gahow/quant-trading
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(backtest): settle delisted US holdings (engine v3), PIT membership before 2011, registry causality test

Browse Source 
- execution/trader: a held US stock with no print for 5 sessions that has left
  the index is sold at its last print (execution.SETTLE_AFTER_SESSIONS), in both
  the batch replay and the live session ledger. Acquired names used to stay at
  their last mark forever: lowrisk_top20 ended 2017-12..2026-10 with 11 of 26
  dead positions, and CAGR rises 1.7-5.5pp once they settle. ENGINE_VERSION
  cash-ledger-pit-v3; the API cache key follows.
- universe_history: current members whose pre-2011 addition the Wikipedia change
  log missed now start at the list page's "Date added" (80 of 94 current members
  added 2001-2010 were open-start, e.g. ISRG, MA, BKNG). Membership from 2016 on
  is unchanged; the research panel cache now keys on the membership intervals.
- research/lab/registered: pure-ETF strategies replay on the ETF panel (they ran
  as all-cash books or crashed); prepare_strategy refuses an ETF strategy that
  lacks its ETF universe.
- research/lab/factors: trend_hzz uses a month's regression betas only from the
  session its forward returns end (short months read 1-2 sessions ahead).
- tests: every registered strategy (US and CN) is truncation-invariant on the
  production signal path; batch and live ledgers settle a delisting identically.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Gahow Wang
2026-10-10 00:21:49 -07:00
parent a9683b6d1f
commit cd0721eada
12 changed files with 279 additions and 41 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
research/lab/factors.py
View File

@@ -445,9 +445,12 @@ def trend_hzz(p):
b = np.linalg.lstsq(Xo, y[ok].values, rcond=None)[0][1:]
betas.append(pd.Series(b, index=lags, name=d))
B = pd.DataFrame(betas)
# the regression at month d uses returns through d+21, so it is usable at d+21;
# shift by one month, then average the last 12 known betas
Ebeta = B.shift(1).rolling(12, min_periods=6).mean()
# the regression at month-end d uses returns through d+21 sessions, so its
# betas are known from that session on (a month can be shorter than 21
# sessions); average the last 12 known betas
known = c.index.get_indexer(B.index) + 21
B = B[known < len(c.index)].set_axis(c.index[known[known < len(c.index)]], axis=0)
Ebeta = B.rolling(12, min_periods=6).mean()
Ed = Ebeta.reindex(c.index, method="ffill")
return sum(feats[L].mul(Ed[L], axis=0) for L in lags)

12
research/lab/panel.py
View File

@@ -10,6 +10,7 @@ Panels are cached in data/research/*.panel.pkl; delete them after refetching.
from __future__ import annotations
import hashlib
import json
import os
import pickle
@@ -103,11 +104,18 @@ def _tiingo_sig() -> list:
return [len(files), round(mtime, 3)]
def _membership_sig() -> str:
with open(os.path.join("data", "sp500_history.json")) as fh:
intervals = json.load(fh)["intervals"]
return hashlib.sha256(json.dumps(intervals, sort_keys=True).encode()).hexdigest()[:16]
def load_us() -> Panel:
"""Cached US panel, rebuilt whenever the Tiingo overlay changes."""
"""Cached US panel, rebuilt whenever the Tiingo overlay or the S&P 500
membership intervals change."""
path = os.path.join(RAW_DIR, "us.panel.pkl")
key = os.path.join(RAW_DIR, "us.panel.key")
sig = _tiingo_sig()
sig = _tiingo_sig() + [_membership_sig()]
if os.path.exists(path) and os.path.exists(key):
with open(key) as fh:
if json.load(fh) == sig:

37
research/lab/registered.py
View File

@@ -11,8 +11,9 @@ import pandas as pd
import metrics
import data_manager
from backtest_runner import (ENGINE_VERSION, load_stock_panels, prepare_strategy, replay_prepared)
from trader import make_strategy
from backtest_runner import (ENGINE_VERSION, load_etf_panels, load_stock_panels, prepare_strategy,
replay_prepared)
from trader import ETF_STRATEGY_UNIVERSES, make_strategy
from strategies.robust_momentum import RobustMomentumStrategy
from research.lab.trials import append_trials
@@ -26,8 +27,19 @@ def evaluate_registered(args):
market = 'cn' if args.market.startswith('cn') else 'us'
destination = Path(args.output or f'data/research/results/registered_{market}')
destination.mkdir(parents=True, exist_ok=True)
panels = load_stock_panels(market, args.registered, None, refresh=False)
end = args.end or str(panels['panels']['close'].index[-1].date())
etf_names = [n for n in args.registered if n in ETF_STRATEGY_UNIVERSES]
stock_names = [n for n in args.registered if n not in ETF_STRATEGY_UNIVERSES]
stock = load_stock_panels(market, stock_names, None, refresh=False) if stock_names else None
etfs = load_etf_panels(market, etf_names, refresh=False) if etf_names else None
def source(name):
"""(panels, universe, membership) the strategy is replayed on, as in run_market."""
if name in ETF_STRATEGY_UNIVERSES:
return etfs, ETF_STRATEGY_UNIVERSES[name], None
return stock['panels'], stock['stocks'], stock['member']
ends = [p['close'].index[-1] for p in ([stock['panels']] if stock else []) + ([etfs] if etfs else [])]
end = args.end or str(min(ends).date())
start = args.start or '2018-01-01'
profiles = ['baseline', 'resize20', 'buffer3'] + (['cash_shy'] if market == 'us' else [])
protocol = {
@@ -40,8 +52,10 @@ def evaluate_registered(args):
for p in [Path('execution.py'), Path('signals.py'), Path('backtest_runner.py')]
+ sorted(Path('strategies').glob('*.py'))},
'data_hashes': {str(p): hashlib.sha256(p.read_bytes()).hexdigest()
for p in [Path('data') / f'{market}.csv', Path('data') / f'{market}_open.csv',
Path('data/sp500_history.json' if market == 'us' else 'data/csi300_history.json')]},
for p in ([Path('data') / f'{market}.csv', Path('data') / f'{market}_open.csv',
Path('data/sp500_history.json' if market == 'us' else 'data/csi300_history.json')]
if stock_names else [])
+ ([Path('data/etfs.csv'), Path('data/etfs_open.csv')] if etf_names else [])},
}
protocol_path = destination / 'protocol.json'
if protocol_path.exists():
@@ -52,7 +66,8 @@ def evaluate_registered(args):
'fee_base': 1.88 if market == 'us' else 0, 'fee_per_share': .009 if market == 'us' else 0}
for name in args.registered:
base_strategy = make_strategy(name, market)
base_plan = prepare_strategy(name, market, panels['panels'], panels['stocks'], member=panels['member'],
panels, universe, member = source(name)
base_plan = prepare_strategy(name, market, panels, universe, member=member,
start=start, end=end, strategy=base_strategy)
if base_plan['warmup_rows'] < 315:
raise ValueError('Registered comparison requires >=315 pre-window sessions; move --start or add history')
@@ -66,17 +81,17 @@ def evaluate_registered(args):
if not isinstance(base_strategy, RobustMomentumStrategy):
continue
changed = copy.deepcopy(base_strategy); changed.buffer = 3.0
plan = prepare_strategy(name, market, panels['panels'], panels['stocks'], member=panels['member'],
plan = prepare_strategy(name, market, panels, universe, member=member,
start=start, end=end, strategy=changed)
elif profile == 'cash_shy':
if 'SHY' not in panels['panels']['close']:
if 'SHY' not in panels['close']:
raise ValueError('SHY bars are required for a traded cash sleeve')
plan = dict(base_plan)
plan['prices'] = base_plan['prices'].copy()
plan['prices']['SHY'] = panels['panels']['close']['SHY'].reindex(dates)
plan['prices']['SHY'] = panels['close']['SHY'].reindex(dates)
if base_plan.get('volume') is not None:
plan['volume'] = base_plan['volume'].copy()
plan['volume']['SHY'] = panels['panels']['volume']['SHY'].reindex(dates)
plan['volume']['SHY'] = panels['volume']['SHY'].reindex(dates)
target = base_plan['targets'].reindex(columns=plan['prices'].columns, fill_value=0).copy()
target['SHY'] += (1-target.sum(axis=1)).clip(lower=0)
plan['targets'] = target