MLB Projection Model

How does the model project a team's runs?

It starts at the league average and scales it by three things: the starting pitcher, the opposing lineup, and the environment. It does that twice, once across the innings the starter is expected to throw and once across the innings the bullpen is expected to throw. Flat adjustments for home field, velocity decline and bullpen fatigue go on at the end.

Each team's projected runs start from the league average (4.27 runs/game) and are scaled by multiplicative factors for the starting pitcher, opposing lineup, and environment (park, weather, umpire). The game is split into starter innings (5.5) and bullpen innings (3.5), with each segment rated separately. When a pitcher is flagged as an opener / bullpen game, the starter-side IP shrinks to ~1.5 and the balance is modeled as bullpen innings. The stacked projection is then shrunk toward league average by 70% trust (mirror of the K-prop shrinkage) to counter winner's-curse amplification at the tails of the factor stack. Additive adjustments for home-field advantage, velocity decline, and bullpen fatigue are applied last.

How does the model rate a starting pitcher?

It blends three windows of the same pitcher: his last three to five starts, his current season, and his prior year. Recent starts catch a mechanical change or a tiring arm that a season average smooths away. The prior year keeps the rating steady in April, when the current sample is too small to trust on its own.

Pitchers are rated by blending three time windows. The current-season line gets the most weight because it reflects the pitcher's actual form, but recent starts are weighted heavily to catch mechanical changes, fatigue, or injury effects that season averages would smooth over. The prior-year component provides stability early in the season when the sample is small.

Why does the model ignore ERA?

ERA mixes a pitcher's own work with his defense and with the order his hits arrived in, so it rates the team around him as much as the man. The model rates pitchers on FIP, xFIP and SIERA instead. All three are built only from outcomes a pitcher controls himself.

The model rates pitchers on a blend of three fielding-independent metrics. ERA is avoided because it conflates a pitcher's performance with team defense and sequencing luck. FIP measures what a pitcher controls directly (strikeouts, walks, home runs). xFIP stabilizes the home run component by assuming a league-average HR/FB rate, filtering out park and luck effects. SIERA goes further by accounting for a pitcher's batted ball profile (ground balls vs fly balls) and how that interacts with strikeout rates.

How does the model rate a lineup?

It rates each team's bats on wRC+, a park- and league-adjusted number where 100 is average. The season figure carries most of the weight, because offense settles into a stable number quickly. A 14-day window carries the rest, so a lineup change or a cold fortnight still moves the projection.

Team offense is measured by wRC+ (Weighted Runs Created Plus), a park- and league-adjusted stat where 100 is average. Season-long data gets more weight because offensive performance stabilizes quickly, but a 14-day window captures lineup changes and hot/cold streaks.

How do the park, the weather and the umpire change the projection?

Each one scales the environment factor in the run formula. The park sets a run, home run and handedness profile. Wind and temperature change how far a fly ball carries. The home plate umpire's strike zone moves strikeouts and walks, which moves runs.

Baseball is played outdoors in 22 of 30 stadiums, so environment matters more than in any other sport. The model adjusts for park dimensions (Coors Field plays very differently from Oracle Park), game-day weather (wind and temperature affect how far the ball carries), and the home plate umpire (whose strike zone tendencies shift strikeout and walk rates, impacting run scoring).

What does the model do in April, when the samples are small?

It leans on prior-year data and league averages, and trusts the current season more as it grows. A pitcher 20 innings into a season can look elite or awful on luck alone, so the model pulls him back toward what he did last year until he clears the innings threshold below.

Small samples are dangerous in baseball. A pitcher who has thrown 20 innings could look elite or terrible purely by chance. Early in the season, the model blends current stats with prior-year data and league averages using Marcel-style regression, gradually trusting the new data more as the sample grows. This prevents the model from overreacting to hot or cold April streaks.

What league averages anchor the numbers?

Every rating on this page is measured against a league-wide average, and those averages refresh through the season. A 3.50 FIP means one thing in a 4.40 run environment and another in a 3.90 one, so the anchors move with the year.

These are the league-wide averages that anchor the model. Every team and player is rated relative to them. A pitcher with a 3.50 FIP in a 4.40 ERA environment, for example, is suppressing runs by about 20%. These values are updated weekly as the season progresses to reflect the current run environment.

What happens when the model and the sportsbook disagree?

Fairline writes an internal observation row and shows you nothing new. A row is recorded when the model's fair odds and a book price differ by a flat 3%, and those rows feed calibration and closing-line research. They are never presented to you as a bet to place.

Fairline records an internal observation whenever the model's fair odds differ from the sportsbook's price by a flat 3%, the same threshold for every market. Each row carries a one-unit research weight for calibration and closing-line tracking. These observations are not surfaced as bets.

Command-line reference. The standalone model runner also includes a quarter-Kelly staking calculator with per-market EV thresholds (below). These are an offline reference and do not drive any user-facing surface. The internal measurement record uses the flat 3% threshold and one-unit research weight described above.

Which MLB markets does the model price?

It prices the moneyline, the run line, the game total, both team totals, first five innings, and starting pitcher strikeout props. Every game market comes out of the same Negative Binomial run distribution, so those prices can never contradict each other. Strikeout props are priced from a distribution of their own.

The model produces fair odds for every major MLB betting market. Here's what each one is and how the model handles it.

How are pitcher strikeout props priced?

The model estimates the pitcher's strikeout rate with a Bayesian posterior, then adjusts it for the lineup he faces, how many times he goes through that lineup, the umpire and the park. A Negative Binomial turns the resulting expected count into an over and under probability at every posted line.

The expected K count comes from a Bayesian Beta posterior over pitcher K% (prior centered on the previous-season K% with a Stuff+ adjustment, updated with current-season K/BF counts), combined with Log5 matchup against the lineup, a times-through-the-order penalty, and umpire/park K multipliers. When confirmed lineups are available, a 100,000-trial Monte Carlo simulation produces per-batter prop distributions (hits, total bases, HR) walking through the 9-man lineup; the K count mean from that pipeline is then used as the location parameter of the Negative Binomial that produces over/under probabilities for each posted line.

The Negative Binomial uses dispersion r=20, the same value the totals model uses for run scoring. Real MLB K count distributions sit at Var/Mean ≈ 1.25, slightly more dispersed than Poisson but tighter than runs. The shape is calibrated to ~40 years of historical K-count data.

How did the model do in the 2025 backtest?

Replayed across the full 2025 regular season, it called 55.8% of moneylines and scored a positive Brier skill against a coin flip. That is a retrospective measurement of one finished season, and it is pending data against the forward model-validation gate. It says nothing about what the model will do next season.

MLB is the most mature of the four models and the only one with a published full-season backtest. The numbers below come from replaying the pipeline across the entire 2025 regular season (2,421 games) using walk-forward data, meaning each game was projected with only the information that was available beforehand, with no peeking at results. All bets were graded at the prices that would have been available on a standard -110 market.

The Brier skill score is arguably the most important number here. It measures how well the model's probabilities are calibrated: whether a 60% forecast wins 60% of the time, whether a 30% forecast wins 30% of the time, and so on across the whole distribution. A model can pick winners at a high rate while still losing money if its probabilities are miscalibrated, and the Brier score penalizes that directly. The positive Brier skill and above-breakeven accuracy describe this backtest across more than 2400 games. This retrospective result is pending data against the forward model-validation gate.

When should you not trust this model?

Trust it least from Opening Day through mid-April, on bullpen and opener games, and when a lineup changed after the last pipeline run. Trust it most from mid-April on, with a confirmed starter, between teams at least 20 games into their seasons. The two lists below give the full version.

Every model has soft spots. Being honest about MLB's keeps you from over-betting situations the math doesn't actually cover well.

Where can I read the pitching side in more detail?

This page covers the starter and the bullpen in pieces. How starting pitchers and bullpen fatigue change a projected run total walks one game through both, with the fatigue adjustment worked out in runs.

The parameter values on this page are rendered from the running system and refresh periodically; when a weight or threshold changes, this page reflects it automatically.