Learn

The concepts behind the numbers

Short, precise explanations of the statistical ideas Kestrel Signal implements. No prerequisite assumed beyond curiosity and a passing familiarity with trading.

Deflated Sharpe Ratio

Why a Sharpe of 2.4 is probably noise — and how to adjust for the number of strategies you tried.

Walk-Forward Analysis

Train on the past, test on the future. Repeated. The simplest honest test of out-of-sample performance.

Combinatorial Purged Cross-Validation

A rigorous way to estimate the probability that your backtest result is pure noise — PBO.

Overfitting and Curve-Fitting

The core problem with retail backtesting. What it is, why it happens, and how to detect it.

The Sharpe Ratio

What it measures, what it doesn't, and why you should never trust it in isolation.

Maximum Drawdown

Risk measured from the perspective of the capital you'd actually lose at the worst moment.

What Is a Systematic Trading Strategy?

A systematic trading strategy is a deterministic rule set mapping market data to positions, designed for reproducibility and statistical evaluation rather than discretionary judgment.

Simple vs Log Returns — Which to Use and Why

Log returns aggregate across time, simple returns aggregate across positions — the choice silently biases volatility, compounding, and portfolio math.

Volatility: what it measures and what it does not

Volatility quantifies return dispersion but ignores asymmetry, tails, regime shifts, and drawdown — making it a necessary but insufficient risk metric.

Win Rate vs Expectancy

Why a high win rate can still lose money, and how expectancy captures the full per-trade outcome distribution.

Position Sizing: Fixed Fractional Basics

Fixed fractional sizing allocates a constant percentage of current equity per trade, producing geometric compounding governed by the choice of fraction f.

How to Read an Equity Curve

A forensic guide to interpreting cumulative return curves, including what they reveal about strategy quality and what they systematically hide.

Transaction Costs and Slippage

The decomposition of explicit and implicit trading costs, how to size them realistically, and why constant-cost backtests systematically overstate live performance.

The Sortino Ratio

A risk-adjusted return metric using only downside deviation, designed for strategies with asymmetric return distributions.

The Calmar Ratio

A drawdown-normalized return metric that measures CAGR against worst observed loss, with known instability and tail-blindness limitations.

Beta and Market Correlation

Beta quantifies a strategy's systematic exposure to a benchmark and separates market-driven returns from genuine alpha.

In-Sample vs Out-of-Sample Data

A time-ordered split of historical data into a calibration set and a held-out set, used to measure how much of a strategy's backtest performance survives contact with unseen observations.

Mean Reversion vs Trend Following

The two foundational price-series regimes, the statistics that distinguish them, and the timescale-dependence that breaks naive classification.

Parameter Sensitivity Analysis

A diagnostic that measures how strategy performance responds to small changes in its parameters, distinguishing robust edges from curve-fit spikes.

Survivorship Bias

The distortion introduced when a backtest universe contains only assets that survived to the present, inflating returns and suppressing drawdowns.

Annualisation and the Square-Root-T Rule

How return and volatility statistics scale across time horizons, why the sqrt(T) identity breaks under realistic return dynamics, and how to read annualised tearsheets without being misled.

Kelly Criterion and Fractional Kelly

The growth-optimal position-sizing rule, its closed-form solution, and why practitioners trade a fraction of it.

The Information Ratio

Active return per unit of tracking error — the standard metric for evaluating systematic strategies against a benchmark.

Value at Risk and CVaR

VaR quantifies the threshold loss at a given confidence level; CVaR measures the expected loss beyond that threshold, capturing tail severity that VaR alone ignores.

Bootstrap Methods for Strategy Evaluation

Resampling-based confidence intervals for backtest statistics that replace parametric assumptions with empirical sampling distributions.

Regime Detection and Conditional Performance

Decomposing strategy performance across distinct market states to distinguish robust edge from regime-dependent luck.

Factor Models: Momentum, Value, and Quality

Reference on the three workhorse equity factors — their construction, interpretation, correlation structure, and the methodological pitfalls that inflate naive backtests.

Market Microstructure and Execution Shortfall

Execution shortfall measures the gap between a strategy's decision price and its realized fill price, decomposed into spread cost, opportunity cost, and fees.

Cointegration and Pairs Trading Statistics

How cointegration tests identify mean-reverting spreads between assets, and the statistical pitfalls that undermine naive pairs trading.

Glossary

Precise definitions of every term used on this platform.