Everything you need to know before you place a bet.

Most bettors lose because they are trying to pick which horse will win the race. That sounds like the right goal — but it isn’t. Picking winners and betting profitably are completely different skills.

A horse that wins 40% of the time at 3-5 odds will cost you money on every single bet. Profitability isn’t about hit rate — it’s about whether the price offered was worth the risk you took. Algorithm Factors is built around that distinction.

The only sustainable path to profit is to bet only when you have a genuine mathematical edge — when the price being offered is greater than what the horse’s true probability of winning justifies.

At a casino, the house sets the odds and always has the edge. At the race track, the public sets the odds through a parimutuel system — and the public makes mistakes. When they underestimate a horse, the expected value equation works in your favor. That’s the edge Algorithm Factors helps you find.

An overlay occurs when a horse’s actual tote board odds are significantly higher than its true mathematical probability of winning warrants. In other words, the public has undervalued that horse — they think it has less of a chance than it actually does.

When you bet a true overlay, the expected value of that bet is positive. That’s when a bet is worth making. When the tote board odds are at or below the true probability — the horse is fairly priced or overbet — you pass. Most races: pass. Some races: act.

You need an independent estimate of each horse’s true probability of winning — one that has nothing to do with public money or handicapping opinion. That’s what the Race Analyzer provides.

Each night you receive data files that, when loaded into the Race Analyzer, show you the fair odds for every horse in every race. You compare that number to the tote board at post time. When the tote is meaningfully higher than our fair odds estimate, a potential overlay exists. When it isn’t, you pass.

No — and that’s not the goal. Even a perfectly placed bet on a genuine overlay will lose sometimes. Probability still controls outcomes race by race.

The goal is positive expected value over time. If you consistently bet only when the math is in your favor, you will be profitable over a meaningful sample of bets. The discipline to pass most races — and only act when a real edge exists — is the most important skill a serious bettor can develop.

All gambling — casinos, lotteries, horse races — is controlled by the expected value equation: Expected Value = (Probability of Winning × Amount Won) − (Probability of Losing × Amount Lost).

A bet is only worth making when this equation produces a positive number — when the true probability of winning multiplied by the payout exceeds the probability of losing multiplied by the stake. Casinos set odds so this equation always favors the house. At the track, the public sets the odds — and sometimes gets it wrong. When they do, the equation can work in your favor.

Edward O. Thorp — mathematician and pioneer of modern probability theory — proved that blackjack could be beaten by skilled players. He discovered that by tracking which cards had been played, a skilled player could identify moments when the remaining deck gave them a mathematical edge, and bet large only in those moments.

Algorithm Factors does the same thing for horse racing. Instead of tracking cards, it analyzes over 13 million data points to identify when the public has mispriced a horse — and the math is in your favor. You bet large when the edge is real. You walk away when it isn’t.

No. Algorithm Factors has done all the hard work. The Race Analyzer takes the 13 million data points, runs them through the algorithms, and presents you with one clear number per horse: the fair odds.

Your job is simple: compare that number to the tote board at post time. If the tote is meaningfully higher, a potential edge exists. If it isn’t, you pass. No calculations required.

The Race Analyzer is built on a database of over 13 million data points — including historical race results, pace figures, trainer patterns, surface preferences, distance performance, and dozens of other statistically significant factors.

The algorithms identify which combinations of factors have the strongest predictive correlation with winning — independent of public money or handicapping opinion. The result is a probability estimate that reflects what the math says the true odds should be.

The Race Analyzer is a proprietary HTML tool built on 13 million data points. You open it in any web browser, drag in your nightly data files, and instantly see win probabilities, fair odds, and a Verdict for every horse in every race at every major track.

It was built by Gary Levin — a court-recognized expert in the mathematics of gambling — and represents decades of research into probability, betting markets, and race data.

For every horse in every race, the Race Analyzer displays:

Win probability — the algorithm’s estimate of each horse’s true chance of winning.
Fair odds — that probability converted to payout odds, for direct comparison to the tote board.
Verdict — a rating of Elite, Strong, Fair, or Weak based on how the horse compares to 10,000 similarly situated horses.
Reliability % — the model’s confidence score for that race. Higher is better.
Fishing Score — signals whether the race has a higher or lower likelihood of a longer-priced winner.

It also includes running style, improvement scores, trainer value ratings, pace data, and more.

The Race Analyzer includes two models you can toggle between:

All Winners — predicts any winning horse regardless of odds. Best suited for win bets where you want the most likely winner at any price.

$10+ Winners — tuned to identify longer-priced winners. Particularly useful for exactas, trifectas, and value plays where you’re looking for overlays at higher odds.

Every horse receives a Verdict based on how it rates against 10,000 similarly situated horses:

★ Elite (38%+) — Dominant pick. Strong single-bet candidate when also showing as an overlay.
✓ Strong (28–37%) — Solid top contender.
~ Fair (18–27%) — Competitive but not dominant. Use with caution.
✗ Weak (below 18%) — Use as underneath in exotics only, never as a single.

Each race tab shows a Reliability % — the model’s accuracy score for that specific race based on the data available.

81%+ — Very reliable. High confidence in the model’s output.
75–80% — Reliable.
70–74% — Moderate. Proceed with some caution.
Below 70% — Use with caution. The model has less certainty for this race.

Click the ✂ Scratch button next to any horse. The horse moves to the bottom of the field and all win probabilities automatically recalculate for the remaining runners. The tool works very fast — if it appears nothing happened, scroll to the bottom to confirm.

To undo a scratch, scroll to the bottom and click ↩ Unscratch.

This button appears only on turf races. If a race is moved off turf to dirt due to weather, clicking 🌧 Off Turf → Dirt rescores all horses using the dirt model instead of the turf model. Click ↩ Restore Turf to revert to the original turf ratings.

Chrome or Edge are recommended for the best experience. The Race Analyzer is an HTML file that runs entirely in your browser — no software installation required. Simply open the file and drag your data files onto the drop zone.

Yes. The Demo page includes a live version of the Race Analyzer with a sample race card loaded. Click “Load Sample Race Card” to see exactly how the tool works — the Verdicts, fair odds, win probabilities, and all columns — before you subscribe.

The subscription is $139 per month. This includes the Gary Levin Race Analyzer and nightly TXT data files for all major tracks, delivered by email every evening.

You receive two things:

The Gary Levin Race Analyzer — a proprietary HTML tool you open in your browser to analyze every race. Delivered once at signup.

Nightly TXT data files — delivered by email every evening for the following day’s races. Drag them into the Race Analyzer and you’re ready before the first race goes off. All major tracks covered. Up to 16 tracks loadable at once.

You can cancel at the end of any month. Cancellations are handled through PayPal. Log into your PayPal account, go to your active subscriptions, and cancel the Algorithm Factors subscription. If you have any trouble, email us directly and we’ll handle it promptly.

You start receiving data files the same night you subscribe. There is no prorated billing — your first month runs from the date you subscribe. Subscribe today and you’ll receive tonight’s data files.

No. This is a month-to-month subscription with no long-term commitment. You can cancel at the end of any month with no penalty.

Data files are delivered by email every evening, typically around 11 p.m., for the following day’s races. This gives you time to review the Race Analyzer before you leave the house the next morning.

If you subscribe after 11 a.m. on a race day, your files will be delivered within 30 minutes of subscribing — so you’ll have them before the first race goes off that same day.

All major tracks are covered, including Aqueduct, Fair Grounds, Gulfstream, Santa Anita, Keeneland, Belmont, Churchill Downs, and more. Up to 16 tracks can be loaded into the Race Analyzer simultaneously.

As a general rule, only bet tracks with a win pool of at least $20,000. Larger pools mean more accurate public pricing and more meaningful overlay opportunities.

First, check your spam or junk folder — emails from algorithmfactors@gmail.com occasionally get filtered. If the files aren’t there, email us directly and we’ll resend them immediately.

Open the Race Analyzer HTML file in your browser (Chrome or Edge recommended). Drag one or more TXT data files onto the drop zone, or click to browse for them. Tracks load and merge in alphabetically — duplicate races are ignored.

To add more tracks after initial load, click + Add more track files in the top left. You can load up to 16 tracks at once.

No. As a general rule, only bet tracks with a win pool of at least $20,000. Smaller pools mean less accurate public pricing — the crowd’s mistakes are harder to exploit because fewer people are betting and the odds are more volatile.

Stick to the major tracks with deep pools. That’s where the most reliable overlay opportunities appear.