How accurate are your MLB picks?

Our XGBoost machine learning model trained on 2,900+ games achieves 62.9% accuracy on head-to-head comparisons. We track every pick with full transparency — win rate, ROI, and edge breakdowns are all public.

What data does the MLB prediction model use?

Our model uses 22 signals including Statcast pitcher xwOBA, FIP/xFIP/SIERA, bullpen fatigue, confirmed daily lineups with platoon splits, park factors, weather and wind, home plate umpire tendencies, catcher framing, rest and travel, and Vegas line movement.

Are these MLB picks really free?

Yes, 100% free with no login required. Every daily pick, edge analysis, pitcher matchup, and player prop projection is available at no cost.

When are today's MLB picks posted?

Picks are generated 1-2 hours before the first game each day once confirmed lineups, pitcher assignments, and umpire data are available from the MLB API.

What is the MLB Edge Finder?

The Edge Finder identifies games where our model's win probability significantly differs from the Vegas implied probability. These gaps represent potential value bets, ranked by edge size with our GOLD, FIRE, and PLATINUM tier system.

Analytics

Pythagorean Wins and Regression: Spotting Lucky and Unlucky MLB Teams

6 min read · Educational guide

Records Lie (Sometimes)

At any point in the baseball season, some teams have records that genuinely reflect their quality — and some don't. A team sitting at 45-35 might be a true .560 club, or they might be a .500 team that got hot in one-run games for two months. Knowing the difference is one of the most reliable edges in MLB futures and moneyline betting.

The tool that helps you see through the noise: Pythagorean win expectancy.

What Pythagorean Win Expectancy Is

Developed by baseball statistician Bill James, the Pythagorean win expectancy formula estimates what a team's winning percentage should be based on how many runs they've scored and allowed — rather than their actual win-loss record.

The simplified formula:

Expected Win% = Runs Scored² / (Runs Scored² + Runs Allowed²)

A team that scores 500 runs and allows 480:

500² = 250,000
480² = 230,400
Expected Win% = 250,000 / 480,400 = 52.0% — roughly 84 wins over 162 games

If that team's actual record projects to 90 wins at that run differential rate, they're significantly outperforming their run differential. That gap is typically attributable to luck — specifically, timing of runs and performance in close games.

Why the Gap Exists: Run Timing and Close Games

Two teams can have the same season runs scored and allowed totals but dramatically different records depending on when those runs came.

Consider:

Team A wins three games 9-1 but loses two games 3-2. Net: +19 runs, 3-2 record.
Team B wins five close games 3-2 and 4-3. Net: +5 runs, 5-0 record.

Over a full season, runs cluster. Teams that consistently win close games outperform their run differential; teams that lose close games underperform it.

Bullpen quality, managerial in-game decisions, and sheer randomness in one-run games all contribute. But research consistently shows that one-run game performance has low year-to-year correlation — teams that outperform their Pythagorean expectancy in one year tend to regress toward it the next.

Using This for Betting

Two types of opportunities emerge:

Overperforming teams (potential sells): A team with a 55-45 record but a Pythagorean expectancy suggesting 48-52 is playing above their true level. Their future moneylines may be overpriced relative to actual quality. The market often prices teams off their record rather than run differential.

Underperforming teams (potential buys): A team with a 45-55 record but a Pythagorean expectancy of 52-48 has been unlucky — often losing close games they should have split. Their lines may be underpriced. This is especially valuable for futures bets (division winner, wild card) when a team's price reflects a bad record that understates their talent.

Third-Order Wins: Going Even Deeper

Standard Pythagorean analysis uses actual runs scored and allowed. More sophisticated versions use component-level run estimators — calculating expected runs from hits, walks, and other underlying events rather than actual runs. This further adjusts for sequencing luck in run scoring itself.

FanGraphs' BaseRuns-based win totals attempt this. They're more complex to calculate but identify teams whose actual run scoring was itself lucky or unlucky, adding another layer of regression signal.

For most bettors, standard Pythagorean analysis is sufficient and already captures most of the actionable signal.

Sample Size and Timing

Pythagorean analysis becomes more reliable as the season progresses and run sample sizes grow. In April with a 15-game sample, run differentials are too noisy to be meaningful. By July with 70+ games, the run differential carries real predictive weight.

The most actionable window is typically mid-season through August — enough sample size to matter, far enough from the end of the season that prices haven't fully corrected.

Combining Pythagorean With Other Metrics

Pythagorean analysis tells you a team's overall quality relative to their record. It works best combined with:

Strength of schedule — Did their run differential come against weak opponents?
Home/road splits — Is a favorable park inflating the run differential?
wRC+ and FIP for key players — Are underlying offensive and pitching metrics consistent with the run differential?

A team that looks lucky by Pythagorean analysis but also has elite underlying FIP and wRC+ may simply be better than their run differential suggests, with the luck running in the other direction. Multiple signals always outperform any single metric.

Bottom Line

Pythagorean win expectancy is one of the most reliable regression tools in baseball analytics. Teams that significantly outperform their run differential are likely to cool off; teams that significantly underperform it are likely to improve. By mid-season, this gap is one of the clearest signals for identifying overpriced and underpriced teams — especially for futures and series-level bets where true team quality matters more than game-by-game variance.