← All tasks
T12 · Factor Investing

Factor Momentum and the Momentum Factor

Sina Ehsani and Juhani T. Linnainmaa, 2022 · The Journal of Finance

49%
pass@1 across 16 systems
15/16
systems with ≥1 successful attempt
5/5
5/5 attempts passed for the best system
80 attempts · 16 systems × 5
Claude Code + Opus 5
60%
Claude Code + Opus 4.8
60%
Claude Code + Opus 4.7
40%
Codex + GPT-5.6-sol
100%
Codex + GPT-5.5
60%
OpenCode + Kimi-K3
60%
OpenCode + Opus 4.7
100%
OpenCode + GPT-5.5
20%
OpenCode + Sonnet 4.6
100%
OpenCode + Qwen3.7-Max
0%
OpenCode + GLM-5.2
20%
OpenCode + DeepSeek V4-Pro
20%
OpenCode + GLM-5.1
60%
OpenCode + Qwen 3.6 Plus
40%
OpenCode + MiniMax-M3
20%
OpenCode + MiniMax M2.7
20%
all checkers pass Type-A clean, Type-B fail fail checking timeout no result / incomplete
Strategy

What the paper asks for.

Time-Series Factor Momentum (TSFM): for each non-momentum factor, take a long (short) position when its trailing 12-month excess return is positive (negative), then equal-weight the resulting per-factor sleeves into a monthly-rebalanced long-short portfolio.

Type
technical
Implementation
pandas
Rebalance
monthly
Long-short
yes
Declared task setup
Data
Bundled public task data
Operational scope
The paper-defined methodological core is retained; the fixed instruction instantiates only the data mappings, interfaces, and evaluation conventions needed for execution.
Six-stage quantitative strategy pipeline
  1. 01 Data loading
  2. 02 Feature engineering
  3. 03 Model fitting
  4. 04 Signal generation
  5. 05 Portfolio construction
  6. 06 Execution
Checker suite

16 properties · 80 attempts.

Each row is one requirement the strategy must satisfy. The bar reports the share of observed checker verdicts that pass; missing checker output is not converted into a property verdict.

TYPE A · DOMAIN-LEVEL (8)
  • A1
    no backfill
    stage · data loading
    77/80
  • A3
    no future data access
    stage · feature eng
    54/80
  • A4
    causal consistency
    stage · signal gen
    46/80
  • A5
    no survivorship bias
    stage · data loading
    80/80
  • A7
    data frequency alignment
    stage · data loading
    80/80
  • A8
    transaction timing
    stage · execution
    54/80
  • A9
    signal delay test
    stage · signal gen
    55/80
  • A10
    e2e metric consistency
    71/80
TYPE B · PAPER-SPECIFIC (8)
  • B1
    lookback window months
    §§3.1 “the explanatory variable is an indicator variable for the factor's performance over the prior year from month t-12 to t-1”
    provenance · paper-explicit
    stage · feature eng
    55/80
  • B2
    formation window excludes current month
    §§3.1 “from month t-12 to t-1”
    provenance · paper-explicit
    stage · feature eng
    54/80
  • B3
    signal rule sign against zero
    §Table 3 caption “long factors with positive returns over the prior one-year period (winners) and short factors with negative returns (losers)”
    provenance · paper-explicit
    stage · signal gen
    69/80
  • B4
    weighting equal weight
    §Table 3 caption “The equal-weighted portfolio invests in all factors with the same weights”
    provenance · paper-explicit
    stage · portfolio
    80/80
  • B5
    weight normalization by count
    §Table 3 caption “long 11.0 factors and short 5.8 factors”
    provenance · paper-explicit
    stage · portfolio
    64/80
  • B6
    universe excludes umd momentum factors
    §p.8 “We exclude the two stock momentum factors, U.S. and global UMD, from the set of factors to avoid inducing a mechanical correlation between factor momentum and individual stock momentum”
    provenance · paper-explicit
    stage · data loading
    75/80
  • B7
    signal uses time series not cross sectional
    §§3.1 / Table 3 caption “long factors with positive returns over the prior one-year period (winners) and short factors with negative returns”
    provenance · paper-explicit
    stage · signal gen
    80/80
  • B8
    no skip month in formation
    §§3.1 “from month t-12 to t-1”
    provenance · paper-explicit
    stage · feature eng
    55/80
What this task reveals

The lowest-passing observed requirement is causal consistency: 46 of 80 observed checker verdicts pass. The full property list above shows whether errors concentrate in domain-level validity or the paper-specific strategy logic.

Open paper → All tasks
← Previous in Factor Investing
T11 · Mimicking Finance: Predicting Mutual Fund Manager Behavior Using Deep Learning
Next in Factor Investing →
T13 · Empirical Asset Pricing via Machine Learning