Reading the Backtest Numbers

A backtest report has dozens of figures. This guide covers the ones that matter most, and how to use them to decide whether a system suits you. Each strategy's real figures are on the Backtest page.

Net profit

Total profit minus total loss over the whole test. On its own it tells you little, because it doesn't show how much risk it took to get there. Always read it alongside drawdown.

Max drawdown

The largest drop in your balance from a peak. This is the most important figure for your decision, because it's what you'll actually have to sit through. If a system's drawdown is 30% but you'd switch the bot off once your account fell 15%, lower your risk per trade before you start.

And future drawdown can be larger than past drawdown. The figure in a backtest is what has happened, not a ceiling.

Win rate and payout ratio

Win rate alone can't tell you whether a system is any good. Read it together with the payout ratio: the average winning trade divided by the average losing trade.

Example: a system wins 40% of its trades. Winners average $150, losers average $90. Its payout ratio is 150 ÷ 90 ≈ 1.67. It loses more often than it wins, yet still makes money, because each win is bigger than each loss.

The break-even win rate is 1 ÷ (1 + payout ratio). The system above needs to win at least 1 ÷ 2.67 ≈ 37.5% of trades to avoid losing money. If a system's real win rate sits only a little above its break-even point, even a mildly unfavorable market can push results close to flat.

Most trend-following systems win less than 50% of the time. That's normal: they cut losses quickly and let winners run.

Profit factor

Total gross profit divided by total gross loss.

Below 1: losing money
1: breaking even
Above 1: making money — e.g. 1.2 means that for every $1 lost, the system made back $1.20

An unusually high value (say, 3 or more across thousands of trades) deserves suspicion. It can be a sign the settings were fitted too closely to the past.

Max consecutive losses

Worth knowing before you go live, because it tells you how long a losing streak you need to be ready for. This table shows roughly how much a balance falls during a losing streak:

Losses in a row	0.5% risk per trade	1%	2%
10	~5%	~10%	~18%
20	~10%	~18%	~33%
40	~18%	~33%	~55%

A streak like this can come at any time — including right after you start. If the numbers in your column look too scary, lower your risk per trade.

Stagnation

The longest stretch where the balance didn't make a new high. If it's 200 days, there was a period of nearly 7 months where the bot seemed to "go nowhere". Most people quit a system during a stretch like that. Knowing it in advance makes it easier to stay calm.

Number of trades

The more trades, the more statistically meaningful the results. A few dozen trades can be pure luck; hundreds to thousands are enough to show a system's real character.

Return / drawdown ratio

Net profit divided by max drawdown. It tells you how many times over the system earned its worst pain. Higher is better, and it's a good way to compare systems — or different settings of the same system.

Look year by year, not just the total

A strong total can hide very different years — some very profitable, some flat or losing. Looking by year (or month) shows how consistent a system really is, and how bad its worst year was.

Check the buy / sell split

If almost all trades are on one side, the results depend on the market's direction during the test. A trend-following system tested in a rising market, for example, will be almost all buys. If the market turns down or ranges for a long time, results can look very different.

In short

Don't judge a system by profit alone. Before you start, ask yourself three questions:

Can I really live with a drawdown this size?
If I hit a losing streak as long as the worst one on record, or several flat months, would I keep the bot running?
If not, how much should I lower my risk per trade?

Past results don't guarantee future performance.