You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit cda51d5
Browse filesBrowse the repository at this point in the historyBrowse files
Copy file name to clipboardExpand all lines: README.md
+8-8Lines changed: 8 additions & 8 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -26,7 +26,7 @@
26
26
27
27
TradingView's strategy tester is the reference every Pine author trusts, and nothing outside TradingView reproduced it — until now. PineForge is a C++17 runtime with a stable C ABI that runs PineScript v6 strategies exactly the way TradingView's broker emulator does: same fills, same sizing, same margin calls, same trailing stops, same `request.security()` buckets, on any OHLCV you give it, in microseconds per bar.
28
28
29
-
-**Proven, not promised.** All 4,190 probes — 312 open reference strategies plus 413 real community scripts on 15 markets and timeframes — grade *excellent* or *strong* against TradingView's own trade lists: **4,177 excellent, 13 strong, zero moderate**. The current full sweep evaluates 2,819,967 TradingView trades, with 2,817,337 matched by the verifier.
29
+
-**Proven, not promised.** All 4,190 probes — 312 open reference strategies plus 413 real community scripts on 15 markets and timeframes — grade *excellent* or *strong* against TradingView's own trade lists: **4,178 excellent, 12 strong, zero moderate**. The current full sweep evaluates 2,819,967 TradingView trades, with 2,817,448 matched by the verifier.
30
30
-**Open.** Engine, transpiler, corpus, benchmarks and the validation tooling are all public and Apache-2.0. The only thing you cannot download is the closed test set, because TradingView's Terms of Service forbid redistributing community scripts.
31
31
-**Fast.** In-process, no interpreter: median **162× faster than PyneCore** on 99 timed strategies. Parameter sweeps re-run a loaded `.so` with new inputs — no recompile, no fork.
32
32
-**Deterministic to the bit.** Two runs with the same inputs produce identical trade lists. Same on Linux and macOS.
@@ -108,18 +108,18 @@ Every PineForge-compiled strategy `.so` exports this same ABI — write the harn
108
108
109
109
## Validation scoreboard
110
110
111
-
**Round 34 · 2026-09-08:** **4,177 excellent / 13 strong / zero moderate** across all **4,190 scored probes**. This round adds one excellent result, with zero regressions on any canonical metric.
111
+
**Round 35 · 2026-09-08:** **4,178 excellent / 12 strong / zero moderate** across all **4,190 scored probes**. This round adds one excellent result, with zero regressions on any canonical metric.
112
112
113
113
| Board | Test set | Result | TradingView trades evaluated |
114
114
|---|---|---|---|
115
115
| **Public** — [open corpus](https://github.com/pineforge-4pass/pineforge-corpus) | 312 reference strategies, Apache-2.0, reproducible by anyone | **309/309 graded excellent** (ETH/USDT-perp 15m; the corpus' declared engine-only / anomaly probes are not graded) | 429,866 |
116
-
| **Closed test** — the parity campaign | 413 community-shared TradingView scripts across 15 market/timeframe lanes: **3,881 script-lane probes** — private under TradingView's Terms of Service | **3,868 excellent + 13 strong + zero moderate** = 3,881/3,881 (100%) excellent-or-strong | 2,390,101 |
116
+
| **Closed test** — the parity campaign | 413 community-shared TradingView scripts across 15 market/timeframe lanes: **3,881 script-lane probes** — private under TradingView's Terms of Service | **3,869 excellent + 12 strong + zero moderate** = 3,881/3,881 (100%) excellent-or-strong | 2,390,101 |
117
117
118
-
**2,819,967 TradingView trades** evaluated, **2,817,337 matched by the verifier** (99.91%), from the round 34 full Cloud Run sweep. **18 TradingView-side anomalies** remain excluded under the unchanged population; each was documented before exclusion. No scored probe remains below *strong*.
118
+
**2,819,967 TradingView trades** evaluated, **2,817,448 matched by the verifier** (99.91%), from the round 35 full Cloud Run sweep. **18 TradingView-side anomalies** remain excluded under the unchanged population; each was documented before exclusion. No scored probe remains below *strong*.
119
119
120
-
Round 34 corrects admission for ordinary process-on-close long MARKET orders and places the covered money-margin event at the next bar's open, before order-fill recalculation. **Overnight Gap Capture** on **OANDA:EURUSD 15m** moves from strong to excellent: zero count gap, 100% matched, and zero entry-price, exit-price, PnL and quantity error at the 90th percentile. Its **3,669 closed trades** match TradingView on entry/exit time, side, quantity and prices.
120
+
Round 35 corrects ordinary pairs of unlinked stop-entry orders placed while flat. After the first stop opens a position, the later stop consumes its own transaction quantity: it can partially close, flatten, or reverse only the remainder, in either path direction. Default percent-of-equity stops retain their placement quantity consistently through admission and execution. **Decoded Volatility Expansion** on **OANDA:XAUUSD 15m** moves from strong to excellent: zero count gap, 100% canonical match, and zero entry-price, exit-price, PnL and quantity error at the 90th percentile. All **514 closed-trade identities** match TradingView on entry/exit time, side, quantity and prices.
121
121
122
-
The changes use broker state, physical order/position provenance and instrument tick/lot facts. A four-cell Cloud experiment separates admission from opening-event timing; the combined change is required for the conversion. The other **4,189 trade CSVs are byte-identical** to round 33, and their canonical grades and metrics are unchanged, including all **704 hard-surface probes**. Codegen, verifier, grading rules, profile-selection code, reference tapes, feeds, input files and scored population are unchanged.
122
+
The rule uses order-book and position provenance, with no strategy, symbol or date lookup. Sixteen independent TradingView controls and an unchanged-output Cloud trace pin the cause. Seven hard-surface corpus probes gain **72 exact trade matches**, with no previously exact match lost; their canonical grades and metrics remain unchanged. The other **4,182 trade CSVs are byte-identical** to round 34, and all **4,189 other probes** retain their canonical grades and quantity metrics. Verifier code, grading rules, profile-selection code, reference tapes, feeds, input files and scored population are unchanged.
123
123
124
124
### The closed test, lane by lane
125
125
@@ -138,9 +138,9 @@ The changes use broker state, physical order/position provenance and instrument
0 commit comments