ThoroughWiz

Horse racing pace analysis

How ThoroughWiz projects a race's shape: the early and late bars, the shape read, and the two warnings printed on every sheet.

The projection (PROJECTED-PACE v5)

Each horse's early and late bars are gap bars: the field's best at that stage sets the full length, and every other bar is shortened by its measured gap behind that best. The early bar can be rescued upward — a horse the market and its lines both say is quicker than its raw figure shows is drawn at the higher mark — but never downward. The late bar is the pure gap read; no market input enters it. Nothing on the sheet is an absolute number: every bar is relative to this field, because races are.

The projected early order sorts the field into Lead, Press, Mid and Back slots from early-speed points alone — the figure no longer dilutes the ordering, a change the lab adopted after testing both forms against charted races.

The race shape read

Each race prints one shape word. Soft: at most one genuine speed horse, and it stands clear of the next on early points — the lone-speed setup. Hot: three or more speed horses, or three horses within a point of the leader — a duel that can melt the front down. Honest: everything in between. The shape is a description of how the race sets up, printed so the reader can judge the pace picture at a glance.

Two warnings, printed first

Two plain warnings print ahead of the race context on every sheet. The wide-open warning fires when the model's highest win chance in the race is under 31.1% — the race has no strong favorite by the model's own numbers, and the number prints with the warning. The no-figure warning counts the starters with no prior speed figure — usually first-time starters and horses whose lines carry no figure — because a race with unmeasured horses is a race where the numbers know less.

What pace does not do here

Pace is annotation, never a ranking. The lab tested pace as a selector three ways on the full historical population — re-ranking by pace, pace as a veto, and pace-scenario narrowing — and none of the three beat the ranking the site files. So the bars describe the race; they never move a pick. The tests, with their counts, are in the research tier of the record and on What We Tested.