Every stat and term you'll see on the picks and results pages.

Pitcher Stats

ERAEarned Run Average — runs allowed per 9 innings. Lower is better.
FIPFielding Independent Pitching — ERA estimate using only strikeouts, walks, and home runs. Strips out defense and luck.
SIERASkill-Interactive ERA — advanced ERA estimator that accounts for batted-ball type and strand rate. More predictive than FIP.
K%Strikeout rate — percentage of plate appearances ending in a strikeout.
BB%Walk rate — percentage of plate appearances ending in a walk.
Barrel%Rate of hard, well-hit balls allowed. Lower = harder to square up.
Avg EVAverage exit velocity allowed off the bat. Lower = weaker contact.
Stuff+Pitch quality rating (100 = league average). Higher = nastier stuff.
Velo TrendRecent fastball velocity change vs season baseline. Positive = gaining velo.
Season IPInnings pitched this season. Context for workload and sample size.

Bullpen, Lineup & Defense

BP xwOBABullpen expected wOBA allowed. Lower = tougher pen.
BP Whiff%Bullpen whiff rate on swings. Higher = more swings and misses.
Lineup xwOBALineup expected wOBA — PA-weighted, platoon-adjusted estimate from batter K%, BB%, and barrel rates. Higher = stronger lineup.
Lineup Barrel%How often this lineup barrels the ball. Higher = more power.
Def OAAOuts Above Average — team fielding quality. Higher = better defense.

Betting Terms

Win ProbabilityModel's estimated chance the picked team wins, derived from predicted run differential and calibrated against historical results.
TierConfidence level based on win probability. Higher tiers indicate stronger model conviction. Tier names and thresholds are set by the model — not manually adjusted.
Kelly SizeRecommended bet size using the Kelly criterion (fractional). Shown in units, where 1u = 1% of bankroll. Sizes bets proportional to edge.
EV%Expected value — the percentage edge over the market odds. Positive EV means the model sees value the line doesn't reflect.
O/U EdgeDifference in runs between the model's predicted total and the posted line. Larger edges indicate stronger over/under signals.
Cover ProbabilityFor run line bets (-1.5), the estimated chance the favored team wins by 2 or more runs.
Model ScoreComposite confidence score (0–52) combining neighbor accuracy (similar historical games), regime accuracy (feature-combination track record), and game familiarity.

Bet Sizing

Bets are sized by the edge and the odds, not the tier. A Strong play with favorable odds can hit the 5u maximum the same way a Max play can — and a Max play with shorter odds might size to less. 5u is the ceiling, not a tier.

Confidence Tiers

Every game the model evaluates receives a confidence tier based on how strongly it favors one side. Higher tiers reflect stronger model conviction.

MaxThe model’s highest conviction picks. These are rare and represent the largest disagreements between the model and the betting market in favor of a clear favorite.
StrongHigh-confidence selections where the model sees a significant edge.
Moderate FavoriteSolid picks where the model favors a team the market also has as the favorite, but sees additional value beyond what the line reflects.
Moderate DogThe model favors a team the market has as an underdog, at a mid-level confidence range. These are flagged with a caution indicator — the model is directly disagreeing with the market, and that disagreement has been less reliable at this confidence level. Use your judgment.
Lean DogThe model likes a team the market doesn’t, at a lower confidence level. These are surfaced as active picks because the plus-money payouts compensate for the lower conviction — you don’t need to win as often when the payout is larger.
Lean FavoriteLower-conviction reads on a market favorite. These are tracked but not recommended. The edge at this confidence level hasn’t historically justified laying juice on a favorite.

Dog vs. Favorite

This distinction reflects whether the model's picked side is a market underdog (Dog) or market favorite (Favorite). The same confidence level plays very differently depending on which side of the line you're on. A Dog bet needs to win less often to be profitable. A Favorite bet needs to win more often to overcome the juice. The model accounts for this.

Over/Under Tiers

O/U picks are tiered by the size of the gap between the model's projected game total and the posted line. Larger gaps indicate stronger conviction that the line is mispriced.

OU MaxThe model sees a large discrepancy between its projected total and the posted line.
OU StrongA significant gap. High confidence in the direction.
OU ModerateA meaningful gap. This is the minimum tier DaftyDimes recommends betting in production.
OU LeanA smaller gap. These are tracked for transparency but not recommended as bets — the edge is too thin to reliably overcome the vig.

What determines the tier?

The model produces run projections for each side of every game. Those projections, combined with how the market is pricing the game, determine both the tier and whether the pick is recommended. Not every game the model evaluates becomes a pick — only those where the conviction and market pricing align favorably.