How TrackWiz Ratings Work

TrackWiz ratings are designed to do the slow work of cross-referencing dozens of variables across every horse in every race, so that the time you spend handicapping is focused on the few decisions that actually move the needle. This page explains, in plain English, what goes into the rating and how to use it alongside your own analysis.

What the rating tries to answer

For each horse in a race, the rating attempts to estimate the relative probability of finishing first — given the conditions of today's race, the horse's past performances, the connections, and the projected pace shape. It is a relative number: useful for ranking horses within a single race, not for comparing one race to another in absolute terms.

The inputs

The rating combines signals across several categories that are well-established in handicapping literature:

• Recent form — finishes, beaten lengths, and speed figure trends in the most recent starts, weighted by recency.
• Class — the class level of recent races compared to today's conditions, with explicit credit or penalty for class moves.
• Surface and distance fit — past performance restricted to today's surface and within a reasonable distance band.
• Pace projection — running-style mix in today's field and how that setup historically favors each horse's style.
• Connections — trainer and jockey win rates, including trainer pattern bumps (1st-off-claim, layoff returns, second start back, etc.).
• Equipment and medication changes — first-time Lasix, blinkers on/off, first start as a gelding, and other commonly cited form-cycle signals.
• Workouts — recency, sharpness, and ranking of published workouts going into today's race.

What the rating does not do

Ratings are a tool, not an oracle. Specifically, the rating:

• Does not see today's odds. By design, it stays independent of the betting market so it can be compared against the market to spot value.
• Does not watch race replays. Trip notes that a human handicapper would catch — a checked horse, a wide turn, a stumble at the gate — are not in the data.
• Does not know about today-only factors a chartcaller can't encode. Weather, track bias developing during the day, late equipment changes, and stable-side chatter are outside the dataset.
• Does not guarantee outcomes. Horse racing is a high-variance sport, and even the best probability estimate is just that — an estimate.

How to use the rating

The most productive workflow we've seen:

1. Look at the top-rated horse in each race as a starting point — not as a play.
2. Read the race card and apply your own handicapping. The five-step framework is a good template.
3. Use the rating to confirm or challenge your own opinion. When you and the rating agree, conviction is high. When you disagree, ask why — sometimes you'll have a trip note the rating doesn't see; other times the rating will catch a pattern you missed.
4. Compare your final probability to the live odds. The bet lives in the gap — see the value and longshot guide.

Where the rating is published

Free best bets — typically the top one or two ratings on each card — are published every morning on the race sheets page. Full-card ratings, plus pace projections and expert tip sheets, are part of the premium subscription. See pricing for details.

A note on transparency

This page is intentionally a high-level overview rather than a full algorithmic spec. Detailed weighting and the proprietary parts of the model are not published. What we can commit to is that the rating is built from data bettors have been using for decades, weighted in ways that backtesting on historical results justifies, and recalibrated as new data comes in.