Distributed edge survives it
Remove the top three trades and a real strategy sags but stays positive. The edge lives in the ordinary trades.
The good outcomeAnalytics
The return is the least useful number on the chart.
Everyone reads an equity curve the same way — end point minus start point, is it up. That number is almost the only thing on the chart that does not tell you whether the strategy is worth running. The shape does.
What you are actually looking at
An equity curve plots account value over time or over trade number. Plotting against trade number is generally more honest, because it removes the flattering effect of quiet periods — a strategy that made money in twelve trades over a year looks like a smooth upward line against time and like twelve isolated events against trade count.
The end point is a single sample from one path the market happened to take. Run the same strategy over a different year and that number changes enormously. What is more stable across paths — and therefore more informative — is the character of the line: how consistent the slope is, how deep and how long the drawdowns are, and whether the winners are spread out or concentrated.
Four shapes and what each one means
Most curves are recognisably one of these, and each has a specific implication.
- Steady slope, shallow drawdowns
- The one everyone wants. Consistent edge, sized sensibly. Worth checking that the sample is long enough to have contained a bad regime — a smooth year can simply mean the market suited the strategy.
- Staircase: long flats, sharp jumps
- A strategy whose edge arrives in bursts, typical of trend following. Genuinely viable, and the hard part is psychological — the flat sections are long enough that most people abandon it during one.
- Smooth up, then one cliff
- The signature of an unhedged tail risk. Selling premium, martingale sizing, or a strategy that quietly relies on mean reversion until the day it does not. The smoothness before the cliff is not evidence of safety, it is the mechanism.
- Up and to the right, but jagged
- Positive expectancy with size that is too large for the strategy's variance. The edge is real; the position sizing is not. This is fixable and the others mostly are not.
The questions worth asking of any curve
Five, in order. The first two eliminate most curves you will be shown.
- 01Is a backtest and a live record shown as one line? If so, where is the publication date? Everything before it was produced by someone who knew what came next. A blended curve hides exactly the transition that matters.
- 02How many trades? Not how many months. Thirty trades is not evidence regardless of how good the line looks; a 60% win rate over thirty trades is entirely consistent with a 45% strategy and a good run.
- 03What is the worst drawdown, in depth and in duration? Depth decides whether the account survives. Duration decides whether the trader does — and the second one ends more strategies.
- 04Is it net of costs? Commission and slippage take a fixed slice of every trade, so a curve gross of costs is steepest exactly where the strategy trades most.
- 05How concentrated are the winners? If removing the best three trades turns the curve flat, you are looking at three lucky trades rather than an edge. This is the single most revealing test and almost nobody runs it.
The best-three-trades test
Worth doing on any curve you are considering, including your own.
Remove the three largest winning trades and re-plot. A strategy with a genuine edge sags but stays recognisably positive, because the edge is distributed across many ordinary trades. A strategy that was three lucky trades goes flat or negative, and you have just learned that its record describes a handful of events rather than a repeatable process.
This is the same logic a prop firm's consistency rule encodes, from the other direction — they cap the share of profit from a single day precisely because a curve dominated by one event is not evidence of a process. Our consistency calculator shows where you sit against that.
The reason nobody runs the test is obvious: it is designed to disappoint you about a curve you already like.
Concentrated edge does not
If the curve goes flat, the record describes three events. More trades will not reproduce them.
Worth knowing earlyIt applies to your own
Easiest to run on a journal built from real fills, because the losing trades are all in there rather than the ones you remembered to log.
No survivorship in a fill logGetting a curve you can trust
Every test above depends on the underlying data being complete, which is where self-reported journals fail. A record you write by hand is missing the sessions you did not feel like writing up, and those are systematically the bad ones.
Quanify's analytics are computed from broker executions rather than typed entries — the platform placed the order, so the fill is recorded whether or not anyone wanted it recorded. Win rate, expectancy, profit factor and maximum drawdown all derive from that.
Two ratios are deliberately withheld when the sample cannot support them: Sortino below three losing months, Sharpe and Sortino below twelve. A ratio computed from a handful of trades is a number rather than a measurement, and showing it confidently would be worse than showing nothing.
The expectancy calculator does the same arithmetic on your own averages if you want to check a strategy before attaching it to anything.
Common questions
What is an equity curve?
A plot of account value over time or over trade number. Plotting against trade number is usually more honest, since it removes the flattering effect of quiet periods where nothing happened.
What makes a good equity curve?
A consistent slope with shallow, short drawdowns, built from many ordinary trades rather than a few large ones. The end return is the least informative thing on the chart — the shape is what generalises to the next year.
How do I know if an equity curve is real?
Ask where the backtest ends and the live record begins, how many trades it contains, whether it is net of costs, and what happens if you remove the best three trades. If removing three trades flattens it, the record describes three events rather than an edge.
Why does a smooth curve with one big drop matter?
Because the smoothness is often the mechanism rather than evidence of safety. Strategies with unhedged tail risk — selling premium, martingale sizing — produce exactly that shape, and the drop is the strategy working as designed rather than a failure.
Should I plot equity against time or trades?
Trades, mostly. Against time, a strategy that took twelve trades in a year looks like a smooth line; against trade number it correctly looks like twelve isolated events, which is what it is.
How many trades before an equity curve means anything?
More than most people assume. Thirty is not enough for any conclusion — a 60% win rate over thirty trades is statistically consistent with a 45% strategy having a good run. Hundreds, spanning more than one market regime, is where it starts to generalise.
Start on a simulated account.
Connect a sim or evaluation account and run the whole product against it before you point anything at live money. Nothing about the setup changes when you do.