Fill

A fill is an order becoming a trade. Every fill has a price, a quantity, and a timestamp, and the gap between requested and filled is where execution cost lives.

Full vs partial

An order can fill completely in one print or accumulate through several partial fills at different prices. Your effective entry is the volume-weighted average of them. Large orders relative to available liquidity fill partially and push the price as they go.

Fill simulation is the heart of a backtester

Two backtests with identical signals can disagree wildly if their fill models differ. The questions that matter: does a market order fill on the bar it was decided (a look-ahead cheat) or on the next bar's open? Do stops fill at the trigger price, or with gap-through slippage the way live stops do? Does size worsen the price, as it does against a real order book?

AlphaProve's engine takes the honest branch of each: decisions on close, fills on next 1-minute open, stop fills penalized, and L2-depth-aware pricing where order-book history is available.

On AlphaProve

Each fill surfaces twice. During a run it streams live over SSE — every signal, fill, and trade emitted in order as the backtest works through the data — and afterward the tearsheet groups the closing fills into an exit-reason breakdown that tags each one TP, SL, FLIP, or LIQUIDATION, so you can see at a glance what actually closed your trades.