When Key Factors Agree: What 1,482 Races Reveal About Win Rate and ROI

A quantitative look at what changes when a horse leads one CHV factor, two factors, or three—and why a strong win signal is not automatically a profitable wager.

CHV Analysis
This study reviews clean-condition, non-maiden Thoroughbred races. It measures horses ranked first in nine CHV factors, then tests what happened when several first-place rankings belonged to the same horse. The results separate two questions that bettors often combine: How likely is the horse to win? and Is the available price high enough to create a positive return?

The handicap report contains several ways to describe a horse: modeled win probability, speed, class, pace, connections, fitness, and value. Each factor can identify a useful part of the race. The harder question is whether any one of them is enough to justify a wager.

The data says no. A single factor can be highly predictive and still lose money at the windows. The most useful historical results appeared when complementary factors agreed—especially when ability, race shape, connections, or fitness supported one another.

What the study measured

The base sample contains 1,482 clean-condition, non-maiden races. "Clean condition" means the race stayed on its scheduled surface and was run over a fast dirt track or firm turf. Maiden races were removed because limited past-performance data can leave several factors incomplete.

For each race, the analysis identified horses ranked first in:

• W% — modeled win probability
• PFIG — Performance Figure
• BFIG — Best Figure
• ENY — Energy
• ES — Early Speed
• LS — Late Speed
• TJR — Trainer Jockey Rating
• FWR — Fitness Work-Out Rating
• VAL — value relative to the adjusted morning line

Ties can produce more starts than eligible races. Every ROI figure is a historical $2 win-bet result for the qualifying horses.

One factor answers only part of the question

The single-factor results show the difference between predicting winners and finding profitable prices.

Top-ranked W% horses won 33.0% of their starts, the highest rate among the nine individual factors. ENY followed at 31.2%, while PFIG reached 28.2%. These factors were effective at identifying contenders.

None of the nine individual factor leaders produced a positive historical ROI. TJR came closest at –11.4%, followed by ENY at –15.1% and BFIG at –15.4%. W% led the win-rate column but returned –21.7%.

That is not a contradiction. The betting public can recognize an obvious contender and reduce its price. A factor can improve the estimate of who is most likely to win without creating a profitable bet at the odds offered.

Two-factor agreement reveals complementary evidence

There are 36 possible two-factor combinations among the nine factors. The strongest historical result came from a combination that is easy to miss if each factor is judged alone.

Horses ranked first in both TJR and FWR won 33.3% of 276 starts and returned +12.1% ROI. FWR by itself had the lowest single-factor win rate at 17.0% and returned –22.7%. Paired with the top trainer-jockey signal, however, fitness became useful confirmation rather than a standalone selection method.

The second-best two-factor return came from BFIG and ENY. Those horses won 42.8% of 437 starts and returned +0.4% ROI. The profit was modest, but the combination joined a strong ability signal with a strong energy signal and produced a large, high-win-rate sample.

Other pairs generated excellent win rates without positive ROI. ES–LS horses won 51.5%, and ENY–TJR horses won 43.2%, yet both lost money historically. Agreement strengthens the probability case; price still determines whether the wager has value.

Three-factor agreement produced the strongest profiles

The leading three-factor combinations pushed many historical win rates above 50%. They also introduced smaller samples, so these results should be treated as evidence for further testing rather than permanent rules.

The highest ROI belonged to BFIG–ES–LS: 28 winners from 41 starts, a 68.3% win rate and +23.4% ROI. The sample is small, but the profile is coherent—best figure plus control of both early and late speed.

The most compelling blend of return and sample size was ENY–LS–TJR. It produced 90 winners from 164 starts, a 54.9% win rate and +20.6% ROI.

Across the ten positive-ROI combinations shown in the study, BFIG appeared eight times, TJR appeared five times, and ENY appeared four times. Modeled W% did not appear in that top group, despite being the strongest individual factor for identifying winners.

That distinction matters. W% estimates probability across the whole field. Factor combinations help explain why a horse may deserve more attention and where the market may be underweighting a specific strength.

What the combinations actually tell us

The study supports four practical conclusions:

1. Win probability and ROI are different jobs. A high hit rate can still be unprofitable when the market price is too short.
2. Weak standalone factors can become valuable in context. FWR was far more useful when paired with TJR or included in the right three-factor profile.
3. Complementary evidence matters more than repetition. Class, pace, fitness, and connections can confirm one another because they describe different parts of the horse and race.
4. More factors do not guarantee value. Some three-factor combinations with win rates above 50% still produced negative ROI. The offered odds remain part of the decision.

How to use Key Factor ROI in the CHV workflow

Start with the model's W% to understand the race's probability structure. Then review the factor ranks to learn whether the contender's case is broad or narrow.

A horse leading several complementary factors deserves closer attention, but the combination is not an automatic bet. Compare the horse's fair odds with the live price, review the race conditions, and decide whether the edge belongs in the win pool, as an exotic key, as a backup, or as a pass.

The goal is not to collect the most green rank-one cells. The goal is to understand what those cells say together—and whether the market is paying enough for the risk.

What comes next

These results can guide continued research into the CHV + suffix, which is designed to flag historically useful situations that may improve the return profile of a letter grade. Any change to the algorithm should be tested on new races and holdout samples before it becomes part of the live product.

This article reports exploratory historical results. Combination groups overlap, the smaller samples can be influenced by a few large payoffs, and past performance does not guarantee future results.

Related CHV resources

• Understand CHV factors and abbreviations
• Read the probability calibration analysis
• Review how CHV letter grades represent modeled win probability
• Watch CHV tutorial videos
• Compare the mobile app and Handicap Report

Continue your CHV journey

Use the CHV app to evaluate today's races with modeled probability, factor rankings, fair odds, and value in one workflow. Let the model identify the contenders, let the factor agreement explain the case, and let the price decide the wager.