Sprint Calculator uses real race data — thousands of recorded splits from 100m, 200m, 400m, and hurdle races — to predict how your times connect to each other. Give it one or more known times (a full race, or a split within a race) and it estimates another distance using the same underlying pace patterns real athletes show at that level.
With a single input, it uses a simple statistical fit. With two or more, it switches to a machine-learning model (XGBoost) trained on the database, which captures pacing patterns a straight line can't. Every prediction also reports its Mean Absolute Error (MAE) — a measure of how much predictions like this one typically miss by, so you know how much to trust the number.
Some inputs are marked as splits of a longer race (e.g. "150m (split 400m)") because a mid-race checkpoint is run at a different pace than an all-out effort over that same distance — the calculator accounts for that difference automatically when it converts between a standalone mark and a split.