The Book

1. Data-Driven Decisions: Leverage Run/Win Expectancy and wOBA

What you are about to read is an attempt to quantify or qualify the ideal strategies in baseball.

Context is everything. Baseball strategy, like any complex system, is meaningless without understanding its context. The game unfolds through 24 distinct base/out states, each carrying a specific "run expectancy" (RE) – the average runs a team will score from that point until the inning ends. Similarly, "win expectancy" (WE) quantifies the probability of winning from any game state.

Quantifying value. Events like singles, doubles, and home runs have measurable "run values" (RV) based on how they shift RE. For instance, a home run with bases empty adds exactly one run, but a grand slam adds 3.302 runs above what an average batter would produce in that situation, accounting for expected runs from runners already on base. This precise quantification allows for objective evaluation of player actions.

Weighted On-Base Average (wOBA). To assess player offensive contributions more accurately than traditional OBP or SLG, wOBA assigns run values to each offensive event (walks, singles, doubles, triples, home runs) and scales them to an OBP-like scale. This metric, along with Markov chains (which model state-to-state transitions), provides a robust framework for analyzing strategy and player performance, moving beyond intuition to empirical data.

2. Streaks and Matchups: Randomness Dominates, Except for Pitcher Health and Platoon Effects

Knowing that a hitter has been in or is in the midst of a hot or cold streak has little predictive value.

Hitter streaks are mostly luck. Despite common belief, a hitter's recent "hot" or "cold" streak has minimal predictive power for future performance. Small sample sizes mean observed fluctuations are largely random. Managers should assume a hitter will perform at their projected norm, regardless of recent results, using past performance only as a tiebreaker.

Pitcher streaks have some carryover. Unlike hitters, pitchers, especially starters, show some predictive value in their streaks.

• Hot starters tend to perform better than expected in their next outing (around 11 wOBA points better).
• Cold starters, particularly inexperienced ones, tend to continue performing worse, possibly due to underlying health issues or mental factors.
• Experienced pitchers who are "hammered" early may show some carryover, but less dramatically.

Matchups are often overrated. Small sample sizes in head-to-head batter/pitcher matchups (e.g., 20-30 plate appearances) are unreliable. A player's overall career performance against the league is far more predictive than their limited history against a specific opponent. Even against "famiglias" (similar pitchers), the effect is often negligible.

3. Clutch Performance: Mostly a Myth; Overall Skill and Situational Penalties Matter More

For all practical purposes, a player can be expected to hit equally well in the clutch as he would be expected to do in an ordinary situation.

Clutch skill is minimal. While "clutch performance" (actual results in high-pressure situations) exists due to randomness, "clutch skill" (an inherent ability to perform better under pressure) is statistically very small. The difference between a good clutch player and an average one is only about 6 points of wOBA, dwarfed by overall hitting skill variations.

Pinch hitting penalty. Players are significantly less effective as pinch hitters than as starters, suffering a roughly 34-point wOBA penalty. This is likely due to a lack of physical and mental readiness from sitting on the bench.

• There's no evidence of a "pinch-hitting specialist" skill.
• A pinch hitter must be substantially better than the player he replaces to make the substitution worthwhile.

Pitchers and the stretch. Pitchers perform slightly worse (around 5 wOBA points) when pitching from the stretch compared to the windup. This "stretch penalty" varies among pitchers, suggesting some are better at it than others. However, there's no evidence that batters perform differently when facing a pitcher from the stretch.

4. Lineup Optimization: Small Gains, But Strategic Placement of Pitchers is Crucial

Your three best hitters should bat somewhere in the #1, #2, and #4 slots.

Maximize plate appearances for best hitters. The batting order is a continuous loop, meaning top spots get more plate appearances. Therefore, a team's best hitters should occupy the #1, #2, and #4 slots to maximize their opportunities.

• #1 and #2 hitters should prioritize OBP (walks).
• #4 hitter should prioritize power (extra-base hits).
• #3 and #5 hitters should be the team's fourth and fifth best, with the #5 hitter often having slightly better opportunities than #3.

Lineup impact is modest. Optimizing a lineup can yield 10-15 extra runs per season, equivalent to about one to one-and-a-half wins. While small, these marginal gains are valuable. Factors like strikeout rates or the ability to move runners over on outs are largely irrelevant for starting lineup construction.

Pitchers in the NL are a major drain. In National League parks, the pitcher's spot in the lineup is a huge offensive liability, costing approximately 68 runs per season compared to having a designated hitter.

• Moving the pitcher from the #9 to the #8 spot (the "second leadoff hitter" theory) can gain a couple of runs per season by setting up the top of the order.
• Managers should aggressively pinch-hit for below-average pitchers, especially if the bullpen can cover the innings.

5. Platoon Advantage: A Powerful, Underutilized Tool for Both Hitters and Pitchers

Player-to-player variations in platoon splits are more important for pitchers than for hitters.

Platoon splits are real and variable. The average platoon advantage (performing better against opposite-handed opponents) is a significant 44 points of wOBA. However, this split is not constant for all players.

• Pitchers and switch hitters show the most variation in their platoon skills.
• For regular left- and right-handed hitters, individual platoon splits are generally small, so assuming an average split is often sufficient.

Exploiting pitcher platoon splits. Pitchers' platoon splits are more reliably measurable and impactful.

• A right-handed pitcher's platoon split is reasonably accurate after about 700 plate appearances against left-handed hitters; for a lefty, it's around 450.
• Pitchers with small platoon splits (e.g., those relying on change-ups, overhand curves, or knuckleballs) are valuable as they are less susceptible to lineup adjustments.

Strategic applications:

• Platooning starters: An effective way to gain about one extra win per season by ensuring the platoon advantage.
• LOOGYs (Left-handed One-Out Guys): Highly valuable specialists who are excellent against lefties and at least average against righties, allowing them to face key left-handed batters and potentially lesser right-handers.
• Bullpen construction: Teams should prioritize acquiring pitchers with diverse platoon splits to maximize matchup advantages.

6. Reliever Usage: Maximize High-Leverage Opportunities and Challenge Traditional Workload Limits

In terms of leverage, for every one win you can get pitching your ace reliever with a three-run lead, you get two wins with a two-run lead, and three wins with a one-run lead.

Leverage is paramount. The impact of an ace reliever varies dramatically with game context.

• A three-run lead in the 9th inning is a low-leverage situation; an ace reliever only adds about 0.02 wins over an average reliever.
• A one-run lead in the 9th offers triple the leverage (0.06 wins), and a one- or two-run lead in the 8th inning provides even better opportunities for an ace.

Avoid wasted outings. Managers frequently misuse ace relievers in low-leverage situations (e.g., 19% of appearances for top aces are in games with minimal impact on the outcome). This wastes their valuable contributions.

• Aces should be deployed when the "leverage" (impact of a single play on win probability) is high, ideally 1.50 or greater.
• This often means bringing them in earlier than the 9th inning, especially with runners on base.

Relievers are underutilized. Modern relievers are capable of handling much heavier workloads than currently imposed.

• Relievers from past eras (e.g., 1970s) threw more pitches per outing and had comparable or longer careers.
• Pitching an ace for one or two batters in the 8th inning does not diminish their effectiveness in the subsequent 9th inning.
• Even heavy workloads (16-24 batters over a five-day period) show no negative impact on next-day performance.

7. Sacrifice Bunt: A Nuanced Strategy, Often Misunderstood, with Context-Dependent Value

If the opposing manager is thinking about attempting a sacrifice (with a runner on first and no outs and a non-pitcher at the plate), tell him that you will gladly give the runner second base in exchange for an out.

Sacrifice attempts are better than guaranteed outs. While simply trading an out for a runner advancing is a terrible strategy (significantly reducing run/win expectancy), a sacrifice attempt (which can result in singles, errors, or other outcomes) is often more valuable.

• In low-scoring environments or early in games, a surprise bunt attempt can yield as many or more runs than swinging away, especially for weak hitters.
• Late in close games, the increased chance of scoring at least one run can make a bunt attempt as good as or better than swinging away.

Context is critical. The decision to bunt depends on numerous factors:

• Defensive alignment: If the defense expects a bunt and plays in, it's harder to succeed. If they play back, a bunt is more effective.
• Batter's skill: Good bunters and fast runners are more effective. Poor/slow bunters should rarely sacrifice.
• On-deck hitter: Bunt more often if the next hitter has a low OBP/walk rate.
• Pitcher bunting: Average-hitting pitchers should bunt about half the time with a runner on first, no outs. Worst-hitting pitchers should bunt with one out, runner on first. Never sacrifice with 1st & 3rd, 1 out.

Game theory implications. The fact that bunting and swinging away often yield similar run/win expectancies suggests an equilibrium point. Managers must sometimes bunt (even if slightly suboptimal) to prevent the defense from always playing back, and vice versa.

8. Intentional Walks: Rarely Beneficial, Reserve for Specific High-Leverage Situations

Issuing an intentional walk to an outstanding hitter will almost always increase the opposing team’s run production.

Generally counterproductive. Intentionally walking a batter, even an outstanding one, almost always increases the opposing team's run production. The negative impact of putting an extra runner on base usually outweighs the positive of facing a weaker batter.

• Walking to set up a force on the lead runner is dangerous, as a subsequent unintentional walk would still advance the lead runner.
• If all batters are equal, intentional walks are rarely beneficial, winning an extra game only once every tens of thousands of games.

Specific high-leverage exceptions:

• Two outs, men on second and third: This is the most justifiable situation for an intentional walk, especially with an elite hitter at the plate and average hitters behind him.
• Bottom of the 9th (tied/one-run lead): An unprotected star hitter can be walked if it doesn't advance the lead runner.
• One out, man on third (and possibly second), with a double-play prone hitter on deck: This situation becomes highly attractive for an intentional walk.

"Pitching around" is ineffective. Trying to avoid a hitter by throwing outside the strike zone (pitching around) primarily increases walks and strikeouts, but does not significantly affect the quality of contact if the ball is put in play. If you'd rather have a walk than a normal non-walk outcome, just issue the intentional walk.

9. Baserunning: Disruptive to Batters; Stealing is Often Overused and Suboptimal

The stolen base attempt reduces the wOBA of the batter by 22 points, compared to the situation if the runner elected not to attempt to steal.

Runner on first disrupts the game. A runner on first base with less than two outs significantly disrupts the defense, giving the batter a 14-point wOBA advantage. However, this advantage is often negated or reversed by the runner's actions or the batter's profile.

• Young, fast, or groundball hitters are most disrupted by a runner on first.
• A "fast and disruptive" runner (one who attempts many steals) has an enormously negative influence on the batter, almost completely offsetting the defensive disruption.

Stolen base attempts are costly. On average, a stolen base attempt (including caught stealings, pickoffs, and balks) is a net negative in terms of run production.

• The break-even success rate for a stolen base is around 72% (ignoring batter effect), significantly higher than the league average of 67.8%.
• A stolen base attempt reduces the batter's wOBA by 22 points compared to the runner staying put.

Contextual break-even points. The optimal stolen base success rate varies by game situation:

• Most desirable: Tied games in later innings (break-even as low as 60%).
• Least desirable: Down by at least two runs in later innings (break-even 75-90%).
• Stealing against left-handed pitchers is much harder (8% lower success rate).
• Artificial turf provides a 3% advantage to the runner, especially for slow runners.

10. Game Theory: Unpredictability is Key to Optimal Strategy in Bunting, Pitchouts, and Pitch Selection

Even if bunting is not technically correct (the RE or WE is less than that from swinging away), given that the infield is anticipating a possible bunt, if it is at all close, which it often is (see the chapter on sac bunting), the batting team must sometimes attempt a sacrifice to keep the defense from playing all the way back.

Equilibrium in strategy. Many baseball strategies, like the sacrifice bunt, pitchouts, and pitch selection, involve game theory. Optimal play often means adopting a mixed, unpredictable strategy to prevent opponents from exploiting predictable patterns.

• If bunting is slightly suboptimal but the defense plays in anticipation, the offense must still occasionally bunt to keep the defense honest.
• Conversely, the defense must sometimes play agnostic to prevent the offense from always swinging away.

Pitchouts are a calculated risk. Pitchouts are costly (especially on a 2-1 count) but necessary to discourage steals, hit-and-runs, and squeezes.

• Managers must balance the cost of a pitchout against the probability of thwarting an opponent's running play.
• Against managers who rarely pitch out, teams should run more aggressively. Against "pitchout happy" managers, running should be curtailed, but occasionally attempted to prevent them from stopping pitchouts entirely.

Pitch selection is a chess match. Pitchers don't throw first-pitch strikes 100% of the time (even if capable) because batters would then always swing at them.

• Pitchers must mix up their first-pitch strikes and balls, and vary pitch types and locations, to keep batters guessing.
• The optimal mix depends on the count, game situation, and batter's tendencies. Rarely should a pitcher plan to throw only one pitch in any given situation.