← All tasks
T14 · Factor Investing

Timing the Factor Zoo

Andreas Neuhierl et al. (4 authors), 2024 · SSRN

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

What the paper asks for.

Aggregate many factor-timing signals via Partial Least Squares with one component (PLS1, expanding window) to forecast next-month excess returns for each equity risk factor; sort factors into quintile portfolios on these forecasts and trade the equal-weighted Q5 minus Q1 long-short multifactor portfolio (H-L).

Type
ml
Implementation
pandas
Rebalance
monthly
Long-short
yes
Declared task setup
Data
One-command fetcher for WRDS-sourced 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

20 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 (7)
  • A1
    no backfill
    56/77
  • A2
    no global normalize
    59/77
  • A3
    no future data access
    56/77
  • A4
    causal consistency
    30/77
  • A8
    transaction timing
    76/77
  • A9
    signal delay test
    31/77
  • A10
    e2e metric consistency
    0/77
TYPE B · PAPER-SPECIFIC (13)
  • B1
    pls components k equals 1
    §§3.2.2 / Table 2 “partial least squares regressions with 1 component”
    provenance · paper-explicit
    58/77
  • B2
    expanding window pls estimation
    §§3.2.2 “We use each factor's first half of the sample to obtain initial estimates, and use the second half to form out-of-sample (OOS) forecasts”
    provenance · paper-explicit
    43/77
  • B3
    quintile sort on predicted returns
    §§3.3 “Each month t we sort factors into five portfolios based on their t+1 predicted excess return”
    provenance · paper-explicit
    63/77
  • B4
    high low portfolio q5 minus q1
    §§3.3 / Table 5 “we construct a high-low (H-L) portfolio”
    provenance · paper-explicit
    34/77
  • B5
    momentum signals vol scaled capped
    §Appendix B, MOM1 “Annualized return in month t scaled by annualized past return volatility over 3Y, capped at ±2.”
    provenance · paper-explicit
    63/77
  • B6
    tsmom ewma halflife 60
    §Section 2.2 and Appendix B “Ex-ante volatility is the square root of an exponentially weighted moving average of squared daily returns.”
    provenance · paper-inferred
    stage · signal gen
    67/77
  • B7
    naive benchmark historical mean sort
    §§3.3, Eq. (11) “quintile portfolios which are sorted based on the historical average factor return. We name these benchmark portfolios 'naively timed'”
    provenance · paper-explicit
    50/77
  • B8
    pls maximizes covariance
    §§3.2.1, Eq. (7) “To find the vectors v_m, we iteratively solve the following problem: v_m = arg max_v [cov(f_i, W_i v)]^2, s.t. v'v = 1”
    provenance · paper-explicit
    19/77
  • B9
    reversal signals definition
    §§2.2 “The 60 (120) month reversal signal translates into a weight w = 1 − annualized factor return over the past 60 (120) months.”
    provenance · paper-explicit
    58/77
  • B10
    oos r squared definition
    §§3.2.2, Eq. (8) “R²_OOS = 1 − Σ(f_{i,t+1} − f̂_{i,t+1})² / Σ(f_{i,t+1} − f̄_{i,t+1})² where f̄_{i,t+1} is the historical average return estimated through period t”
    provenance · paper-explicit
    34/77
  • B11
    per factor pls regression
    §§3.2.1 (footnote 6) “Note that we run a separate PLS regression for each factor to capture differential dynamics in factor risk premia.”
    provenance · paper-explicit
    58/77
  • B12
    momentum sign signals binary
    §§2.2 “we derive momentum signals that assign a weight of w_{i,t} = ±1, conditional on the sign of the past factor return over an n-months horizon”
    provenance · paper-explicit
    63/77
  • B13
    pls predictive lag signal t return t plus 1
    §§3.2.1, Eq. (5) “f_i is a T×1 vector of factor i's one-period ahead excess returns”
    provenance · paper-explicit
    29/77
What this task reveals

The lowest-passing observed requirement is e2e metric consistency: 0 of 77 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
T13 · Empirical Asset Pricing via Machine Learning
Next in Factor Investing →
T15 · Growing the Efficient Frontier on Panel Trees