Files
signal-platform/tests/unit/test_signal_eval.py
T
dennisthiessen 7060b9a019
Deploy / lint (push) Successful in 6s
Deploy / test (push) Successful in 38s
Deploy / deploy (push) Successful in 25s
parallelize the backtest across worker processes (true multi-core)
The replay was CPU-bound and single-core: the earlier asyncio.to_thread offload
kept the API responsive but, because of the GIL, ran on one core. Per-ticker
replay is independent, so fan it out across worker processes (which sidestep the
GIL) for real multi-core speedup.

- New `settings.backtest_workers` (default 4), capped to cpu_count-1 so a core
  stays free for the web server.
- Uses a `forkserver` context (workers forked from a clean single-threaded
  server — avoids the fork-with-threads deadlock); falls back to `fork`. On
  spawn-only platforms (Windows) and for 1-ticker runs it uses the thread path,
  so dev/tests are unaffected.
- Worker takes primitive column arrays (cheap to pickle), rebuilds bars, and
  returns (candidates, plain-dict signal series) — both picklable across the
  process boundary. Bars are still fetched in the event loop (ORM-safe).
- Pool creation is guarded: if the pool can't start, the job falls back to the
  sequential thread path instead of failing.

334 backend tests pass (parallel path is POSIX/server-only, so it's covered by
construction + the picklability/worker-count tests; the thread fallback is
exercised by the run_backtest smoke test).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-23 23:20:20 +02:00

196 lines
8.0 KiB
Python

"""Tests for the cross-sectional signal-evaluation (factor IC) pass."""
from __future__ import annotations
import os
import pickle
import random
from datetime import date, timedelta
from types import SimpleNamespace
from app.services import backtest_service as bt
# ---------------------------------------------------------------------------
# Rank-correlation primitives
# ---------------------------------------------------------------------------
def test_spearman_monotonic_is_one():
xs = [1.0, 2.0, 3.0, 4.0, 5.0]
ys = [10.0, 20.0, 30.0, 40.0, 50.0]
assert bt._spearman(xs, ys) == 1.0
def test_spearman_inverse_is_minus_one():
xs = [1.0, 2.0, 3.0, 4.0, 5.0]
ys = [5.0, 4.0, 3.0, 2.0, 1.0]
assert bt._spearman(xs, ys) == -1.0
def test_spearman_handles_ties_without_crashing():
xs = [1.0, 1.0, 2.0, 2.0, 3.0]
ys = [1.0, 2.0, 2.0, 3.0, 3.0]
ic = bt._spearman(xs, ys)
assert ic is not None and 0.0 < ic <= 1.0
def test_spearman_none_when_degenerate():
# A flat array has zero variance → correlation undefined.
assert bt._spearman([1.0, 1.0, 1.0, 1.0], [1.0, 2.0, 3.0, 4.0]) is None
assert bt._spearman([1.0], [2.0]) is None
def test_quintile_spread_sign_follows_signal():
# signal == fwd return: top quintile clearly beats bottom → positive spread.
pairs = [(float(i), float(i)) for i in range(20)]
spread = bt._quintile_spread(pairs)
assert spread is not None and spread > 0
# Top quintile mean (17,18,19,16) - bottom (0,1,2,3) = 16.0
assert spread == (17 + 18 + 19 + 16) / 4 - (0 + 1 + 2 + 3) / 4
def test_quintile_spread_none_when_too_few():
assert bt._quintile_spread([(1.0, 1.0)] * 9) is None
# ---------------------------------------------------------------------------
# Signal value extraction (point-in-time, price-only)
# ---------------------------------------------------------------------------
def test_signal_values_momentum_and_trend():
# Steadily rising series so every lookback is positive and trend is above SMA.
closes = [100.0 * (1.01 ** k) for k in range(300)]
i = 299
vals = bt._signal_values(closes, closes, i)
assert vals["mom_12_1"] > 0 # up over the 12→1 month window
assert vals["trend_200"] > 0 # price above its 200-bar SMA in an uptrend
# 12-1 momentum skips the last month: close[i-21] / close[i-252] - 1
assert vals["mom_12_1"] == closes[i - 21] / closes[i - 252] - 1.0
# Strictly rising → today IS the 52-week high (highs==closes here) → ratio 1.0
assert vals["high_52w"] == 1.0
assert vals["vol_6m"] > 0 # realized vol is defined and positive
def test_signal_values_drops_signals_without_enough_history():
closes = [100.0 + k for k in range(80)] # only 80 bars
vals = bt._signal_values(closes, closes, 79)
assert "mom_3_1" in vals # needs 63 bars of lookback — present
assert "mom_6_1" not in vals # needs 126 — absent
assert "mom_12_1" not in vals # needs 252 — absent
assert "trend_200" not in vals # needs 200 — absent
assert "high_52w" not in vals # needs 252 — absent
assert "vol_6m" not in vals # needs 126 — absent
# ---------------------------------------------------------------------------
# End-to-end aggregation: a predictive signal scores, noise does not
# ---------------------------------------------------------------------------
def _records(closes: list[float]) -> list[SimpleNamespace]:
start = date(2020, 1, 1)
return [
SimpleNamespace(date=start + timedelta(days=k), close=c, high=c)
for k, c in enumerate(closes)
]
def test_signal_evaluation_separates_edge_from_noise():
rng = random.Random(42)
# 120 consecutive weeks, 40 names each. After non-overlapping thinning
# (stride = HORIZON/5 = 6) that leaves 20 independent windows — above the
# reliability bar. "edge" perfectly orders the forward return; "noise" is
# independent of it.
collected: dict = {"edge": {}, "noise": {}}
for week in range(120):
edge_recs = []
noise_recs = []
for _ in range(40):
fwd = rng.gauss(0, 0.05)
edge_recs.append((fwd, fwd)) # signal == fwd → IC = 1
noise_recs.append((rng.gauss(0, 1), fwd)) # signal ⟂ fwd → IC ≈ 0
collected["edge"][(2020, week)] = edge_recs
collected["noise"][(2020, week)] = noise_recs
rows = {r["signal"]: r for r in bt._signal_evaluation(collected)}
assert rows["edge"]["mean_ic"] == 1.0
assert rows["edge"]["weeks"] == 20 # 120 weeks thinned to non-overlapping
assert rows["edge"]["reliable"] is True
assert rows["edge"]["ic_positive_pct"] == 100.0
assert rows["edge"]["mean_quintile_spread"] > 0
assert abs(rows["noise"]["mean_ic"]) < 0.15 # indistinguishable from zero
# Rows are sorted by mean_ic descending: the real signal ranks first.
assert bt._signal_evaluation(collected)[0]["signal"] == "edge"
def test_signal_evaluation_flags_too_few_windows_unreliable():
# 5 adjacent weeks collapse to a single non-overlapping window → unreliable.
collected: dict = {
"edge": {(2020, w): [(float(i), float(i)) for i in range(40)] for w in range(5)}
}
row = bt._signal_evaluation(collected)[0]
assert row["weeks"] == 1
assert row["reliable"] is False
def test_nonoverlapping_weeks_thins_by_stride():
weeks = [(2020, w) for w in range(1, 13)] # 12 consecutive ISO weeks
kept = bt._nonoverlapping_weeks(weeks, stride=6)
assert kept == [(2020, 1), (2020, 7)] # 6 apart, no overlap
# Stride 1 keeps everything; ordering is chronological.
assert bt._nonoverlapping_weeks(list(reversed(weeks)), stride=1) == weeks
def test_signal_evaluation_skips_thin_weeks():
# A week with fewer than MIN_CROSS_SECTION names is ignored entirely.
collected: dict = {"edge": {(2020, 1): [(1.0, 1.0)] * (bt.MIN_CROSS_SECTION - 1)}}
assert bt._signal_evaluation(collected) == []
def test_accumulate_signal_series_emits_weekly_pairs():
closes = [100.0 * (1.005 ** k) for k in range(400)]
collected: dict = {}
from collections import defaultdict
collected = defaultdict(lambda: defaultdict(list))
bt._accumulate_signal_series(_records(closes), collected)
# The long, rising series should yield momentum + trend observations...
assert "mom_12_1" in collected and len(collected["mom_12_1"]) > 0
# ...one per ISO week, with a forward return attached to each pair.
sample = next(iter(collected["mom_12_1"].values()))
assert all(len(pair) == 2 for pair in sample)
# ---------------------------------------------------------------------------
# Parallel-replay plumbing (process pool): plain/picklable results, worker count
# ---------------------------------------------------------------------------
def test_signal_series_is_plain_and_picklable():
from collections import defaultdict
closes = [100.0 * (1.003 ** k) for k in range(400)]
series = bt._signal_series(_records(closes))
# Must be plain dicts (no defaultdict/lambda) so it survives a process boundary.
assert type(series) is dict
assert all(type(weeks) is dict for weeks in series.values())
pickle.dumps(series) # the worker's return is pickled to the parent — must not raise
# ...and equivalent to the in-place accumulator.
acc = defaultdict(lambda: defaultdict(list))
bt._accumulate_signal_series(_records(closes), acc)
assert series == {name: dict(w) for name, w in acc.items()}
def test_worker_count_caps_to_cpu_minus_one(monkeypatch):
monkeypatch.setattr(bt.settings, "backtest_workers", 1000)
assert bt._backtest_worker_count() == max(1, (os.cpu_count() or 1) - 1)
def test_worker_count_one_disables(monkeypatch):
monkeypatch.setattr(bt.settings, "backtest_workers", 1)
assert bt._backtest_worker_count() == 1
def test_mp_context_is_none_or_posix():
ctx = bt._mp_context()
# None on spawn-only platforms (Windows); a safe POSIX context otherwise.
assert ctx is None or ctx.get_start_method() in ("fork", "forkserver")