Daily MLB picks driven by starting pitcher quality, bullpen leverage, ballpark factors, and platoon splits — calibrated against three-plus years of historical data. Verified track record. Full transparency.
Every MLB pick on this site comes from one pipeline: a quantitative model trained on three-plus years of game-level MLB data, pitch-by-pitch outcomes, current odds across major sportsbooks, and a fixed set of edge thresholds that decide whether a game becomes a play.
The model scans the day's slate as soon as confirmed starting pitchers are announced. For each game it produces a fair-value win probability and projected run total, then compares that probability to the implied probability priced at the moneyline, run line, and total. When the gap clears the edge threshold — and the bullpen, park, and lineup context confirm — the game becomes an official play with explicit unit sizing.
Baseball is the highest-volume sport on the slate, which means edges are smaller individually but compound across hundreds of plays. Discipline matters more here than in any other league.
MLB outcomes are dictated by who's on the mound, what the opposing lineup looks like against that pitcher, and where the game is being played. The model weights four core inputs:
Strikeout rate, walk rate, home-run-per-nine, and underlying contact quality (xFIP, SIERA-style metrics) drive the projection. A pitcher's true talent over the most recent rolling 10 starts gets weighted more heavily than season-aggregate ERA, which is too noisy across small samples.
Modern MLB games are decided by relief usage as much as by starters. The model tracks each team's rolling bullpen ERA, current rest state of high-leverage arms, and projected innings-remaining handed off after the starter. A tired closer-setup combination playing the third game in three days is a different pen than the one ranked on a season-long leaderboard. More on why F5 markets reward pitcher analysis and full-game markets reward bullpen analysis →
Run scoring varies by 15 to 25 percent across MLB parks for handedness-specific batters. Coors Field plays differently than Petco; Yankee Stadium's right-field porch matters; wind direction at Wrigley shifts totals. The model encodes park run-scoring multipliers and current-game weather (wind, temperature, humidity) into the projection.
Right-handed-heavy lineups against soft-tossing lefties price differently than the same teams' season averages suggest. The model uses confirmed lineups and platoon splits to project expected runs scored and allowed, then converts that to win probability and run-total expectation.
Each MLB play arrives with a fixed structure so you know exactly what you're betting and why:
Plays are released after starting pitchers are officially confirmed by both teams and the early-afternoon odds settle — typically by 1pm ET on weekdays for evening games and earlier on weekend day games. Confirmed pitcher identity is the largest single input to MLB projection, so we wait for it.
Coverage runs full regular season plus playoffs. MLB has the largest daily slate of any major US sport (15 games on a typical night) which means slate volume is highest here. The model is calibrated to fire only on real edges, so a typical day might produce two to five plays. On low-edge days it stays out entirely.