Concept
Live Decay Tracking
Live Decay Tracking, also known as published-strategy forward performance, alpha decay, is a Performance, Backtesting & Validation concept. First implementations are in the build queue: the write-up leads, the indicators follow.
the public differentiator
What is live decay tracking?
Live decay tracking is the ongoing comparison of a deployed strategy's realized performance against the expectations set by its backtest, with the specific goal of detecting alpha decay: the gradual erosion of an edge after it goes live. Rather than asking once whether a strategy works, it treats validation as continuous, monitoring whether the live Sharpe ratio, win rate, average trade, and drawdown profile remain consistent with the distribution the backtest implied.
The practice exists because edges are not permanent. Crowding erodes them as other participants find the same signal, market structure changes remove the inefficiency that fed them, and some were never real to begin with, only overfit noise that decays instantly. The phenomenon is well documented for published research: McLean and Pontiff's 2016 study of academic equity anomalies found that returns fell materially after the sample period ended and fell further after publication, which is why forward performance of published strategies is watched as a barometer of how fast disclosed edges fade.
Traders care because the alternative is discovering decay through the account balance. The hard problem is statistical: live samples are small and noisy, and every strategy spends time below expectations by chance. Decay tracking is therefore about pre-defined comparisons against the backtest's own variability, distinguishing an ordinary cold streak from evidence that the process generating returns has changed.
How traders use it
- By defining tolerance bands before deployment: simulate the backtest's trade sequence, for example with Monte Carlo resampling, to establish how bad a drawdown or losing stretch is plausible under the null of an intact edge, then compare live results against those bands rather than against the average.
- Through rolling metrics: rolling Sharpe ratio, expectancy, and slippage-adjusted average trade over the last N trades, plotted against the same metrics rolled through the backtest, make gradual erosion visible before it is conclusive.
- By monitoring inputs as well as outputs: fill quality, signal frequency, and average holding period drifting away from historical norms often flag regime or microstructure change earlier than the profit series does.
- As a capital allocation input: many practitioners scale exposure down as live evidence weakens rather than waiting for a binary kill decision, a relative of equity-curve-based throttling, and retire strategies whose live statistics breach pre-committed limits.
- Its core limitation is power: with few live trades, real decay and bad luck look identical for a long time, so decisions rest on thresholds chosen in advance, when judgment was not under pressure from an open drawdown.
Live decay tracking vs related concepts
Forward/paper-trading Confirmation: Confirmation is the pre-deployment gate that asks whether the strategy earns capital at all; decay tracking is the post-deployment watch that asks whether it keeps earning it. The comparison machinery is similar, but the stakes and sample grow over time.
Drawdown Statistics: Drawdown statistics describe the depth and duration of equity declines in any record. Decay tracking uses them as one input, asking specifically whether the live drawdown profile has left the range the backtest made plausible.
Parameter Stability: A pre-deployment robustness probe: fragile parameter peaks predict fast decay. Decay tracking is the live-side measurement of whether that prediction, or crowding, is actually materializing.
Related concepts · Validation methodology
Concept family
Performance, Backtesting & Validation
30 concepts mapped · 30 in the Library
Live Decay Tracking FAQ
How do I tell alpha decay from a normal losing streak?
Statistically, only by comparing the live stretch to the distribution of stretches the backtest itself produces. If the current drawdown or rolling expectancy sits within what resampled backtest sequences show routinely, there is no evidence of decay yet; if it breaches pre-set bands, the edge is suspect. Certainty arrives slowly either way.
How fast do edges decay after publication?
Evidence from academic equity anomalies suggests post-publication returns are substantially lower than in-sample returns, with part of the drop attributable to overfitting and part to crowding. Speed varies widely: capacity-constrained and widely publicized edges tend to fade fastest.
Should I re-optimize a decaying strategy?
Cautiously. Re-tuning on recent data can adapt to genuine regime change, but it can also chase noise and quietly convert live data into in-sample data. A disciplined approach re-optimizes on a schedule fixed in advance, as in walk-forward operation, rather than in reaction to losses.
What metrics decay first?
Often the subtle ones: average profit per trade shrinks through worse fills and thinner follow-through before win rate visibly moves. Tracking slippage-adjusted expectancy tends to give earlier warning than tracking the equity curve alone.
Build Live Decay Tracking your way.
Quant writes, tests, and refines it with you — then it runs on LuxAlgo charting or ports to TradingView.