Slippage

Slippage is the gap between the price your strategy expected and the price it actually got filled at. You click, or the algo fires, and by the time the order reaches the market the price has already ticked once or twice against you. That gap isn't an occasional glitch. It shows up on every single trade, and across a few thousand trades it compounds into something that can swallow a strategy's whole edge.

Most backtests skip it entirely. The default assumption is a fill at the exact signal price, as if the order arrived instantly and the market waited politely for it. No broker works that way. Model fills honestly — real latency, a real order book, not a bar's printed close — and a strategy's paper edge starts shrinking before a cent of commission comes off.

Some systems barely feel it. A strategy trading a handful of times a year on deep, liquid futures can mostly shrug slippage off. A system firing dozens of times a day, or sizing large relative to what's actually resting on the book, pays for it on every single trade. We build fills from tick data with a real bid/ask spread instead of a bar's printed close, because a close price was never something you could actually transact at.

Slippage is one of three costs, alongside the spread and the commission, behind what we call cost-fatal: a strategy whose entries genuinely predicted something, that was profitable on paper, and that dies the moment real execution costs get subtracted. Gross-positive, net-negative. The logic worked. The market's plumbing didn't let it through.

A fill you assumed was free is a number borrowed from a curve that never agreed to pay it back. That's the real cost of ignoring slippage: the backtest reads like proof, when it's actually a loan against results the live market has no obligation to honor.