MLB Fantasy Projections: Baseball Analytics and Forecasting
Baseball projections sit at the intersection of a 162-game marathon and the cold mathematics of probability — a combination that makes MLB one of the most statistically rich environments in all of fantasy sports. This page covers how baseball-specific projection models are built, what inputs drive their outputs, and where the real decision-making complexity lives. Whether the context is a March draft or a September waiver wire claim, understanding the mechanics separates reactive guesswork from structured analysis.
Definition and scope
A fantasy baseball projection is a forward-looking statistical estimate — typically expressed as a season-long or rest-of-season line — that quantifies how a player is expected to perform across categories like ERA, WHIP, batting average, home runs, stolen bases, and strikeouts. Unlike one-number NFL projections (fantasy points per game), MLB projections often carry 8 to 12 discrete statistical outputs per player, because standard rotisserie and points-based formats score across a wide category range.
The scope of baseball projection systems generally covers three populations: starting pitchers, relief pitchers, and position players. Each population requires a different model architecture. A starting pitcher projection is heavily weighted by stuff metrics, workload, and opponent quality. A position player projection leans on plate discipline rates, exit velocity, sprint speed, and park factors. The starting pitcher projection methodology that feeds into any serious MLB model is substantially more complex than its NFL counterpart, largely because a pitcher's role, health, and team context can shift multiple times within a single season.
How it works
The foundation of any MLB projection is a multi-year weighted average of player performance, adjusted for aging, role changes, and sample reliability. The Steamer projection system — one of the most publicly documented models in baseball analytics, hosted at FanGraphs — applies a three-year weighted mean with current-year data receiving the heaviest weight, then regresses each component toward league average based on sample size.
The regression step is not cosmetic. A hitter who posts a .380 BABIP over 200 plate appearances will have that figure pulled toward the league mean of roughly .300 because 200 PA is insufficient to treat as a stable signal. This connects directly to the principles behind regression to the mean in fantasy, which is one of the more consequential — and most ignored — forces in season-long baseball.
Beyond the baseline average, modern MLB projection builds incorporate:
- Batted ball and exit velocity data — Statcast metrics from MLB's Baseball Savant database, including xBA, xSLG, and barrel rate, translate underlying contact quality into expected production.
- Plate discipline rates — Walk rate and strikeout rate are among the most stable year-over-year statistics, making them reliable anchors in multi-year models.
- Park factor adjustments — A hitter moving from Petco Park in San Diego to Coors Field in Denver faces a dramatically different environment; projection systems apply multiplicative park factors drawn from multi-year data.
- Aging curves — Position players typically peak between ages 26 and 28 (Baseball Prospectus publishes aging curve research openly); pitchers often show decline signals earlier.
- Role and lineup context — A leadoff hitter in a strong offense accumulates more plate appearances and run-scoring opportunities than the same hitter batting seventh. Projection systems account for lineup position and team run environment.
The statistical inputs for fantasy projections page covers the broader input taxonomy across sports, but baseball's Statcast layer makes it uniquely granular.
Common scenarios
The three most common decision points where MLB projections are applied are draft preparation, in-season roster management, and trade evaluation.
Draft context is where projection output — a full season statistical line — gets compared against average draft position (ADP). When a pitcher's projected 185 strikeouts and 3.40 ERA is available in the 12th round of a 12-team draft, that gap between projected value and draft cost is the signal. Applying projections to draft strategy covers how to systematically exploit these gaps.
Waiver wire decisions are where in-season or rest-of-season projections become critical. A second baseman called up in late April carries a full-season projection that may not yet reflect his new role. Systems like ZiPS (from Dan Szymborski, documented at FanGraphs) update frequently enough to capture these context changes. Using projections for waiver wire decisions addresses the timing and weight to assign these updates.
Trade evaluation requires comparing two players' rest-of-season projections adjusted for positional scarcity and scoring format. A closer with 30 projected saves in a standard 5×5 rotisserie league has a different trade value than the same closer in a points-only format where saves receive minimal weight. Scoring format impact on projections quantifies exactly how format reshapes player values.
Decision boundaries
The clearest analytical boundary in MLB projections sits between process-driven forecasting and outcome-chasing — the difference between using a projection model and just rostering whoever went 3-for-4 last night.
A secondary but important boundary separates floor projections from ceiling projections. A player with a floor of 20 home runs and a ceiling of 38 is a fundamentally different asset than one with a floor of 28 and a ceiling of 32. The first profile suits a best-ball format where upside is rewarded; the second suits a head-to-head league where consistency matters more. Floor and ceiling projections details this distinction with position-specific examples.
Projection confidence intervals are the tool that makes this distinction operational. Rather than a single point estimate, a well-built model outputs a distribution — and the width of that distribution carries real information. A pitcher with a history of elbow inflammation carries a wider distribution than an equally talented pitcher with 400 consecutive innings of clean health. Projection confidence intervals explains how to read and apply that uncertainty band in live roster decisions.
The Fantasy Projection Lab home provides the broader modeling framework within which all sport-specific projections, including MLB, are developed and documented.