I Tried to Break My Own System by Deleting Its Best Trades

Every let-winners-run trader lives with a quiet fear. Your system makes most of its money from a handful of huge winners. The rest is small losses and breakevens. So a nagging voice asks: what if those few monster trades were just luck? What if, without them, I have nothing?

I decided to find out the hard way. I deleted my system's best trades — on purpose — and checked whether the edge survived.

Why this fear is rational

My main strategy has a brutal-looking profile. The win rate is about 43%. The median trade is a small loss. The average trade is positive only because a few trades run very, very far. In statistics terms, the returns have a "skew" of +5.96 — which is a fancy way of saying the distribution has a long, fat right tail. A small number of trades carry everything.

That's normal for trend-following. It's also exactly the setup for self-deception. If your entire 20-year track record rests on, say, three lottery-ticket trades, then you don't have a system — you have three lucky stories and a lot of noise around them. And no backtest's headline number will warn you, because the average looks great either way.

The usual statistical tests don't catch this. They confirm the edge is "real" and "not curve-fit." Neither answers the specific question: is the edge concentrated in a few trades that might never repeat?

The test: delete the winners and look again

So I built a blunt instrument. Take the full 20-year trade record. Delete the single best trade. Recompute the edge. Then delete the best two. Then the best three. Then the best five. At each step, ask: is the average return still positive, and is it still statistically distinguishable from zero?

If the edge collapses when you remove one or two trades, it's fragile — the tail is the strategy, and you're one unlucky decade away from nothing. If it survives deleting the best five trades over twenty years, it's robust — the big winners sit on top of a genuinely positive base, not in place of one.

The result

Here's what my main system did as I deleted its best trades, one by one:

Delete the five biggest winners from twenty years of trading, and the system still earns +0.51R per trade, with the edge still statistically real. The top three trades account for only 26% of total profit — meaningful, but nowhere near "the whole thing."

The skew is real, but it isn't fragility. The monster winners are a bonus sitting on a broad, healthy base — not a crutch holding up an otherwise-empty system.

Why I find this more reassuring than a high return number

A backtest that shows a big average return tells you the system worked. It doesn't tell you why, or whether the "why" was repeatable. A system that survives having its best trades deleted is telling you something deeper: the edge is distributed across hundreds of trades, not balanced on a few. That's the difference between a method and a memory of good luck.

I ran the same test on the strategy variants I had considered but rejected. Several of them failed it — delete one or two trades and their edge vanished. That gave me a second piece of confidence: the rejected versions weren't just lower-return, they were structurally fragile, and the selection process had quietly avoided them for the right reason.

The takeaway

If you trade a few-big-winners style — breakouts, trend-following, anything with a fat right tail — run this test on your own record before you trust it. Delete your best trades and see what's left. If the edge survives, you have a system. If it evaporates, you have a story.

I'd rather know which one I'm holding before real money is on the line. That's the entire reason I do this in public, on paper, before going live.

Personal research. Not investment advice. Robustness analysis on a 20-year walk-forward trade record; deleting trades is a stress test, not a forecast of future magnitudes. Past results do not guarantee future performance.

[Get the free 36-year research brief →](https://moeasymmetry.netlify.app/the-36-year-test.html)