Overall top-100 hits
26.14
average across horizons
Predicted top-100 avg actual rank
225.71
lower is better
Rank IC
0.0808
picks the better of two stocks
52.6% of the time
IC reliability
13.53
t-stat;
IC positive 75.5% of months
not adjusted for overlapping windows
not adjusted for overlapping windows
Horizon Breakdown
top 100 only| Horizon | Average Top-100 Hits | Predicted Top-100 Avg Actual Rank | Rank IC |
|---|---|---|---|
| 3m | 25.53 | 231.11 | 0.0617 |
| 6m | 26.39 | 224.73 | 0.0876 |
| 12m | 26.51 | 221.29 | 0.0931 |
Run Note
test summaryHypothesis: rank IC and hits@100 want different models, so build directly for IC. Scanning every column as a date-local rank shows from_low_252d alone reaches IC 0.0590 while vol_63d -- which drives nearly all of the hits@100 score -- is only 0.0179 at t 1.55. Four signals are individually positive in all 3 time blocks and both halves and are near-uncorrelated (from_low_252d vs volume 0.02): from_low_252d, volume, free_cash_flow_margin_pct, revenue_growth_pct. Combined as a plain unweighted average of date-local rank percentiles with zero fitted parameters. Change from strategy_000079: drops all 11 fitted family weights and the z-score construction. Result IC 0.08077 (t 13.53, positive 75.5%) vs 079's 0.06572 (t 8.81, 67.5%), a 23% improvement, and it beats 079 in all 3 blocks and both halves (0.1026/0.0492/0.0907 vs 0.0764/0.0419/0.0791). Avg actual rank improves to 225.71 from 232.69. Hits@100 falls to 26.144 from 27.965 because the volatility term that pays for hits is gone. IMPROVED on IC and on avg actual rank, worse on hits. Nested holdout: selecting on b1+b2 and scoring untouched b3, and the reverse, picks this same family both times and it beats 079 out of sample in each. Also found: 079's second largest IC contributor is a coverage artifact -- a bare has_market_cap indicator scores IC 0.0459 t 11.89 while market cap with missing left neutral scores 0.0044 t 0.99. Next: the price-only half of this model (from_low_252d + volume, both 100% coverage, no fundamentals) already scores IC 0.0712 t 10.48, beating 079 with no coverage artifact at all -- worth isolating.
Run Details
strategy_test_results #228| id | 228 |
|---|---|
| strategy_id | strategy_000109 |
| strategy_title | Robust Rank Core: Four Signals Positive In Every Sub-Period |
| strategy_hash | 50390b52b1ab003208e0e5ff0c281799cc56149b0a27acf43828e4f0935dbddf |
| evaluator_version | v3_top100_monthly |
| start_date | 2015-01-01 |
| end_date | 2025-06-01 |
| created_at | 2026-07-25 13:44:40 |
| updated_at | 2026-07-25 13:44:40 |
Sample Top-100 Lists
2020-03-31 / 6m / 27 hitsPredicted top 100
AAPL, ABBV, ABT, ADBE, AES, AFL, AIG, AMD, AMT, AMZN, ANET, AVGO, BKNG, BMY, BSX, CAG, CCL, CDNS, CF, CI, CMCSA, CMG, CNC, CNP, CRM, CSCO, DAL, DRI, DVN, DXC, DXCM, ETSY, EXC, EXPE, EXR, FANG, FITB, FTNT, GILD, GOOGL, HPE, HWM, INTC, ISRG, IVZ, KEY, KLAC, KMI, KO, KR, LLY, LNC, LRCX, LYV, MA, MCHP, MCO, MDT, MET, MGM, MO, MPC, MS, MSFT, NCLH, NEE, NEM, NFLX, NKE, NOW, NVDA, NWL, ODFL, OKE, ORCL, OXY, PLD, PPL, PRU, PYPL, QCOM, RCL, ROST, RTX, SPG, SPGI, SRE, SYF, SYK, SYY, TER, TJX, TSLA, UAL, URI, V, VRTX, VTR, WELL, WMB
Actual top 100
AAP, AAPL, ABMD, ADBE, ADSK, ALB, ALGN, AMD, AMP, AMZN, APD, APH, APTV, ATVI, AVGO, BBY, BWA, CARR, CDNS, CE, CHRW, CMG, CMI, CPRT, CRM, CTAS, CTLT, CTSH, DD, DE, DHI, DHR, DRI, DXCM, EBAY, EMN, ETSY, EXPE, FCX, FDX, GLW, GPS, HAL, HD, IDXX, IPGP, JCI, KMX, LEG, LEN, LH, LOW, MAS, MCHP, MGM, MOS, NCLH, NKE, NOW, NSC, NVDA, NWS, NWSA, ORLY, PAYC, PH, PHM, PNR, POOL, PVH, PWR, PYPL, QCOM, QRVO, RCL, ROK, ROL, SHW, SIVB, SNPS, SWK, SWKS, SYF, TDG, TEL, TER, TGT, TMO, TSCO, TSLA, TT, UPS, URI, VAR, VIAC, VTR, WHR, WMB, WST, WY
Matched tickers
AAPL, ADBE, AMD, AMZN, AVGO, CDNS, CMG, CRM, DRI, DXCM, ETSY, EXPE, MCHP, MGM, NCLH, NKE, NOW, NVDA, PYPL, QCOM, RCL, SYF, TER, TSLA, URI, VTR, WMB
Strategy Script
50390b52b1"""Generated V3 stock-ranking strategy."""
from __future__ import annotations
import numpy as np
import pandas as pd
STRATEGY_ID = "strategy_000109"
STRATEGY_TITLE = "Robust Rank Core: Four Signals Positive In Every Sub-Period"
DESCRIPTION = """
Built directly against the rank IC metric rather than hits@100. Scoring every
column as a date-local rank shows the IC leaderboard is nothing like the hits
leaderboard: from_low_252d alone reaches IC 0.0590 while vol_63d, which drives
almost all of the hits@100 score, is only 0.0179 at t 1.55.
Four signals are individually positive in all three time blocks and both halves,
and they are close to mutually uncorrelated (from_low_252d vs volume is 0.02):
from_low_252d distance held above the 252-day low
volume traded share volume
free_cash_flow_margin_pct cash conversion
revenue_growth_pct top-line growth
They are combined as a plain unweighted average of date-local rank percentiles.
There is no fitted parameter anywhere in the model, which matters because every
prior attempt to fit this objective has lost to unfitted rank composites out of
sample. Perturbing any single weight to 0.5, 0.75, 1.5 or 2.0 makes the result
worse in both directions, so equal weighting sits on a plateau rather than on a
tuned peak.
Two of the four carry no fundamental data at all, so half the model is immune to
the coverage artifact that inflates the incumbent: strategy_000079's second
largest IC contributor is its log-market-cap term, and a bare has_market_cap
indicator scores IC 0.0459 at t 11.89 while market cap ranked with missing
values left neutral scores 0.0044 at t 0.99. That term is measuring which rows
have fundamentals in the database, not size.
"""
FACTORS_USED = [
"from_low_252d",
"volume",
"free_cash_flow_margin_pct",
"revenue_growth_pct"
]
PARAMETERS = {
"from_low_252d": 1.0,
"volume": 1.0,
"free_cash_flow_margin_pct": 1.0,
"revenue_growth_pct": 1.0
}
COMPLEXITY = 4
PARENT_STRATEGY_ID = "strategy_000079"
GENERATION = 109
RANDOM_SEED = 20261090
CREATED_AT = "2026-07-25T00:00:00+00:00"
def _date_group(df):
if "trade_date" in df.columns:
return df["trade_date"]
return pd.Series(0, index=df.index)
def _rank_pct(df, column):
"""Date-local rank percentile centred on zero; missing values stay neutral."""
if column not in df.columns:
return pd.Series(0.0, index=df.index)
clean = df[column].replace([np.inf, -np.inf], np.nan)
ranked = clean.groupby(_date_group(df)).rank(pct=True)
return ranked.fillna(0.5) - 0.5
def calculate_score(df):
score = pd.Series(0.0, index=df.index)
for column, weight in PARAMETERS.items():
score = score + weight * _rank_pct(df, column)
total = sum(abs(w) for w in PARAMETERS.values()) or 1.0
score = score / total
return score.replace([np.inf, -np.inf], np.nan).fillna(0.0)