How it works: The tool takes the natural log of each IC observation and fits a line — ln(IC) = ln(IC₀) − λh — using ordinary least squares. From the slope λ it derives the half-life t½ = ln(2)/λ and a rebalancing guideline based on the point where IC falls below 75% of its peak. Only positive IC rows are used in the regression.
Open source — view or fork the calculator on GitHub (MIT, single dependency-free HTML file). Related tools: Information Coefficient Calculator · Backtest Overfitting Simulator · Cointegration & Pairs Trading Simulator · Signal Combination Simulator · Correlation Heatmap · VPIN & the Volume Clock · Signal Skill Explorer.
Signal IC at Multiple Horizons
Enter the Information Coefficient (IC) of your signal at each forward-return horizon. Only positive IC values are used in the fit.
| Horizon | IC value |
|---|---|
| 1d | |
| 5d | |
| 21d | |
| 63d |
Frequently asked
- What is a signal's half-life?
- The horizon over which its predictive power (IC) falls to half of its peak. The tool fits a log-linear model to your IC observations by least squares and reports the half-life as ln(2) divided by the decay rate. A short half-life means the edge is gone quickly and must be traded fast; a long one tolerates slower rebalancing.
- What do I enter?
- Your signal's IC measured at several forward-return horizons — for example 1, 5, 10 and 21 days. Enter your own measured ICs to get your signal's half-life; only positive-IC rows are used in the fit.
- How does it choose a rebalancing frequency?
- From the point where the fitted IC drops below about 75% of its peak. Trade more often than that and you mostly pay extra transaction costs for IC you have already captured; trade less often and you let alpha decay before acting.
- Doesn't transaction cost matter too?
- Yes. The half-life tells you how fast alpha fades, but the optimal rebalance also depends on your costs: a fast-decaying signal can still be untradeable if costs exceed the alpha captured per turn. Pair this with the Information Ratio and transaction-cost analysis.
- Is this investment advice?
- No. It is an educational estimation tool. The results depend entirely on the ICs you supply and assume a simple exponential decay.