← All tasks
T19 · Volatility Management

Downside risk and the performance of volatility-managed portfolios

Feifei Wang and Xuemin Sterling Yan, 2021 · Journal of Banking and Finance

88%
pass@1 across 16 systems
16/16
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
100%
Codex + GPT-5.6-sol
100%
Codex + GPT-5.5
100%
OpenCode + Kimi-K3
100%
OpenCode + Opus 4.7
100%
OpenCode + GPT-5.5
100%
OpenCode + Sonnet 4.6
100%
OpenCode + Qwen3.7-Max
100%
OpenCode + GLM-5.2
60%
OpenCode + DeepSeek V4-Pro
80%
OpenCode + GLM-5.1
100%
OpenCode + Qwen 3.6 Plus
60%
OpenCode + MiniMax-M3
60%
OpenCode + MiniMax M2.7
40%
all checkers pass Type-A clean, Type-B fail fail checking timeout no result / incomplete
Strategy

What the paper asks for.

Scaling each equity factor's monthly return by the inverse of its prior-month downside realized volatility (square-root of sum of squared negative daily returns), normalised so the managed and unmanaged factor share the same full-sample volatility, generates a downside-volatility-managed factor portfolio with higher Sharpe ratio than the unmanaged factor.

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

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 (7)
  • A1
    no backfill
    stage · data loading
    74/80
  • A3
    no future data access
    stage · feature eng
    78/80
  • A4
    causal consistency
    stage · signal gen
    77/80
  • A7
    data frequency alignment
    stage · data loading
    79/80
  • A8
    transaction timing
    stage · execution
    77/80
  • A9
    signal delay test
    stage · signal gen
    77/80
  • A10
    e2e metric consistency
    76/80
TYPE B · PAPER-SPECIFIC (11)
  • B1
    downside vol uses only negative returns
    §§2.2, Eq. 3 “σ_{Down,t} = sqrt(Σ_{j=1}^{N_t} f_j² I[f_j < 0]) ... we compute downside volatility using only negative daily returns in month t.”
    provenance · paper-explicit
    stage · feature eng
    79/80
  • B2
    volatility computed within calendar month
    §§2.2, Eq. 3 “f_j represents the return on day j in month t, and N_t is the number of daily returns in month t.”
    provenance · paper-explicit
    stage · feature eng
    79/80
  • B3
    managed return formula with lagged vol
    §§2.2, Eq. 5 “f_{σ_Down,t} = (c̃*/σ_{Down,t-1}) f_t”
    provenance · paper-explicit
    stage · signal gen
    78/80
  • B4
    monthly rebalancing frequency
    §§2.2, Eq. 5 “f_{σ_Down,t} = (c̃*/σ_{Down,t-1}) f_t, where σ_{Down,t-1} is the downside volatility in month t-1”
    provenance · paper-explicit
    stage · execution
    79/80
  • B5
    factor universe 8 factors
    §§2.1 “the nine equity factors considered by Moreira and Muir (2017), i.e., the market (MKT), size (SMB), and value (HML) factors... MOM... RMW and CMA... ROE and IA... and BAB”
    provenance · benchmark-specified
    stage · data loading
    79/80
  • B6
    realized vol is sqrt sum squared not std
    §§2.2, Eq. 2–3 “σ_{Total,t} = sqrt(Σ_{j=1}^{N_t} f_j²) and σ_{Down,t} = sqrt(Σ_{j=1}^{N_t} f_j² I[f_j<0])”
    provenance · paper-explicit
    stage · feature eng
    79/80
  • B7
    c star computation via vol matching
    §§2.2, below Eq. 1 “c* is a constant chosen such that f_t and f_{σ,t} have the same full-sample volatility.”
    provenance · paper-explicit
    stage · signal gen
    77/80
  • B8
    edge case few negative returns
    §§2.2 “If the number of negative daily returns is less than three in month t, then σ_{Down,t} is measured using negative daily returns over both month t and month t-1.”
    provenance · paper-explicit
    stage · feature eng
    78/80
  • B9
    uses raw not demeaned returns
    §§2.2, Eq. 2–3 “σ_{Total,t} = sqrt(Σ_{j=1}^{N_t} f_j²) ... σ_{Down,t} = sqrt(Σ_{j=1}^{N_t} f_j² I[f_j<0])”
    provenance · paper-explicit
    stage · feature eng
    79/80
  • B10
    scaling by inverse vol not variance
    §§2.2, Eq. 5 + Footnote 7 “Moreira and Muir (2017) scale factor returns by lagged realized variance. We decide to use lagged realized volatility primarily because it leads to less extreme investment weights and hence lower turnover and trading cost.”
    provenance · paper-explicit
    stage · signal gen
    78/80
  • B11
    rf not added in main
    §Section 2, Equations 1 and 5 “where f_t is the monthly excess return for the original portfolio”
    provenance · paper-explicit
    80/80
What this task reveals

The lowest-passing observed requirement is no backfill: 74 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 Volatility Management
T18 · Conditional Volatility Targeting
Next in Volatility Management →
T20 · Option Momentum