Evaluates signal half-life, turnover implications, and capacity constraints for systematic alpha factors. Use when analyzing signal persistence, evaluating factor decay, or estimating strategy capacity.
インストール
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
Evaluates signal half-life, turnover implications, and capacity constraints for systematic alpha factors. Use when analyzing signal persistence, evaluating factor decay, or estimating strategy capacity.
Evaluating whether an alpha signal retains predictive power across different holding periods
Determining optimal rebalance frequency by measuring how quickly signal strength deteriorates
Estimating strategy capacity limits before market impact erodes expected returns
Comparing decay profiles across candidate factors during signal selection or portfolio construction
Assessing whether a live signal has degraded relative to its backtested decay curve (alpha erosion monitoring)
Inputs To Gather
Signal return series: Period-by-period returns for portfolios sorted by the signal (e.g., decile or quintile long/short spreads)
Holding-period returns matrix: Returns measured at 1-day, 5-day, 10-day, 21-day, 63-day, and 126-day forward windows
Turnover data: Portfolio turnover rate at each rebalance frequency; round-trip transaction cost estimate (bps)
Universe and dates: Investable universe definition, backtest start/end dates, any regime breaks or structural changes
AUM or notional size: Current or projected strategy size for capacity analysis
Market impact model parameters: Participation rate assumptions, ADV percentiles, spread estimates [VERIFY — impact model choice varies by asset class and execution venue]
Workflow
Construct the decay curve
Compute the information coefficient (IC) or long/short spread return at each forward horizon (1d through 126d)
Plot IC vs. holding period; fit an exponential decay model IC(t) = IC₀ · e^(−λt) to estimate the decay constant λ
Derive the signal half-life: t½ = ln(2) / λ
Report confidence intervals around half-life using bootstrapped IC series
Assess turnover cost breakeven
Calculate implied turnover at each rebalance frequency from the signal's rank-change rate
Multiply turnover by estimated round-trip cost to get the cost drag per period
Identify the breakeven rebalance frequency where gross alpha minus cost drag is maximized
Flag if optimal rebalance is faster than operationally feasible settlement or execution cycles [VERIFY — settlement cycles differ by market and instrument type]
Estimate capacity constraints
For each rebalance, estimate total dollars traded per period
Apply market impact model (e.g., square-root impact: cost ∝ √(participation rate)) to compute expected slippage at increasing AUM levels
Find the AUM level where net-of-impact alpha falls below a minimum threshold (e.g., 0 bps or a target Sharpe hurdle)
Produce a capacity curve: net alpha vs. AUM
Diagnose decay regime and crowding risk
Compare decay profile across sub-periods (pre/post publication, different volatility regimes)
Check if half-life has shortened over time — an indicator of signal crowding or information diffusion
Cross-reference with short-interest, ETF flow, or factor-crowding proxies if available
Note any structural breaks where decay accelerated (new data vendors, regulatory changes, competing strategy launches)
Synthesize findings into actionable recommendations
Recommend hold period and rebalance cadence consistent with the decay profile
State capacity ceiling with assumptions made explicit
Identify risk: if half-life is near or below the cost-breakeven horizon, flag the signal as capacity-constrained or potentially non-viable at scale
The deliverable is a structured Signal Decay Analysis Report containing:
Signal summary table: signal name, universe, backtest period, IC₀, half-life, decay constant
Decay curve chart: IC or spread return vs. forward horizon with fitted exponential overlay
Turnover-cost analysis: table of rebalance frequency vs. gross alpha, cost drag, and net alpha
Capacity curve: net alpha vs. AUM chart with annotated capacity ceiling
Regime comparison: sub-period half-lives and any crowding indicators
Recommendation: optimal rebalance cadence, maximum deployable AUM, and risk flags
Quality Checks
Verify that IC calculations use point-in-time data with no lookahead bias in signal construction
Confirm transaction cost assumptions reflect realistic execution for the asset class and AUM level [VERIFY — cost estimates should be calibrated to actual fills or TCA data where available]
Ensure half-life estimate is stable across bootstrap samples; flag if confidence interval spans more than 2× the point estimate
Check that capacity estimate accounts for correlated liquidation risk (other funds trading the same signal)
Validate that decay curve uses non-overlapping or bias-adjusted overlapping returns to avoid autocorrelation inflation
Cross-check turnover calculations against actual portfolio rebalance logs if the signal is already in production