We ran the bot at 10× leverage. It liquidated 34 times.

The bot on this site is deliberately boring. Over eleven years it turned a pretend $10,000 into about $177,000 — roughly a tenth of what simply holding Bitcoin did — but it did it while never falling more than about 16% from a high, versus Bitcoin’s 84% faceplant. It wins by not blowing up.

So the obvious temptation: if it’s that steady, why not multiply it? Take every call it makes and put 10× leverage behind it. We ran exactly that, candle by candle, across the full history — with real trading fees and real funding costs — and then tried every “smart” version we could think of. Here’s what actually happens.

10× isn’t aggressive. It’s a countdown.

Leverage has a hard floor called liquidation: borrow 10× and a move of only about 9% against your position wipes it out entirely. That sounds survivable until you remember what this bot does — it holds trend positions for days or weeks and sits through the noise. Its average trade only nudges a few percent against it, but the worst position it ever held rode a 40% move against entry… and then recovered into a winner. At 1× that patience is the whole point. At 10× you’re liquidated long before the recovery.

Getting liquidated isn’t about being wrong — it’s about being early. The 10× account gets stopped out at the bottom of moves the bot was ultimately right about, 34 separate times, the first within the opening months of the record. And the first time a full-sized position gets liquidated, the account is simply gone. Leverage doesn’t punish bad calls here. It punishes patience — which is the one thing this bot is good at.

The money, at 1× vs 2×

Below the danger zone, leverage does exactly what you’d expect, and it’s worth seeing in dollars. From the same pretend $10,000:

Two× turns $10k into $2.37M instead of $177k — a much bigger pile. But notice it still trails just buying and holding Bitcoin ($2.75M). Leverage here isn’t beating the market; it’s roughly matching a buy-and-hold return with a 30% drawdown instead of 84%. And that 30% is a $550,000 paper loss you’d have to sit through without flinching. Most people can’t — which is the real reason leverage ruins accounts, long before liquidation does.

Leverage is a dial, not an edge

This is the part people miss. Multiplying a strategy multiplies its wins and its losses equally — so the return-per-unit-of-risk doesn’t improve at all. In our runs the risk-adjusted score was flat across every leverage level, and it stayed flat when we checked it on years the model had never seen. More leverage is not a better strategy. It’s the same strategy, louder. The only thing that genuinely earns more without more risk is a better signal — and that’s the hard part we’ve failed to crack six different ways.

“But I’d use a stop-loss”

The natural fix: put a stop at entry, cut the loss if it goes against you, re-enter later. We tested that too. It backfires — for a very specific reason.

After the bot enters, the first few hours are a coin flip: price swings almost exactly equally both ways, and it’s right barely half the time. The edge is slow — it builds over days, and by two weeks the average entry has gone about twice as far its way as against it. But 72% of trades never even dip 5% against you, and the winners routinely take 5–15% of heat before they work. So a stop tight enough to make leverage safe sits right in the middle of normal noise: it sells the bottom of good trades. Bolt a 12% stop onto the plain 1× bot and its 11-year result falls from 18× to about 7× — and the drawdown doesn’t even improve. On a patient trend-follower, cutting losses fast is the loss.