Tool in Action: 20 Trades, +$220, and the Capital Quant Says It Needs

5 November 2026, 10:07
Michael Thomas Harty
0
7

The case (illustrative numbers): a trend EA has closed 20 trades at 0.10 lots. It won 12, averaging +$85, and lost 8, averaging −$100. That is +$220 net, +$11 a trade, and a worst drawdown so far of $140. The owner plans to run it on $1,500.

What the panel shows, figure by figure:

Figure Value How to read it
Expectancy +$11 a trade 95% range roughly −$30 to +$52
p95 drawdown (2,000 resamples) $675 1 run in 20 goes deeper
Suggested capital (p95 ÷ 20%) $3,375 money the EA should be given

Reading 1. The edge isn't proven. With a trade standard deviation of about $94, the standard error of the mean is $94 ÷ √20 ≈ $21. The 95% range runs from −$30 to +$52, so zero sits comfortably inside it. Twenty trades can't tell this EA apart from a coin flip.

Reading 2. The drawdown so far is the lucky ordering. Resampled 2,000 times (drawing from the same 20 trades with replacement), they give a median drawdown of $315 and a p95 of $675, nearly five times the $140 seen.

Reading 3. The decision. On $1,500, a $675 drawdown is 45%. The owner has two honest choices: bring $3,375, or scale the lot to fit the money. Drawdown scales with lot size, so the lot multiplier is (20% × $1,500) ÷ $675 = 0.44. That turns 0.10 lots into 0.04.

The part people miss: run it again at 40 and 80 trades. Early on, the suggested capital moves more than the EA does, and a figure that keeps rising tells you more than any single run.

VRD Quant produces these per-EA estimates from 10 closed trades upward, with each EA estimated on its own. With few trades it can only reshuffle what has already happened, so early figures are rough and live results can exceed them.

What to do today: for each EA, divide 20% of the money you actually give it by its p95 drawdown. If the answer is below 1, cut its lot by that factor until the sample grows.

Try it: Try the free demo of VRD Quant in the Strategy Tester to see p95 drawdown and suggested capital on its labelled sample data, then run it on your own history. $69 to buy, or rent for $30 for 1 month.
https://www.mql5.com/en/market/product/204796

Virtual Risk Desk

All figures are an illustrative case: made-up trades, a seeded resample run and a normal-approximation range. Educational content, not trading advice.