Performance analysis·10 min read
How to find your trading edge — in your own trades, not someone else's
An edge is not a strategy you bought or a pattern you were told about. It is a measurable fact about your own trading: under some condition — a setup, a session, an hour, a tag — your results are better than your overall average, by more than luck would explain, over enough trades to believe it. Finding one is a matter of asking your journal the right questions in the right order, and being suspicious of the answers.
Published 9 September 2026 · by the SageTradingJournal team
Start with the baseline
Before you can find a condition that beats your average, you need the average. Compute your overall expectancy in R — total R divided by number of trades — along with win rate, average winner and average loser. That is your baseline. Everything else is measured as lift against it: a slice with expectancy +0.5R when your baseline is +0.1R has a lift of +0.4R per trade.
If you have not been recording the stop at entry, you cannot compute R and this whole exercise is unavailable to you. Start recording it today (how to keep a trading journal); the analysis becomes possible after thirty or so trades.
Slice by what you actually recorded
| Dimension | The question it answers |
|---|---|
| Setup / play | Which of my ideas make money, and which just feel good to trade? |
| Session | Am I better in London than New York? Should I trade Asia at all? |
| Hour of day | Is there an hour where I consistently give money back? |
| Day of week | Do Mondays or Fridays look different — and is it the market or me? |
| Instrument | Is my edge in one market, and am I diluting it by trading three? |
| Tag | Does "waited-for-confirmation" beat "anticipated"? By how much? |
| Tag pairs | Two conditions together — a setup in a session — often carry what neither does alone. |
| State at entry | Do my low-focus mornings lose? Do trades after a loss lose? |
Each dimension gives you a small table: for every value, the number of trades, the expectancy, and its lift over the baseline. Sage's Edge Lab builds these tables from your journal and backtest sessions, ranks the slices by lift, and prints the sample size and a confidence interval next to every rate — for the reason the next section explains.
Confidence intervals, or you will fool yourself
A slice with five trades and an 80% win rate is not a discovery; the plausible range for that win rate runs from roughly 30% to 99%. Every rate you look at should carry its interval, and a slice only counts as an edge when the lower end of its interval still clears the baseline. This rule alone eliminates most "edges" a trader finds in their first month of analysis, which is exactly its job.
The many-comparisons trap
If you slice your trades twenty different ways, one slice will look good by chance alone — that is what "one in twenty" means. The more dimensions you examine, the more false edges you will find. Two defences: prefer slices that make sense before you look (a setup you deliberately trade in a session you deliberately chose), and confirm any candidate on trades it was not found in. The cleanest way to do the second is the backtesting habit of an untouched out-of-sample stretch: find the edge in one period, check it in the next, and only believe it if it survives (how to backtest).
Leaks: the edge you fix by stopping
The same tables that reveal edges reveal leaks — slices that consistently lose. Leaks are worth more than edges to most traders, because fixing one requires no new skill. If your trades between 15:00 and 16:00 have a −0.6R expectancy over 40 trades, you do not need to understand why. Stop trading that hour and your overall expectancy rises immediately. The usual leaks are a session you have no business in, an instrument you trade out of boredom, and anything entered within minutes of a loss (tilt and revenge trading).
From a slice to a play
When a slice survives the interval and the confirmation, write it down as a play: the condition, the setup, the session, the stop rule, the exit rule, the size. Give it a lifecycle — it is live now — and journal every trade against it, so the edge you found keeps being measured rather than assumed. Edges decay; a play that is still recording its own results tells you when. Sage's Playbook is built for this hand-off: from a discovered slice to a written play with its own tested and live statistics.
The metrics beyond expectancy
Expectancy tells you the average; the shape of the returns around it matters too. Profit factor (gross profit ÷ gross loss) and maximum drawdown describe the ride. Sharpe and Sortino ratios relate return to volatility and to downside volatility. SQN (system quality number) scales expectancy by its consistency over the sample. Kelly sizing tells you the theoretical bet fraction your edge supports — always trade a fraction of it. And a Monte Carlo resample of your own trades gives you a distribution of possible drawdowns and a probability of ruin at a given risk per trade, which is the number that should set your position size. Sage computes all of these in Edge Lab; what they mean and when they lie is in trading statistics that matter.