Studies

Pro-Handicap 2.0: The Evidence Behind the Model

Over the last few months, Pro-Handicap 2.0 has conducted extensive historical back-tests at major tracks, including Belmont Park, Keeneland, Parx Racing, Laurel Park and Gulfstream Park. The studies were designed to identify effective handicapping factors and weights for different racing conditions.

Race distances and surfaces were examined separately, including dirt and turf sprints and routes. This allows Pro-Handicap to develop specialized configurations tailored to each track and race category.

How the Studies Were Conducted

Historical race data was matched with recorded finishers and win payouts. For each qualifying race, the model ranked the horses, and its highest-ranked horse was evaluated against the result.

The studies measured how often the top-ranked horse won and calculated the return from a hypothetical $2 win bet on that horse in every qualifying race. This allowed selection accuracy and return on investment to be evaluated together.

Numerical optimization was used to test combinations of handicapping factors. The studies also used log loss and, in several cases, Brier scores to evaluate the model’s probability estimates. Some configurations were developed using earlier races and evaluated on later races; others were fitted using the full historical sample.

The specific factor weights and formulas are proprietary.

What Sets Pro-Handicap Apart

Some handicapping products present indicators or ratings without disclosing how those measures performed when tested against historical results. Pro-Handicap evaluates combinations of factors against recorded outcomes and payouts, then examines the results by track, surface and distance.

Rather than relying solely on a judgment about which factors matter, the studies measure how different combinations perform. The goal is to understand how those factors work together—and to evaluate both the model’s ability to identify winners and the returns those winners produce.

Why the Results Matter

Horseplayers must overcome the built-in cost of pari-mutuel wagering. Before winning tickets are paid, a percentage is deducted from the win pool: 17% at Parx, 17% at Gulfstream Park, 16% at Keeneland, 16% at Belmont under NYRA’s schedule, and 18% at Laurel Park.

The studies calculated returns using actual published payouts, which already reflect those deductions. A positive historical ROI therefore means the winning selections returned enough to overcome the losing wagers and the effect of the takeout within that sample.

The back-tests provide evidence for developing and refining Pro-Handicap’s track- and race-specific configurations. Results fitted to the full historical sample show how a configuration performed on the races used to develop it. Results on later races provide a separate measure of performance, subject to the limitations of the available data. Continued testing on new races will help assess how the historical findings carry forward.

https://prohandicapanalytics.com/wp-content/uploads/2026/10/Pro_Handicap_2_Backtest_Studies-4.pdf