Exit levels on the wrong side of the entry are kept

[alexgrover]

A strategy.exit stop / limit on the wrong side of its entry was dropped unless the exit was re-placed every bar or placed with its trade already open. The drop was a heuristic ("variables scoped in an if-block are na on TradingView"), detected through the call cadence of each strategy.exit call site.

On TradingView such a leg is already breached when the entry fills, and it fills at once. Example from a public strategy (BINANCE:BTCUSDT 1h): an exit placed while flat for a long limit entry, with its stop at 64,066.01 above the 63,974.01 fill, closes the trade at that fill on the entry bar.

Change

• The wrong-side drop in processExitOrders is removed, along with the cadence tracking behind it: _isPersistent, _coversOpenTrade, _exit_call_history, and the exit call-site fallback counter.
• In a calc_on_order_fills run after a fill, exit levels snap to the tick grid relative to the fill point's price, as entries and orders already do since calc_on_order_fills: the script runs again after the bar points with fills #433.

Tests

tests/namespaces/strategy/exit-wrong-side.test.ts (2 tests): a stop above a long entry and a take-profit below it both fill at the entry fill on the entry bar. Both fail on dev.

Results against TradingView (compared with dev)

• Public strategies:
  • ema200-regime-bos-choch: first divergence trade 1 → 11;
  • harmonic-pulse-xabcd: trade 2 → 9, and its trade count now matches TradingView's (48);
  • bull-wolf and hvn-fade: same first divergence, trade counts change by one (hvn-fade now matches TradingView's 23).
• Controlled cases and premium backtesters: no change.
• Unit suite: all committed tests pass. Oracle suite: 0 regressions.

---

Note

Medium Risk
Changes core strategy.exit evaluation and removes cadence-based heuristics, which can shift trade counts on real strategies but is covered by new regression tests.

Overview
strategy.exit no longer drops stop/limit legs that sit on the “wrong” side of the entry. TradingView treats those as already breached when the entry fills and closes immediately at the fill; PineTS previously stripped them using call-site cadence heuristics (_isPersistent, _coversOpenTrade, _exit_call_history, synthetic callsite IDs).

That machinery is removed from exit.ts, types.ts, and processExitOrders in utils.ts, along with strategy-state initialization for exit cadence maps.

Mintick snapping for exit prices now uses context._fillPointPrice when the script re-runs after a fill (calc_on_order_fills), matching how entries and orders already round relative to the fill point.

Tests in exit-wrong-side.test.ts cover a long with stop above entry and take-profit below entry—both should exit on the entry bar at the entry fill price.

^{Reviewed by Cursor Bugbot for commit 793cfd1. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: fa7b1417-f5db-4247-a212-db3cc49445c2)