How we win
Chasing wins is a losing strategy.Exploiting value is the only one that works.
The bettors who win consistently - across poker, blackjack, baseball, and every other market - aren't the ones who pick the most winners. They're the ones who only bet when the math is on their side. This page explains what that actually means, and why our model lives by it.
01 - The Moneyball thesis
The books don't price for sharps. They price for the room.
In 2002, the Oakland A's won 103 games on a payroll that was a fraction of the Yankees'. Billy Beane didn't buy better players - he bought undervalued ones. On-base percentage was a known stat. The market just hadn't priced it correctly. Beane noticed, and exploited the gap until the rest of the league caught up.
Sportsbooks are no different. Their odds are a price, and that price is set to balance bets across the entire betting public - not to reflect what's actually going to happen on the field. When the average bettor over-weights name recognition, recent narrative, or last night's box score, the price moves with them. That gap - between what a player's home-run probability should be and what the book is selling it at - is the edge.
We're not predicting who will hit a home run. We're finding the players the books have priced wrong.
02 - The poker proof
Pocket aces lose all the time.
Heads-up at the table, you look down at two aces - the strongest starting hand in poker. You're a heavy favorite. And almost one time in five, you lose anyway.
The strongest hand. Still loses 18% of the time.
Second strongest. Loses almost a quarter of the time.
A losing poker player sees aces crack to a flushed seven-five and folds aces the next time. A winning player doesn't. The math doesn't change because the last hand went sideways. You play premium hands every time you get them, because the math is on your side over the long run.
The same rule applies to our picks. Some days a +12% edge HR pick won't cash. That doesn't mean the model was wrong. It means variance happened - exactly as the math predicted it would, some of the time.
03 - The blackjack lesson
A tiny edge × a lot of hands = a lot of money.
The MIT Blackjack Team didn't have a magic system. They had card counting - a method that gave them about a 1.5% edge over the house when the deck was favorable. That's it.
1.5% sounds like nothing. They won millions. Why? Because they only played when the count was in their favor, they sat out when it wasn't, and they played thousands of hands. Variance gets smaller the more hands you play. Edge doesn't.
Our model regularly surfaces HR picks at edges of 5%, 10%, even 15%. Far thicker than what the MIT team had. The catch is the same one: you have to actually wait for the edge - and stay out when it isn't there.
04 - A real-world pick
What an edge actually looks like.
Same player, same matchup. Two different prices. One is a bet. The other is a trap. The difference isn't whether the player will hit a home run - it's whether the book's price gives you a margin worth taking.
Vegas overpricing the player
The book thinks the player is more likely to homer than our data says. Even at +233 - looks fun, looks payable - every bet here is a slow leak. We pass.
Vegas underpricing the player
Same player, different night. The book is shading him short. Our data says the real probability is higher than the price implies. That is the edge - and the math says hammer it.
The trap: the negative-edge pick might still hit. The player might homer. You'll feel like a genius. But run the same play a thousand times and you lose money on every one. Positive edge is the opposite - you'll lose plenty of individual nights and still come out ahead over a season.
05 - Why edge compounds
One season. Two bettors. Same start. Different math.
Below is a simulated season. Each line is a bettor who places 150 bets, 1u each, at the same starting bankroll. The only difference is the math behind their picks.
Simulated bankrolls - 150 bets
Same starting bankroll. Same number of bets. Only difference: one side has edge, the other doesn't.
Illustrative simulation. Each bet wagers 1u at typical HR odds (~+700 for the positive-edge series, +600 for negative). Outcomes are deterministic - the same numbers render every time - but the takeaway is the same one variance can't hide: edge compounds, and so does its absence.
Variance is real. Both lines have ugly stretches. But one line trends up because the math is positive, and the other trends down because the math isn't. The longer you play, the louder the math gets - and the quieter variance becomes.
06 - How the model thinks
The theory is the easy part. The work is in the data.
Finding +10% edge isn't guesswork. It's 100+ features per matchup, three full seasons of backtesting, and a model that publishes every pick - wins and losses both.
Predictive Modeling
A machine-learning engine that turns millions of Statcast data points into confident, high-edge picks. Every prediction is built on real pitch-level data: bat speed, barrel rate, exit velocity, launch angle, and over 100 other features.
Mispricing Detection
We compare our model's HR probability against the sportsbook's implied odds and surface only the plays where the gap is wide enough to bet with confidence. No gut calls. Just math.
Full Transparency
ROI and profit are published live every day. Three full seasons of historical results are public. We don't hide losing months or cherry-pick stats - you see the full picture, including the days the model doesn't deliver.
Now you know how it works.
Want to see it working?Or just step in the box.
Three winning seasons of evidence are already public. Or, if the math's enough for you, pick a package and join us for season four.