MLB Stolen Base Predictions Tonight: Daily SB Probability Rankings (2026)

The stolen base narrative has shifted dramatically since 2023. Larger bases (introduced under the new collective bargaining agreement) measurably improved success rates league-wide, and several teams began deploying aggressive base-running strategies that had been dormant since the stolen base peak of the 1980s. The result is more steals per game than any recent season — which increases the value of the SB category in fantasy formats that track it.

Even the most prolific base stealers steal in only 25-35% of their games at peak production. The league average batter steals in under 7% of games. This rarity is what makes per-game SB probability — not season totals — the correct metric for daily lineup optimization and waiver wire decisions.

An elite base stealer running at a 40-steal season pace over 150 games steals in approximately 27% of his games — roughly 4x the 6.8% baseline. A 60-steal pace player steals in approximately 40% of games.

These per-game rates are season averages. The per-night rate varies significantly based on the opposing pitcher's hold rate, the catching arm behind the plate, the score of the game, and whether the batter gets on base at all. Seasonal rates tell you the player's true SB talent. Per-game model predictions tell you which nights that talent is most likely to express.

Some fantasy resources express SB value as “expected SBs per 100 games” or “season projection at current pace.” These are useful for draft rankings. For daily start/sit and waiver wire decisions, the relevant metric is a binary per-game probability: will this batter steal a base tonight?

P(SB ≥ 1) answers that question directly. A probability of 28% means there is roughly a 1-in-3.5 chance of a steal tonight — not 1-in-15 like the baseline, and not a guarantee. The model surfaces the batters where tonight's specific conditions (opposing pitcher, catcher, recent form) point toward elevated steal probability — distinguishing which nights of a fast player's season are most likely to produce the stat.

The model was backtested across multiple recent MLB seasons. Because SB is a low-frequency event, the relevant metric is not absolute accuracy but relative lift — how much more likely are top-ranked batters to steal versus a random pick?

The 8.9 percentage-point net signal — top decile vs. the shuffle test — confirms the model is capturing genuine information, not noise. A shuffle test takes the same batter-games, randomly assigns P(SB) rankings, and checks whether the “top decile” in a random ranking also hits at 15.7%. It does not — the shuffle test produces 6.8%, the base rate. The gap between 15.7% and 6.8% is the model's real contribution.

Beyond the 20-game rolling rate and Bayesian prior, the model encodes several matchup-level signals:

These players have sustained 30+ steal seasons and offer enough multi-category value that they are strong picks even in leagues where you do not desperately need SB:

Players whose primary fantasy value is SB production, with secondary contributions in R and occasionally BA. Draft to anchor the SB category; accept lower production elsewhere.

A player ranked #1 in season SB is not necessarily the best P(SB) pick on any given night. The per-game model adjusts for several factors that season rankings cannot capture:

In standard 5x5 Roto, stolen bases are one of five equally weighted hitting categories. Every steal adds one to your SB total, and the standings are determined by your cumulative SB rank across the league. Finishing last in SB costs the same standings points as finishing last in home runs. Finishing first in SB is as valuable as finishing first in any other category.

The practical implication: in a 12-team league, the difference between first and last in SB can represent 11 standings points — a large enough gap to swing the title. Drafting 1-2 elite stealers at appropriate ADP and using per-game P(SB) to optimize lineup decisions daily is the correct approach. Do not treat SB as a secondary category in 5x5.

In 6x6 formats (which typically add OBP or SLG as the sixth hitting category and QS or K/9 as the sixth pitching category), SB remains a direct scoring category with identical weight. The addition of a sixth hitting category does not dilute SB — it adds a sixth independent dimension. In fact, players who contribute to SB, OBP, and HR simultaneously (Acuña, Witt Jr., Carroll) become even more valuable in 6x6 because they help in more categories per draft slot.

In points leagues, stolen bases convert to a fixed point value per event rather than a category rank. The value varies significantly across platforms:

In points leagues, the primary driver of a speed player's value is batting average and run production — the counting stats that accumulate across every plate appearance. SB adds meaningful bonus points but is not the defining valuation lever unless your specific league's SB weighting is unusually high. Use P(SB ≥ 1) to identify the nights a speedster is most likely to contribute the SB bonus on top of his normal line.

A player who stole 45 bases last season but is on pace for only 15 through the first six weeks is not the same SB asset as he was at draft time. Season totals anchor perception in ways that resist updating — the “46-SB last year” number stays in the fantasy manager's head long after the player has changed his role, lost a step to injury, or been dropped in the batting order where steal opportunities are fewer.

Our model corrects for this explicitly. The 20-game rolling rate with Bayesian shrinkage updates nightly. A player who has not stolen a base in his last 18 games will show a materially lower P(SB) in the model output than his season total or draft ranking would suggest — giving you an early signal to consider alternatives before the season narrative catches up to the data.

This is the most significant matchup factor in stolen base prediction that fantasy managers routinely ignore. Elite defensive catchers — Adley Rutschman (BAL), William Contreras (MIL) — throw out runners at 30-35%, well above the 25% league average. Running against an elite arm reduces expected steal probability by roughly 20-25%.

Conversely, catchers with below-average arms (caught-stealing rate below 20%) are exploitable by even moderate-speed runners. A catcher throwing out only 15% of steal attempts is such a soft target that managers green-light runners who would normally not be running. The per-night catcher matchup often matters as much as the base stealer's own speed.

Our model encodes opposing catcher CS rate as an explicit feature — one of the most defensible differentiators versus generic SB projection tools that rank players only by season rate.

Pitcher tempo to the plate is the other half of the steal equation that season totals cannot capture. Left-handed pitchers have a structural advantage over base stealers because they naturally face first base during their delivery — but more importantly, pitchers with quick step deliveries, slide-steps, or exceptionally fast times to the plate suppress steal attempts regardless of handedness.

A manager who green-lights a base stealer every night without checking the opposing pitcher's hold tendencies is leaving information on the table. The per-game model incorporates pitcher-level hold tendencies, which means nights where a known hold-specialist is starting will show depressed P(SB) even for elite speedsters — a signal that season totals and generic projections cannot provide.

For 5x5 Roto managers who flex players across positions, the per-night P(SB) ranking helps identify the maximum-value lineup configuration. A speedster with a 25% P(SB) on a night where his catcher matchup is favorable and the opposing pitcher is slow to the plate is a must-start. The same player at 12% against an elite catcher and quick-release lefty is a lower-priority start in a tight race for the SB category.

Speed is consistently undervalued on the waiver wire because casual managers check season SB totals and see a player's hot steal streak as “used up” after the fact rather than as a signal. The P(SB) model helps identify:

For multi-category fantasy managers, the /home-runs page and the SB projections page together identify the rare nights where a multi-tool player (Witt Jr., Carroll, De La Cruz) is highly ranked in both categories simultaneously. These “confluence nights” are the most valuable single-player lineup slots for managers contending in both SB and HR categories. Also check /pitcher-quality-starts to see if tonight's opposing pitcher is a high-K arm — high-K pitchers tend to post faster paces and may suppress steal opportunities on the same night.