.002 -0.001 -0.001 -0.000 0.001 -0.000 0.001 0.000 0.000 0.002 0.005 0.002 -0.003
(-1.70) (-0.81) (-0.58) (-0.05) (0.53) (-0.10) (0.98) (0.26) (0.19) (1.63) (3.27) (1.46) (-3.41)

Twelve Months

EW -0.002 -0.001 -0.000 0.000 0.001 -0.000 0.001 -0.000 -0.000 0.001 0.003 0.001 -0.002
(-0.99) (-0.33) (-0.24) (0.03) (0.42) (-0.05) (0.69) (-0.07) (-0.23) (0.87) (2.51) (0.71) (-2.31)

41
 Table 6. Stocks sorted on Max Pain: Placebo Test
This table shows average stock returns that are sorted based on the Max Pain measure. We form a portfolio from two weeks to
expiration until one week to expiration. Panel A shows weekly stock returns. Panel B shows alphas based on CAPM and the Fama
and French (1993) 3-factor model. We use weekly factors, and we use the third weekly return of every month. Stock returns are
calculated using the midpoint of the bid and ask prices at market close adjusted for stock splits. We report Newey and West
(1987) t-statistics with 6 lags in parenthesis. The data is from CRSP, Compustat, and Optionmetrics and contains monthly series
from January 1996 to December 2021.

Panel A: Stock returns of portfolios of stocks sorted based on Max Pain

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P10-P1 t-stat

EW 0.000 0.001 0.000 0.000 0.001 0.001 0.001 0.001 0.000 0.002 0.002 (1.06)
VW 0.001 0.000 0.000 0.001 0.000 0.001 0.000 0.002 0.001 0.001 0.000 (0.08)

Panel B: Alphas of portfolios of stocks sorted based on Max Pain

CAPM FF3

EW 0.001 0.001
(0.91) (0.79)
VW -0.000 -0.000
(-0.09) (-0.13))

42
 Table 7. Option Order Imbalances
This table displays option order imbalance for calls and puts and the total number of trades for the positive view of the investors
(open/close buy call and sell put options), the negative view of the investors (open/close sell call and buy put options) and the
net effect which is the difference between positive and negative view. Panel A presents results for imbalances that are calculated
based on the total number of trades the week before expiration. Panel B shows results for imbalances that are calculated on the
2nd Friday of the month. We report Newey and West (1987) t-statistics with 6 lags in parenthesis. The data is from CRSP,
Compustat, and Optionmetrics and contains monthly series from January 1996 to December 2021.

Panel A: Week before expiration

I M B C al l I M B Put Posi t ive N eg at ive N et

Max Pain 1 -1.309 -0.159 28.019 29.169 -1.150

(-8.82) (-2.89) (6.80) (6.96) (-7.06)
Max Pain 5 -0.743 -0.432 23.663 23.973 -0.311
(-6.80) (-7.58) (8.42) (8.53) (-2.42)
Max Pain 10 0.495 -0.652 19.397 18.250 1.147
(3.02) (-8.48) (8.64) (8.85) (5.16)

Panel B: Formation Period (2nd Friday of the month)

I M B C al l I M B Put Posi t ive N eg at ive N et

Max Pain 1 -0.349 -0.047 5.512 5.813 -0.301

(-7.34) (-3.12) (7.51) (7.77) (-5.62)
Max Pain 5 -0.271 -0.130 4.497 4.639 -0.142
(-9.41) (-6.97) (9.67) (9.91) (-4.00)
Max Pain 10 0.160 -0.222 3.984 3.602 0.382
(3.39) (-8.02) (8.95) (9.21) (5.48)

43
 Table 8. Cross-Sectional Regressions: Option Order Imbalances
This table displays cross-sectional regressions of stock returns on lagged option order imbalances and a number of controls.
Specifically, our cross-sectional regression takes the following form:
5
X 5
X 5
X
Ret i,t = α + β j I M Bi,t−
C all
j+ γ j I M Bi,t−
Put
j+ δ j Ret i,t− j + "i,t
j=1 j=1 j=1

where Ret i,t+1 represents the stock return at time t + 1, I M B C all is the order imbalance of the call options and I M B Put is the
order imbalance of the put option. We report Newey and West (1987) t-statistics with 6 lags in parenthesis. The data is from
CRSP, Compustat and Optionmetrics and contains daily series from January 1996 to December 2021.

All Option Expirations

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Max Pain P1 Max Pain P5 Max Pain P10
C all
I M B t−1 0.297*** 0.435*** 0.069** 0.107*** 0.250*** 0.384***
(4.06) (4.56) (2.10) (2.87) (2.97) (3.44)
C all
I M B t−2 0.124 0.235** 0.017 0.013 0.101 0.138
(1.63) (2.23) (0.49) (0.27) (1.10) (1.24)
C all
I M B t−3 -0.033 -0.035 -0.060 -0.093* -0.012 -0.043
(-0.35) (-0.33) (-1.53) (-1.65) (-0.14) (-0.36)
C all
I M B t−4 0.047 0.024 -0.063** -0.081* -0.037 0.109
(0.55) (0.24) (-2.07) (-1.95) (-0.45) (1.03)
C all
I M B t−5 -0.109 -0.118 -0.084** -0.092** 0.054 -0.018
(-1.34) (-1.24) (-2.53) (-2.49) (0.64) (-0.16)
Put
I M B t−1 0.058 0.016 -0.166*** -0.178*** 0.078 0.092
(0.46) (0.10) (-3.54) (-3.19) (0.75) (0.73)
Put
I M B t−2 -0.169 -0.053 0.055 0.060 -0.318*** -0.345**
(-1.33) (-0.36) (1.35) (1.08) (-2.93) (-2.39)
Put
I M B t−3 0.138 0.110 -0.007 -0.004 -0.009 -0.040
(1.37) (0.90) (-0.16) (-0.07) (-0.08) (-0.32)
Put
I M B t−4 0.086 0.122 0.059 -0.012 0.137 0.116
(0.75) (0.82) (1.27) (-0.22) (1.28) (0.80)
Put
I M B t−5 -0.122 -0.223* 0.048 -0.023 0.040 0.113
(-1.12) (-1.66) (0.95) (-0.40) (0.39) (0.78)
Ret t−1 0.007 0.005 0.006 0.018** 0.018** 0.017** 0.009 0.009 0.009
(0.97) (0.69) (0.79) (2.23) (2.30) (2.19) (1.07) (1.18) (1.08)
Ret t−2 0.013* 0.012* 0.012* 0.013* 0.015** 0.014** 0.010 0.010 0.010
(1.85) (1.79) (1.71) (1.89) (2.23) (1.99) (1.26) (1.34) (1.25)
Ret t−3 0.003 0.002 0.004 0.001 0.001 0.003 0.015** 0.021*** 0.018**
(0.57) (0.33) (0.67) (0.11) (0.11) (0.45) (2.28) (2.92) (2.48)
Ret t−4 0.004 0.008 0.006 0.003 0.003 0.003 -0.005 -0.011* -0.007
(0.58) (1.16) (0.85) (0.40) (0.52) (0.42) (-0.74) (-1.71) (-1.00)
Ret t−5 -0.004 -0.003 -0.003 -0.003 -0.003 -0.004 -0.006 -0.009 -0.004
(-0.64) (-0.47) (-0.47) (-0.41) (-0.42) (-0.49) (-1.04) (-1.47) (-0.59)
cons -0.000 -0.000 0.000 0.001** 0.001*** 0.001** 0.000 0.000 0.000
(-0.00) (-0.18) (0.07) (2.48) (2.69) (2.54) (0.65) (0.55) (0.42)

R-squared 0.223 0.220 0.287 0.239 0.240 0.300 0.239 0.239 0.309

44
 Table 9. Cross-Sectional Regressions: Alternative Explanations
This table displays cross-sectional regressions of stock returns on the Max Pain measure and a number of controls.
Specifically, our cross-sectional regression takes the following form:

Ret i,t+1 = α + β1 M ax Paini,t + β2 Zi,t+1 + γC ont r olsi,t + "i,t+1

where Ret i,t+1 represents the stock return at time t + 1, Zi,t+1 include market maker net open interest, firm written
open interest, customer written open interest, the written volume of firms, the written volume of customers,
new delta hedging, covered call and protective put unwinding and Thursday stock price distance to strike. The
volume and open interest variables for firms and public customers are computed only from the currently expiring
call and put with the exercise price nearest to the Thursday closing stock price. The set of controls includes the
stock illiquidity, a dummy variable that takes a value of one if the stock is traded in NASDAQ, price, size, book
to market, debt to assets, stock volume (SVOL), institutional ownership (IOR), idiosyncratic volatility (IVOL),
weekly stock reversal, monthly stock reversal and momentum. We show results for the full sample, small stocks,
illiquid stocks and Nasdaq stocks. We report Newey and West (1987) t-statistics with 6 lags in parenthesis. The
data is from CRSP, Compustat and Optionmetrics and contains monthly series from January 1996 to December 2021.

Stock Returns

(1) (2)
M a x Pain 0.075*** 0.066***
(5.30) (5.52)
Market-maker net purchased open interest 0.000 0.002
(0.07) (0.61)
Firm proprietary trader written open interest 0.008* 0.001
(1.87) (0.45)
Public customer written open interest 0.001 -0.001
(0.28) (-0.42)
Firm proprietary trader open written volume 0.080*** 0.081***
(4.98) (4.16)
Public customer open written volume 0.065** 0.040**
(2.24) (2.45)
New delta hedging 0.062*** 0.060***
(8.75) (8.67)
Covered call and protective put unwinding -0.005*** -0.004***
(-10.58) (-8.14)
Thursday stock price distance to strike 0.002*** 0.001***
(4.10) (3.13)
cons 0.003* 0.001
(1.70) (0.64)

Controls No Yes
R-squared 0.082 0.198

45
 Table 10. Cross-Sectional Regressions: Net Gamma
This table displays cross-sectional regressions of stock returns on lagged values of Net Gamma and a number of
controls. Specifically, our cross-sectional regression takes the following form:
5
X 5
X
Ret i,t = α + β j Gammai,t− j + δ j Ret i,t− j + "i,t
j=1 j=1

where Ret i,t+1 represents the stock return at time t + 1, Gamma is the net gamma. We report Newey and West
(1987) t-statistics with 6 lags in parenthesis. The data is from CRSP, Compustat and Optionmetrics and contains
daily series from January 1996 to December 2021.

All Option Expirations

(1) (2) (3) (4) (5) (6)

Max Pain P1 Max Pain P5 Max Pain P10
Gamma t−1 -0.022 -0.046 0.006 0.040 0.082 0.012
(-0.70) (-1.40) (0.25) (1.27) (0.95) (0.21)
Gamma t−2 -0.017 0.035 -0.057* -0.068* 0.043 0.048
(-0.44) (0.87) (-1.91) (-1.88) (0.94) (1.10)
Gamma t−3 -0.020 -0.047 0.042 0.047 -0.055 0.017
(-0.52) (-1.24) (1.22) (1.50) (-1.05) (0.34)
Gamma t−4 0.034 0.039 0.067* 0.030 -0.045 -0.011
(0.72) (0.96) (1.67) (0.90) (-0.58) (-0.25)
Gamma t−5 0.028 0.023 -0.037 -0.040 0.001 -0.052
(0.69) (0.61) (-1.22) (-1.39) (0.01) (-1.11)
Ret t−1 0.009 0.016** 0.010
(1.25) (2.08) (1.27)
Ret t−2 0.010 0.013** 0.007
(1.47) (2.01) (1.02)
Ret t−3 0.002 -0.000 0.017***
(0.29) (-0.06) (2.72)
Ret t−4 0.001 0.002 -0.010*
(0.22) (0.38) (-1.66)
Ret t−5 -0.002 -0.006 -0.006
(-0.35) (-0.82) (-1.13)
cons -0.000 0.000 0.000 0.001** 0.000 0.000
(-0.26) (0.24) (0.97) (2.52) (0.40) (0.89)

R-squared 0.060 0.200 0.077 0.238 0.064 0.211

46
 Table 11. Put to Call Open Interest Ratios of Portfolios sorted on Max Pain
This table presents time-series averages of median put to call open interest of portfolios of stocks that are sorted based on the
Max Pain measure. We compute the ratio the day which corresponds to one week prior to the expiration of the option every
month. We report Newey and West (1987) t-statistics with 6 lags in parenthesis. The data is from CRSP and Optionmetrics and
contains monthly series from January 1996 to December 2021.

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P10-P1 t-stat

Mean 0.356 0.372 0.378 0.382 0.382 0.386 0.387 0.394 0.406 0.422 0.066 (6.65)

47
 Table 12. Stocks with and without options sorted on Max Pain
This table presents average weekly stock returns that are sorted based on the Max Pain measure. Panel A shows weekly stock
returns. We report results for stocks with and without options that are matched based on size and volume using propensity score
matching. For stocks without options, we use a sorting variable the Max Pain value of the corresponding stock pair with options
and similar size and volume. Panel B shows alphas based on CAPM and the Fama and French (1993) 3-factor model. We use
weekly factors, and we use the third weekly return of every month. Stock returns are calculated using the midpoint of the bid
and ask prices at market close adjusted for stock splits. We report Newey and West (1987) t-statistics with 6 lags in parenthesis.
The data is from CRSP, Compustat, and Optionmetrics and contains monthly series from January 1996 to December 2021.

Panel A: Portfolios of stocks sorted based on Max Pain

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P10-P1 t-stat

Stocks with Options

EW -0.001 0.001 0.002 0.003 0.002 0.002 0.002 0.003 0.003 0.003 0.005 (3.42)
VW -0.000 0.001 0.003 0.003 0.003 0.002 0.003 0.005 0.004 0.004 0.004 (2.40)

Stocks without Options

EW 0.002 0.003 0.003 0.003 0.003 0.002 0.002 0.003 0.002 0.002 -0.001 (-0.56)
VW 0.002 0.002 0.001 0.003 0.003 0.002 0.003 0.003 0.002 0.002 -0.000 (-0.30)

Panel B: Alphas of Stock Returns sorted based on Max Pain

CAPM FF3 CAPM FF3

Stocks with Options Stocks without Options

EW 0.004 0.004 -0.001 -0.001

(3.12) (3.39) (-0.66) (-1.03)
VW 0.003 0.004 -0.001 -0.001
(1.93) (2.73) (-0.74) (-0.88)

48
 Table 13. Different Number of Strikes per Stock
This table shows average stock returns that are sorted based on the Max Pain measure. We keep stock with at least two, three, or
four strikes. We estimate the stock return which corresponds to one week prior to the expiration date until the expiration of the
option every month. Panel B shows alphas based on CAPM and the Fama and French (1993) 3-factor model. We use weekly
factors and we use the third weekly return of every month. Stock returns are calculated using the midpoint of the bid and ask
prices at market close adjusted for stock splits. All portfolios are value-weighted. We report Newey and West (1987) t-statistics
with 6 lags in parenthesis. The data is from CRSP, Compustat and Optionmetrics and contains monthly series from January 1996
to December 2021.

Panel A: Portfolios of stocks sorted based on Max Pain

Low Max Pain P2 P3 P4 High Max Pain HML

Two Strikes 0.000 0.002 0.003 0.004 0.004 0.004
(0.31) (1.41) (2.53) (2.65) (2.20) (2.58)

Three Strikes -0.003 0.002 0.004 0.005 0.005 0.008

(-1.39) (1.14) (2.48) (2.80) (2.35) (3.82)

Four Strikes -0.002 0.001 0.004 0.005 0.005 0.007

(-0.66) (0.54) (2.50) (2.53) (1.70) (2.37)

Panel B: Alphas of Stock Returns sorted based on Max Pain

CAPM FF3 CAPM FF3 CAPM FF3

Two Strikes Three Strikes Four Strikes
Alphas 0.003 0.004 0.007 0.008 0.006 0.007
(2.25) (2.87) (3.61) (4.41) (2.08) (2.67)

49
 Table 14. Max Pain Theory Applied to Index Options
This table presents returns on spread portfolios of indices based on the Max Pain measure. Portfolios are constructed the second
Friday of the month. We report Newey and West (1987) t-statistics with 6 lags in parenthesis. The data is from CRSP, Compustat
and Optionmetrics and contains monthly series from January 1996 to December 2021.

Panel A: Portfolios of stock indices sorted based on Max Pain

P1 P2 P3 HML t-stat

Returns

EW 0.002 0.002 0.001 -0.000 (-0.23)

Max Pain Gain/Loss

EW -0.058 -0.003 0.141 0.199 (4.42)

Panel B: Alphas of Index Returns sorted based on Max Pain

CAPM FF3

EW -0.001 -0.001
(-0.81) (-0.59)

50
 Internet Appendix to
"No Max Pain, No Max Gain: A Case of Predictable Reversal"

by

Ilias Filippou Pedro A. Garcia-Ares Fernando Zapatero

(Not for publication)

1 Example of Max Pain Calculation

In this section, we offer an example of the Max Pain gain/loss calculation, which is the
sorting variable of our main analysis. Let us assume that one stock has three attached
call options and three put options. As it is explained in Table A1, the strike prices for the
call option are $25, $50, and $100 with a corresponding open interest of 10, 20, and 10
outstanding contracts. For simplicity, we assume that put and call options have the same
strike prices but a different number of outstanding contracts. Specifically, the put options
have open interest of 50, 25, and 1.

Table A1. Option Series for Stock A

Strike Price Call Open Interest Put Open Interest

25 10 50
50 20 25
100 10 1

In Table A2, we compute the total loss assuming that the stock price at expiration will
equal the first strike price, which is $25. We offer the strike prices of the calls and puts and
the open interest. In the Call Loss and Put Loss columns, we calculate the potential loss for
each option. At an expiration price of $25, the payoff of the call options will be zero as the
strike price exceeds or is equal to the exercise price. On the other hand, the put options
will be exercised for the strike prices of $50 and $100. In these cases, the writer of the put
will have a loss of (Stock Price-Strike Price)*Open Interest or (50-25)*25 and (100-25)*1,
respectively. Thus, the total loss will be $700. It is worth mentioning that we label as Stock
Price the assumed or theoretical stock price of $25.

Table A2. Stock Price Expires at 25

Price Strike Call OI Put OI Call Loss Put Loss Total Loss
25 25 10 50 0 0 0
50 20 25 0 (50-25) x 25 625
100 10 1 0 (100-25) x 1 75
700

1
 In Table A3, we continue with the calculation of the call and put loss, assuming that
the stock price at expiration will equal the second strike price, which is $50. Here, the
call option will be exercised for a strike price of $25 because the assumed stock price at
expiration is larger. In this case, the writer of the call option will suffer a loss of (50-25)*10.
Similarly, the put option will be exercised when the strike price is equal to $100 because
the strike price will be greater than the assumed price at expiration. The potential loss for
the put option will be (100-50)*1. Thus, the potential total loss for call and put options
will be $300.

Table A3. Stock Price Expires at 50

Price Strike Call OI Put OI Call Loss Put Loss Total Loss
50 25 10 50 (50-25)x10 0 250
50 20 25 0 0 0
100 10 1 0 (100-50)x1 50
300

In Table A4, we show results for the last strike price. We assume that the stock will
expire at $100. Only the call option will be exercised at this price, and the call option
writer will suffer a loss of $1750.

Table A4. Stock Price Expires at 100

Price Strike Call OI Put OI Call Loss Put Loss Total Loss
100 25 10 50 (100-25)x10 0 750
50 20 25 (100-50)x20 0 1000
100 10 1 0 0 0
1750

2
 In Table A5, we summarize the information obtained in the previous tables. Thus, we
report the loss of call and put writers for the three strike prices if we assume that the stock
price at expiration will equal each strike price. In the last column, we sum the call, put
losses per strike price, and select the smaller value, our Max Pain value for this stock. The
Max Pain strike price is the one that is associated with the minimum loss, which is $50.
Thus, the Max Pain Gain/Loss will be the difference between the Max Pain strike price
($50) and the stock price on the second Friday of the month –which is the day that we
form our portfolios– over the stock price the second Friday of the month.

Table A5. Max Pain Price for Stock A

Strike Price Call Loss Put Loss Total Loss

25 0 700 700
50 250 50 300
100 1750 0 1750