Parents
Parent A
Parent B
Legal ranges: Health 15–30 HP, Jump 1.09–5.29 blocks, Speed 4.74–14.23 b/s. Values outside this range get clamped to the nearest legal value before simulating. Both parents must be fed a Golden Apple or Golden Carrot in-game before they will breed.
Foal Outlook
Per-Stat Prediction
How This Is Computed
Real Stat Ranges (Java Edition)
| Stat | Legal range | Displayed as |
|---|---|---|
| Max Health | 15 – 30 | HP (2 HP per heart) |
| Jump Strength | 1.09 – 5.29 | Blocks of jump height |
| Movement Speed | 4.74 – 14.23 | Blocks per second |
Jump Strength and Movement Speed are simplified here. In the actual game they are stored as a raw attribute (Jump Strength 0.4–1.0, run through the jump-height curve height = -0.1817v³ + 3.689v² + 2.128v - 0.343, giving a real block-height floor of ~1.09 rather than 0; Speed 0.1125–0.3375 blocks-per-tick, roughly ×42.16 to convert to blocks/second at 20 ticks/second). This tool works directly in the intuitive blocks / blocks-per-second units, which is also how the numbers actually show up in-game and in most stat-display mods — close enough for breeding-outcome purposes.
The Offspring Formula
Every foal stat — Health, Jump Strength, and Movement Speed — is rolled independently using the exact same shape of formula from the game’s AbstractHorse.createOffspringAttribute code (this also governs donkeys and mules):
midpoint = (parentA + parentB) / 2range = statMax - statMinfor that stat’s legal rangespread = |parentA - parentB| + 0.3 × rangeroll = (random() + random() + random()) / 3 - 0.5— three independent 0–1 rolls averaged together, then re-centered on zerofoalValue = midpoint + spread × roll- If the result falls outside the legal range, the real game reflects it back in; this tool clamps to the nearest legal boundary instead as a simplification.
Averaging three random rolls (step 4) instead of using one flat roll is what makes the result cluster toward the midpoint rather than spreading evenly — most foals land fairly close to the average of their parents, with occasional outliers toward either edge.
Uses the real vanilla offspring-attribute formula (parent midpoint + a difference-based spread, rolled via 3 averaged random values) run as a client-side Monte Carlo simulation — results are probabilistic estimates from many simulated rolls, not a single deterministic answer, matching how the real game works. Jump strength and speed are shown in the more intuitive blocks/blocks-per-second units rather than raw attribute values.
