One chart tells you more about a trading system than any table of statistics, and most people glance at it for two seconds and move on. It is the equity curve: the running balance of an account plotted trade after trade, from the first fill to the last. Learn to read that line and you can spot a fragile system, an honest one, or an outright fake.
This is a short course in reading it: slope, drawdowns, flat stretches, and why a curve that looks too clean is the one to walk away from.
An equity curve is nothing fancy. Start with the account balance, add the result of each trade in the order it happened, and plot the running total. A winner steps the line up, a loser steps it down, and the shape that emerges over thousands of trades is the whole life of the account in one picture.
Two versions exist and they differ. A closed-trade curve moves only when a trade closes, so it looks like a staircase. An intraday or marked-to-market curve also shows the unrealized swing while a position is open, so it dips lower during open trades and makes drawdowns look deeper. Know which one you are looking at, because a closed-trade curve can hide the heat you sat through. Measuring a system this way starts with what a backtest is.
A good equity curve is not a straight line to the sky. It rises on average over a large number of trades, with real dips and flat stretches along the way. A curve with no dips is not a better system. It is a warning that the result was fitted to the past, or invented.
The first thing to read is the overall slope. Does the line rise across the full sample, or drift sideways and only look like it climbs because your eye anchors on a lucky stretch at the end? Cover the last few months with your hand and check whether the earlier part was still rising. A system that only works in the final window is a system that only worked once.
Every equity curve goes down before it goes up again. The distance from a peak to the lowest point before a new peak is a drawdown, and it is the most important feature of the chart, more telling than the height of the finish. Anyone can look at the top-right corner. The careful reader looks at the deepest valley and asks one question: could I have kept the system running through that?
Read three things: how deep the worst one is, in dollars and as a share of the account; how long it lasted from the old peak to the recovery, because time underwater is harder to sit through than depth; and how often drawdowns of a given size repeat. The full treatment is in what drawdown is.
The peak tells you what the system made. The deepest valley tells you what you had to survive to be there for it.
Between the climbs and the dips are the flat stretches, where the account makes no progress for weeks or months. New traders assume a good system is always working. It is not. A strategy built for a particular kind of market goes quiet when that market is not present, and the flat spot is often the system correctly doing nothing rather than forcing trades that do not fit.
What you want to see is that the flat periods end, that the line eventually resumes its climb rather than rolling over into permanent decline. A flat stretch that never recovers is the fingerprint of a system that has stopped working. One followed by a fresh high is just the cost of admission.
Here is the counterintuitive part. A believable equity curve is lumpy. It has good runs and bad runs, sharp steps up and unpleasant steps down, and it wanders around its own trend line. A smooth, almost straight diagonal from bottom-left to top-right, with tiny even dips and no drama, should make you more suspicious, not less.
Why? Because real markets are noisy and a real strategy inherits that noise. A curve that is too smooth usually means the strategy was over-optimized until it fit every wiggle of the historical data (see overfitting and curve fitting), the sample is too short to contain a bad patch, costs were left out, or the record is fabricated. A claimed 100% win rate produces the smoothest curve of all, and it is always a lie, as spelled out in how to spot a fake track record.
The full backtest equity curve, the drawdowns and the losing stretches are on one page, with the live signal log that archives every new signal with its outcome.
Put concrete figures on this. The published backtest of the Algomyth engine, one of the engine behind this site, runs day by day from January 2022 to September 2026 on a mid-size account, one window a day at 9:30 AM ET. Its curve is not a straight line. It is built from 1,164 trades, and because the target sits at three times the risk, the line does not need the system to be right most of the time: break-even sits at a 25% win rate.
| Feature of the curve | What the backtest shows | Why it reads as honest |
|---|---|---|
| Reward to risk | 3:1 (break-even at a 25% win rate) | A system that need not win a majority produces visible losing streaks, not a smooth line |
| Profit factor | 1.46 | Modest, not a jackpot, consistent with a small repeated edge |
| Worst drawdown | $5,273 | A real valley you would have had to sit through |
| Sample size | 1,164 trades | Large enough that the slope is not a fluke |
| Costs | $1.90 a round turn plus a tick of slippage, charged per trade | A net curve; no gross one is published at all |
A profit factor of 1.46, explained in profit factor explained, means that for every dollar the system lost in the test it made about a dollar and forty-six cents, after costs: a real edge, and a small one. A small edge repeated hundreds of times is what an honest curve is made of. What it works out to per month, with every cost subtracted, is in how much a trading bot makes.
Hypothetical performance. The equity curve and every figure above come from a backtest of the system over historical data, not from a live account. Simulated results are prepared with hindsight, carry no financial risk, and cannot fully reflect real execution, slippage or liquidity. Past performance, real or simulated, does not guarantee future results.
Next time someone shows you an equity curve, run it through these questions before you react to the number at the end.
A curve that survives those six questions is worth a second look. One that fails any is telling you something the person showing it may not want you to notice. For more on judging systems, see the trading systems guide and the risk page.
A good equity curve rises on average from left to right across a large number of trades, but unevenly, with real dips, flat stretches and the occasional sharp drop. The trend is up and drawdowns recover to new highs over time, yet the line looks messy because real markets are messy. A smooth, near-perfect diagonal with no meaningful dips is usually a warning sign, not a good one.
Because real trading results are noisy and a genuine strategy inherits that noise. A curve with almost no drawdowns and an unbroken climb typically means the strategy was over-optimized to the historical data, the sample is too short to include a bad patch, costs were left out, or the record was invented. The smoothest curve of all comes from a claimed 100% win rate, which never survives a live market.
For deciding whether you could actually run a system, the drawdown matters more than the headline profit. The profit tells you what the strategy produced, but the deepest drawdown tells you what you would have had to sit through to be there for it, and most people who abandon a working system do so at the bottom of an ordinary drawdown. Read the worst peak-to-trough drop, in dollars and account percentage, before the final balance.
No, and they answer different questions. A backtest shows how a fixed set of rules would have behaved over historical data, which is valuable because you can reproduce it, but it is prepared with hindsight and cannot fully capture real execution, slippage or the discipline of running it live. A live curve reflects real fills but is harder to verify and usually far shorter. Treat a reproducible backtest as evidence, not a promise, and always confirm which one you are being shown.
Everything we write comes back to one idea: verify, then decide. The live terminal exists for exactly that.
Unlock Algomyth Live