← All tasks
T27 · Systematic Allocation

Out-of-Sample Equity Premium Prediction: Combination Forecasts and Links to the Real Economy

David E. Rapach et al. (3 authors), 2010 · Review of Financial Studies

68%
pass@1 across 16 systems
14/16
systems with ≥1 successful attempt
5/5
5/5 attempts passed for the best system
80 attempts · 16 systems × 5
Claude Code + Opus 5
100%
Claude Code + Opus 4.8
100%
Claude Code + Opus 4.7
80%
Codex + GPT-5.6-sol
100%
Codex + GPT-5.5
100%
OpenCode + Kimi-K3
0%
OpenCode + Opus 4.7
100%
OpenCode + GPT-5.5
100%
OpenCode + Sonnet 4.6
40%
OpenCode + Qwen3.7-Max
100%
OpenCode + GLM-5.2
40%
OpenCode + DeepSeek V4-Pro
80%
OpenCode + GLM-5.1
20%
OpenCode + Qwen 3.6 Plus
60%
OpenCode + MiniMax-M3
60%
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.

Combine 15 recursive Welch-Goyal equity-premium forecasts by equal-weighted mean and use the resulting quarterly forecast in a constrained mean-variance allocation between the S&P 500 value-weighted index and the risk-free asset.

Type
fundamental
Implementation
pandas
Rebalance
quarterly
Long-short
no
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

18 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 (9)
  • A1
    no backfill
    stage · data loading
    68/69
  • A2
    no global normalize
    stage · feature eng
    69/69
  • A3
    no future data access
    stage · model fit
    68/69
  • A4
    causal consistency
    stage · signal gen
    69/69
  • A6
    train test temporal order
    stage · model fit
    64/69
  • A7
    data frequency alignment
    stage · data loading
    65/69
  • A8
    transaction timing
    stage · execution
    67/69
  • A9
    signal delay test
    stage · signal gen
    65/69
  • A10
    e2e metric consistency
    62/69
TYPE B · PAPER-SPECIFIC (9)
  • B1
    predictor count is 15
    §Abstract / §2 “We analyze the forecasting performance of combinations for the equity premium based on 15 economic variables from Goyal and Welch (2008).”
    provenance · paper-explicit
    stage · feature eng
    69/69
  • B2
    dp formula
    §§2 “Log dividend-price ratio (log(DP)): log of dividends (D) minus log of prices, where dividends are 12-month moving sums of dividends paid on the S&P 500 and prices are quarter-end levels of the S&P 500.”
    provenance · paper-explicit
    stage · feature eng
    69/69
  • B3
    dy uses lagged price
    §§2 “Log dividend yield (log(DY)): log of dividends minus the log of lagged prices.”
    provenance · paper-explicit
    stage · feature eng
    69/69
  • B4
    infl lagged one quarter
    §§2 footnote “Following Campbell (1987) and Rapach and Wohar (2006), among many others, we lag inflation one period to ensure data availability.”
    provenance · paper-explicit
    stage · data loading
    68/69
  • B5
    univariate predictive regression expanding ols
    §§2 eq. (6) “\(\hat{r}_{i,t+1} = \hat{\alpha}_{i,t} + \hat{\beta}_{i,t} x_{i,t}\), where \(\hat{\alpha}_{i,t}\) and \(\hat{\beta}_{i,t}\) are OLS estimates of \(\alpha_i\) and \(\beta_i\) from regressing \(\{r_s\}_{s=2}^{t}\) on a constant and \(\{x_{i,s-1}\}_{s=2}^{t}\).”
    provenance · paper-explicit
    stage · model fit
    62/69
  • B6
    combination mean equal weights
    §§3 eq. (8) “The mean combination forecast weights each individual forecast equally: \(\hat{r}_{c,t+1} = \frac{1}{N} \sum_{i=1}^{N} \hat{r}_{i,t+1}\)”
    provenance · paper-explicit
    stage · signal gen
    69/69
  • B7
    variance window is 10 year rolling
    §Section 5 “estimate the variance of equity premium returns using a ten-year rolling window”
    provenance · paper-explicit
    stage · signal gen
    69/69
  • B8
    risk aversion gamma is 3
    §§5 “Consistent with prior research, we set the coefficient of risk aversion to γ=3.”
    provenance · paper-explicit
    stage · signal gen
    69/69
  • B9
    weight bounds are 0 to 1 5
    §§5 “We constrain \(w_{t+1}\) to lie between 0 and 1.5, preventing short sales and limiting leverage to 50%.”
    provenance · paper-explicit
    stage · signal gen
    69/69
What this task reveals

The lowest-passing observed requirement is univariate predictive regression expanding ols: 62 of 69 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 Systematic Allocation
T26 · Shrinking the Cross-Section
Next in Systematic Allocation →
T28 · Universal Portfolio Shrinkage