Every section of the report, on a synthetic sample.

Median trade

-152.81 USD

Half these trades did better than this, half worse.

5th percentile

-244.53 USD

One trade in twenty was worse than this.

25th percentile

-198.89 USD

A quarter of trades were worse than this.

75th percentile

+271.42 USD

A quarter of trades were better than this.

95th percentile

+370.17 USD

One trade in twenty was better than this.

Skew

-0.08

Above zero: a few large winners. Below zero: a few large losers.

Fat tails

-1.42

Excess kurtosis. Above zero means extreme trades happen more often than a bell curve would suggest.

Bell-curve shaped?

No

Test p-value 0.0000.

Big losses

1.8%

Share of trades more than two standard deviations below the average trade.

Gain to pain

1.17

Total profit divided by total loss. Above 1 means the wins outweigh the losses.

The normality check is a Jarque-Bera test. It only answers whether the trade results look like a bell curve. Trading results usually do not, and that is not a fault — it matters because most textbook risk formulas assume they do.

Value at risk, 95%

-244.53 USD

One trade in twenty was at least this bad.

Value at risk, 99%

-545.47 USD

One trade in a hundred was at least this bad.

Average of the worst 5%

-479.90 USD

When a trade did land in the worst twentieth, this was the average result.

Average of the worst 1%

-568.45 USD

The same idea for the worst hundredth.

Worst single trade

-593.41 USD

Worst three combined

-1,727.05 USD

If the three worst trades had happened back to back.

Worst five combined

-2,819.26 USD

Chance of eventual ruin at 2% risk per trade, over an unlimited number of trades

0.0%

A textbook gambler's-ruin estimate. It assumes fixed-fractional sizing at 2% of the account per trade (an assumption we chose, not something the sample report states), that the edge in these trades repeats unchanged, and an unlimited number of trades. Over a finite run the risk is lower. For ruin within a run the length of THIS backtest, read the simulated figure in the Monte Carlo section above — it uses a different, explicitly stated definition (falling 50% below the start), so the two numbers are not comparable.

Kelly fraction

7.0%

From the win rate and the average win-to-loss size.

Optimal f

16.8%

The fraction that would have grown this exact sequence of trades fastest — by construction, fitted to it.

These are theoretical upper bounds, not advice and not a recommended position size. Both are computed from this backtest alone and assume the next trades look like these ones. Trading at full Kelly or full optimal f produces drawdowns most people cannot sit through, and either number is badly wrong if the sample is small or the worst loss here is not the worst loss possible. Many traders who use them at all use a fraction of them.

Edge ratio

1.23

Average favourable move divided by average adverse move. Above 1 means trades tended to go their way first.

Average best moment

204.33 USD

How much the average trade was ahead at its best point.

Average worst moment

166.23 USD

How far the average trade went against it before closing.

Profit captured

79.5%

Of the move a winning trade offered, how much it actually kept.

Loss taken

91.7%

Of the move against a losing trade, how much it kept. Near 100% means losers ran to their worst point.

Average R multiple

1.44

Result divided by how far the trade went against it — profit per unit of heat.

Run-up vs drawdown

-0.66

Whether trades that went further against them also went further in their favour.

Correlation, next trade

0.05

Between each trade and the one straight after it.

Two trades later

0.02

Three trades later

0.00

Beyond noise?

No

Anything beyond ±0.10 counts as more than noise at this sample size.

Sequence looks random?

Cannot rule it out

Runs test p-value 0.2386.

Win after a win

51.0%

How often a winning trade was followed by another one.

Win after a loss

45.4%

Loss after a win

49.0%

Loss after a loss

54.6%

High here means losses arrive in runs, which is what makes a drawdown deep.

In the four rates below, a breakeven trade counts as “not a win”, so each pair adds up to 100%. Compare them with the overall win rate of 48.3%: if they are close, one trade tells you nothing about the next.