hjnormey/siloqy
Public Access
1
0
Fork 0
Code Issues 7 Pull Requests 2 Actions Packages Projects Releases Wiki Activity

PINK: fix fee calibration poisoning + fee_settled trade_id linkage

Browse Source 
Two related accounting fixes:

1. _calibrate_fee_model startup guard: before calling calibrate_fee,
   compute raw deviation from the published taker/maker rate (ignoring
   any stale calibration_ratio). If >15%, skip and log WARNING rather
   than letting a bad REST fill set calibration_ratio to ~0.8 and cause
   ESTIMATED fees to understate actuals by 20% for the entire session.

2. fee_settled_events trade_id: BingX WS does not echo back our
   clientOrderId in fill events (field "c" is empty). Was falling back
   to BingX's internal orderId (p-e-mq5.../p-x-mq5...) which can't
   be joined to trade_events. Now reads trade_id from kernel slot 0
   (which retains the trade_id until the next ENTER) so
   fee_settled_events.trade_id = BTCUSDT-T-N. Added venue_order_id
   field to persist_fee_settled for bidirectional reconciliation.

128/128 tests green.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Codex
2026-06-08 20:19:10 +02:00
parent c16b5aaaa4
commit 9e210b5a02
2 changed files with 40 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

39
prod/clean_arch/runtime/pink_direct.py
View File

@@ -473,6 +473,23 @@ class PinkDirectRuntime:
return
order_type = str(row.get("orderType") or row.get("type") or "MARKET").upper()
is_maker = order_type == "LIMIT"
# Guard: check raw deviation (ignoring any stale calibration_ratio) before
# mutating kernel state. A REST fill with >15% deviation from published rate
# poisons calibration_ratio and causes ESTIMATED fees to drift from actuals.
raw_rate = (
self._BINGX_FEE_CONFIG.get("maker_rate", 0.0002) if is_maker
else self._BINGX_FEE_CONFIG.get("taker_rate", 0.0005)
)
raw_expected = fill_price * fill_qty * raw_rate
raw_deviation_pct = abs(actual_fee - raw_expected) / raw_expected * 100 if raw_expected > 0 else 0.0
if raw_deviation_pct > 15.0:
self.logger.warning(
"Fee calibration SKIPPED: REST fill shows %.2f%% deviation from "
"published %.4f%% rate (expected=%.6f actual=%.6f). "
"Holding calibration_ratio=1.0 to avoid poisoning kernel fee model.",
raw_deviation_pct, raw_rate * 100, raw_expected, actual_fee,
)
return
report = self.kernel.calibrate_fee(fill_price, fill_qty, actual_fee, is_maker=is_maker)
status = report.get("calibration_status", "?")
log = self.logger.error if status == "ERROR" else self.logger.info
@@ -536,11 +553,25 @@ class PinkDirectRuntime:
# downstream can reconcile against the ESTIMATED_TAKER row.
if self.persistence is not None:
try:
# trade_id: best-effort from the client_order_id field ("c")
# or order_id ("i") — WS may not carry our trade_id directly.
ws_trade_id = str(event.client_order_id or event.order_id or "")
# BingX WS does not echo back our clientOrderId ("c" field
# is empty). Read trade_id from the kernel slot instead —
# the slot retains its trade_id until the next ENTER.
# Store BingX's own orderId as venue_order_id for
# bidirectional reconciliation.
_venue_order_id = str(event.order_id or "")
try:
_slot_dict = self.kernel.slot(0).to_dict() if self.kernel.max_slots > 0 else {}
_our_trade_id = str(_slot_dict.get("trade_id") or "")
except Exception:
_our_trade_id = ""
# Fall back: if clientOrderId was echoed (future BingX change)
# parse our trade_id prefix from "BTCUSDT-T-N:intent_id"
if not _our_trade_id:
_c = str(event.client_order_id or "")
_our_trade_id = _c.split(":")[0] if ":" in _c else (_c or _venue_order_id)
self.persistence.persist_fee_settled(
trade_id=ws_trade_id,
trade_id=_our_trade_id,
venue_order_id=_venue_order_id,
fee=event.fee,
fee_asset=event.fee_asset or "USDT",
is_maker=event.is_maker,