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.

Strategy

Kelly Criterion for Baseball Bettors: Sizing Bets Without Going Broke

7 min read · Educational guide

The Bet-Sizing Problem Nobody Takes Seriously

Most sports bettors spend the majority of their time picking games — analyzing matchups, reading injury reports, studying pitching data. They spend almost no time thinking about how much to bet on each game. That's backwards.

Bet sizing is arguably as important as pick quality. A bettor who correctly identifies games with a 5% edge but sizes bets randomly will underperform a bettor who finds a 3% edge and sizes every bet optimally. The math behind this is not intuitive, which is why the Kelly Criterion is worth understanding.

What Is the Kelly Criterion?

The Kelly Criterion is a mathematical formula developed by Bell Labs scientist John Kelly in 1956. It calculates the optimal fraction of a bankroll to bet on any given outcome, given a known edge. The formula maximizes long-run bankroll growth while avoiding ruin.

The formula for sports betting:

f = (bp - q) / b

Where:

f = the fraction of bankroll to bet
b = the decimal profit from a winning bet (decimal odds minus 1)
p = your estimated probability of winning
q = probability of losing (1 - p)

Example: You're betting a team at +110 (decimal 2.10) and your model says they win 55% of the time.

b = 1.10 (profit on a $1 bet)
p = 0.55
q = 0.45
f = (1.10 × 0.55 - 0.45) / 1.10 = (0.605 - 0.45) / 1.10 = 14.1%

Kelly says to bet 14.1% of your bankroll on this game. That might feel like a lot — and it is, which is why most practitioners modify it.

Why Most Practitioners Use Fractional Kelly

Full Kelly is mathematically optimal if your probability estimates are exactly correct. In practice, they never are. Model error, overconfidence, and incomplete information mean your edge estimates are noisy.

When you overestimate your edge and bet full Kelly, the volatility is brutal. A string of losses — entirely normal in any betting endeavor — can devastate a bankroll before long-run expected value has time to manifest.

The standard practical solution: Half Kelly or Quarter Kelly.

Half Kelly: Bet half the formula output. This cuts volatility substantially while preserving most of the expected growth rate.
Quarter Kelly: Even more conservative. Used by bettors who are less confident in their edge estimates or prioritize capital preservation.

For recreational baseball bettors, Half Kelly is a sensible default. Quarter Kelly makes sense when you're new to a model or operating a smaller bankroll where a bad run would be psychologically damaging.

Setting a Realistic Bankroll and Unit Size

Before Kelly can be applied, you need a defined bankroll — the total amount set aside exclusively for sports betting. This should be money you can afford to lose in its entirety. Treat it as its own account, completely separate from living expenses or savings.

From that bankroll, define your base unit:

A standard unit is typically 1–3% of total bankroll
Kelly bets are expressed as multiples of your unit
Cap any single bet at 3–5% of bankroll regardless of Kelly output

Capping is important. Kelly can output surprisingly large bet sizes when estimated edges are large, and estimated edges are often over-estimated. A hard cap is a prudent guardrail against model overconfidence.

Kelly and Edge Estimation: The Feedback Loop

Kelly's value extends beyond bet sizing. It creates discipline around edge estimation. Because the formula requires a specific probability, you're forced to produce a number — not a vague feeling that a team is "due" or "playing well lately."

Over time, tracking your Kelly-sized bets against results lets you calibrate whether your edge estimates are accurate. If you're consistently losing on bets where Kelly suggested 8% edge, your model needs revision. This feedback loop is one of the most valuable aspects of systematic betting.

What Kelly Doesn't Solve

Kelly is a bet-sizing tool, not a pick-making tool. It can't tell you which games have edge — only how much to bet once you've identified edge. If your probability estimates are systematically wrong, Kelly will systematically over-size bad bets.

It also doesn't account for correlated outcomes. If you're betting three games on the same day where results might correlate, strict Kelly across all three may over-allocate exposure.

And it assumes a static bankroll for calculation purposes, while real bankrolls fluctuate. Many practitioners recalculate Kelly as a percentage of current bankroll after each bet — the theoretically correct approach.

Bottom Line

The Kelly Criterion gives baseball bettors a principled, mathematically grounded answer to "how much should I bet?" Used at half or quarter Kelly, it prevents the two most common sizing errors — betting too much on uncertain edges and risking ruin, or betting too little on strong edges and leaving growth on the table. Set your bankroll, define your units, calculate your edge honestly, and let the math guide the sizing.