Stealing a base is a bet: succeed and the runner advances, fail and the team loses both the runner and an out. This calculator finds the success rate at which that bet pays off, using run expectancy, the average runs scored from each base-out state.
How it works
Every steal attempt moves the team between run-expectancy states. Define the gain on success and the loss if caught, then solve for the success probability where the expected change is zero:
gain = RE(success) - RE(now)
loss = RE(now) - RE(caught)
break-even = loss / (gain + loss)Because being caught removes the runner and adds an out, the loss is large relative to the gain, which pushes the break-even rate well above half.
Worked example: runner on first, no outs
Using a typical MLB run-expectancy table:
- Current RE (1st base, 0 outs): ~0.83
- Success RE (2nd base, 0 outs): ~1.07 → gain ≈ 0.24
- Caught RE (bases empty, 1 out): ~0.24 → loss ≈ 0.59
- Break-even = 0.59 / (0.24 + 0.59) ≈ 71%
A runner who succeeds 75% of the time adds expected runs on net. A runner who succeeds only 65% of the time is costing the team runs. The math is not symmetrical: the cost of getting caught (losing both the runner and an out) is larger than the gain of advancing, which is why the break-even sits so far above 50%.
How the break-even changes by situation
| Situation | Typical break-even rate |
|---|---|
| 1st → 2nd, 0 outs | ~70–72% |
| 1st → 2nd, 1 out | ~74–76% |
| 1st → 2nd, 2 outs | ~65–67% |
| 2nd → 3rd, 0 outs | ~75–80% |
| 2nd → 3rd, 1 out | ~80–85% |
The two-out situation has a lower break-even for 1st→2nd because getting caught ends the inning, removing what was a modest run-expectancy anyway, while succeeding puts a runner in scoring position with any hit. Two-out steals are actually the easiest to justify mathematically — if your runner has a good success rate and the batter is unlikely to drive them in from first, the expected-run math often favors running.
The two-out, 3rd base situation is the trickiest: the gain of scoring a run versus the loss of an out that ends the inning depends heavily on the batter, score, and inning.
Win probability versus run expectancy
This model uses run expectancy, which measures average runs scored from a state for the rest of the inning. That is the right framework across a full season, but in individual games — especially late-inning, close situations — win probability added (WPA) is a better lens.
For example, stealing second base with a one-run lead in the bottom of the ninth adds different win-probability value than the same steal in a 5-run blowout. In tie games, getting caught can dramatically swing win probability. Managers who think in terms of WPA rather than run expectancy will steal bases more conservatively in high-leverage late-game situations and more aggressively in blowouts or lower-leverage early innings.
Use run-expectancy break-even for strategic baseline analysis; layer in game situation and leverage when making in-game decisions.