Inside the 2026 Kentucky Derby: A Data-Led Look at All 20 Runners

The Kentucky Derby is the hardest puzzle in racing. Twenty 3-year-olds, a trip none of them have ever raced, a track most have never seen, and a market drowning in noise.

We pulled every runner through the same five-part analytical framework we've been refining at G1 Goldmine, ran the queries against our database of 9.6 million runner records across 41 countries, and let the numbers speak.

Here's what they say.

The Framework

Every runner is scored on five components. None on its own picks a Derby winner. Together, they tell you which horses have the most credible case at the line, and which are running on hope.

1. Class-adjusted recent form (35%) Last three starts, scored by race grade (G1, G2, G3, Listed, MSW), finishing position, beaten margin, distance proximity to 10 furlongs, and surface. The most recent start carries the most weight.

2. Sire distance affinity (20%) How the sire's progeny perform in dirt races between 9f and 10.5f globally over the last five years. Strike rate plus place rate, with a confidence adjustment that penalises low sample sizes. A sire with one runner from one start does not get credit for a 100% record.

3. Dosage suitability (20%) The Dosage Index (DI) and Centre of Distribution (CD) measure speed-vs-stamina balance in a horse's pedigree. Derby winners cluster in a narrow band: DI roughly 1.5 to 2.5, CD roughly 0.5 to 0.9. We score each runner by how close they sit to that historical sweet spot.

4. Freshness (15%) Days between the most recent start and Derby Day. The pattern of the last 20 years says 28 to 42 days is optimal. Anything outside that window, in either direction, is a yellow flag.

5. Trainer and jockey strike rate (10%). Combined graded-stakes (G1, G2, G3, Listed) win rate over the last 12 months, globally. International connections get full credit, with the same confidence adjustment applied.

Each component is scored from 0 to 100 across the field, then weighted into a single composite. Higher is better.

The Composite Rankings

The Headline Reads

Further Ado is the model's clear top pick

Brad Cox's colt sits 7.5 points clear of the next runner. The case for him is structural, not sentimental. Top-tier recent form. A sire (Gun Runner) with 324 progeny runs in our 9-10f dirt cohort and a strike rate that holds up under confidence weighting. Perfect 28-day prep cycle. And a trainer-jockey pairing (Cox and Irad Ortiz Jr) that ranks at the top of this field for global graded performance over the last 12 months.

The one caveat: his dosage profile leans speedy (DI 3.18). At 10f, a fast pace would expose him. He needs the trip to suit and his class to carry him home.

Commandment is the form pick

If you bet on raw 3-year-old class alone, Commandment is the horse. He posts the highest form score in the field. Cox-Prat is an elite combination. His Into Mischief sire base has 290 dirt 9-10f runs to draw confidence from.

The friction is dosage: DI 0.37 is the inverse of the historical Derby winner's profile. Stamina-dominant. He'll relish the trip, but if the race becomes a cruising-speed test rather than a stamina test, he's vulnerable.

So Happy has the cleanest balanced profile

Of the top tier, So Happy has the dosage shape (DI 0.93, CD -0.11) closest to a balanced classic horse. Mike Smith aboard. Mark Glatt is a small-yard trainer but holds his own at graded level. The tension here is sire sample (Runhappy, 63 runs in cohort) and a 5-1 morning line that will shorten further if money keeps coming.

The dosage value angle: Pavlovian and Fulleffort

Two horses sit in the historical Derby dosage sweet spot: Pavlovian (DI 1.93, CD 0.50) and Fulleffort (DI 1.71, CD 0.43). Neither is a market headliner. Both rank 8th and 9th overall.

What this means: if pedigree is the dominant lens, both horses suit the trip more naturally than the top picks. Their composite is dragged down by other factors (Pavlovian's connections, Fulleffort's recent form), but for a punter who weights stamina pedigree, they're worth a longer look than the market suggests.

The dosage red zone

Four horses have dosage profiles built for sprinting, not the Derby trip:

• Silent Tactic (DI 11.0)
• Incredibolt (DI 6.14)
• The Puma (DI 5.67)
• Danon Bourbon (DI 5.50)

Historically, horses with this kind of speed-dominant pedigree do not handle 10 furlongs against this calibre of opposition. Each can run a half-mile. Closing the second half is a different test.

The freshness flags

Three horses are racing well outside the 28-42 day prep window:

• Wonder Dean (133 days since last start, plus a Japanese form line that does not translate cleanly)
• Six Speed (99 days)
• Incredibolt (91 days)

Long lay-offs into the Derby are historically punished. Combine that with Wonder Dean's bottom-quartile finish across every other component, and the model has him last.

What the Data Doesn't Capture

We're upfront about what this analysis can and can't do.

It cannot read the weather. The forecast for Saturday matters: a wet track changes the pace dynamics, and several of these horses have never raced in the slop.

It cannot read the trip. Gate draws, pace scenarios, and the wide-out runs that have defined recent Derbies are race-day variables.

It cannot read the moment. Some horses peak at the right time. Some don't.

What it can do is tell you which runners have the structural foundations in place, and which ones are making up the numbers.

On Saturday, those foundations get tested.

How We Built This

The framework runs against the G1 Goldmine and Stallion Match database, which holds:

• 1.4 million horse records across 41 countries
• 9.6 million runner records with full race context (date, venue, class, distance, surface, finishing position, margin, trainer, jockey)
• Pedigree ancestry to 12 generations, with Chef-de-Race-style dosage points calculated for every horse
• Stallion progeny aggregations are updated daily, allowing us to filter sire performance by any combination of country, surface, distance band, and class

For this analysis, we ran the framework across all 20 Derby runners, drew on five years of progeny race data, twelve months of trainer and jockey performance, and applied confidence weighting (Wilson lower bounds) to every rate-based metric so small samples couldn't distort the rankings.

The full breakdown, including the component-level scores for every horse, is what our members see inside G1 Goldmine.

Try the Framework Yourself

The same data infrastructure powering this Derby analysis is available inside G1 Goldmine. Members can run pedigree, sire affinity, and dosage analysis on any horse, in any race, anywhere in the world.

If you want to see how your own sales prospects, broodmare picks, or weekend selections look through the same lens, start a free trial of G1 Goldmine.

We'll be back on Sunday with a post-race review: what the framework got right, what it missed, and why.

Matthew Ennis is co-founder of G1 Racesoft, the company behind G1 Goldmine and Stallion Match. The G1 Goldmine database holds one of the largest privately maintained repositories of global thoroughbred race and pedigree data.