Trading glossary
- Backtest
- Running a trading rule on past prices to see how it would have performed. It shows what happened, not what will happen.
- Deflated Sharpe ratio
- A Sharpe ratio adjusted for how many strategy variants were tried and for non-normal returns. It answers: is this result better than the luckiest of many random attempts?
- Doesn't hold up (verdict)
- ChartingPath verdict: loses money in the recent period once estimated costs are included.
- Expectancy
- The average profit or loss per trade, including losers. On ChartingPath it is measured in R: +0.2R means you gain 0.2 times your risk per trade on average.
- Forward testing
- Running a strategy on new market data as it happens, without changing it. The only test that can't be accidentally fitted to the past.
- Fragile (verdict)
- ChartingPath verdict: profitable after costs recently, but it fails at double costs, was not positive earlier, or lost more than half its edge.
- Market regime
- The broad state of the market: trending, ranging or volatile. Many strategies only work in one regime and quietly fail in the others.
- Maximum drawdown
- The largest peak-to-trough fall in account value. It is the loss you would have had to sit through, and it matters for prop-firm rules.
- Monte Carlo simulation
- Reshuffling or resampling a strategy's trades thousands of times to see the range of outcomes it could plausibly have produced, not just the one lucky path.
- Multiple testing
- Testing many strategies and keeping the best one. The more you try, the more likely the best result is a fluke, so the bar for believing it must rise.
- Out-of-sample
- Testing a strategy on data it was not built or tuned on. In-sample is the data used to design it; out-of-sample results are the honest ones.
- Overfitting
- Tuning a strategy so closely to past data that it captures random noise instead of a real edge. Overfit strategies look great in backtests and fail live.
- Probability of backtest overfitting
- The estimated chance that the best-looking variant from a batch would rank below average on unseen data. High values mean the "winner" is probably luck.
- Profit factor
- Total money won divided by total money lost. Above 1.0 means profitable before other considerations; it says nothing about how reliable that is.
- R-multiple
- Profit or loss expressed as a multiple of the amount risked. Risking $100 and making $200 is +2R; hitting your stop is -1R.
- Robust (verdict)
- ChartingPath verdict: profitable after costs in both the earlier and the recent period, still positive at double costs, and it kept at least half its earlier edge.
- Sample size
- How many trades a result is based on. Small samples swing wildly by chance; ChartingPath gives no verdict below 100 trades.
- Sharpe ratio
- Average return divided by its volatility. Higher is smoother, but a high backtest Sharpe is a poor predictor of future Sharpe.
- Slippage and trading costs
- The gap between the price you expect and the price you get, plus spreads and commissions. Many backtested edges disappear once realistic costs are included.
- Too few trades (verdict)
- ChartingPath verdict: fewer than 50 trades in one of the two periods (or under 100 in total), too few to judge either way.
- Walk-forward testing
- Repeatedly tuning a strategy on one stretch of history, then testing it on the next unseen stretch, and stitching the unseen results together.
- Win rate
- The share of trades that make money. On its own it says little: a 30% win rate can be profitable and a 70% one can lose, depending on the size of wins and losses.