How to Read Player Prop Edges Like a Pro

A prop with a +6% edge sounds like free money. It isn't — and treating it that way is the fastest way to give back any gains you've earned. Here's how a sharp eye reads an edge percentage in context.

What "edge" actually means

Edge is the gap between our model's win probability and the sportsbook's implied probability after stripping out the vig. If our model says a player goes over 1.5 hits 58% of the time and the implied probability of the offered price (-130) is 56.5%, the edge is +1.5%. That's the math; the math is easy. The hard part is everything around it.

Three things that destroy a "+6% edge" in practice

• **Sample size.** A model trained on 8 games of recent form can be wildly overconfident. Our engine penalises confidence by 30% when the sample is under 10 and 15% when it's between 10 and 20 — but the published edge number doesn't shrink. Always check the Confidence column alongside Edge.
• **Stale lines.** By the time you read a pick, the line may have moved. A +6% edge at -120 is a -2% edge at -150. Confirm the price at your book before placing.
• **Late-breaking news.** A pitcher's last-minute scratch, a starting lineup change 30 minutes pre-game, a weather delay — none of these always make the 15-minute model refresh window. Edges that don't reflect current information are not edges.

What a defensible "play this" looks like

A pick is worth taking action on when the edge is meaningful **and** the supporting context is intact:

• Confidence is in the upper third of the day's slate.
• The supporting splits have at least 50 plate appearances / 200 minutes / 100 shifts depending on sport.
• The line at your book is within 3 cents of the line we modeled against.
• No injury news in the last 30 minutes that we haven't reflected.

Treat picks that fail any of those checks as information, not action. Information is still valuable — sometimes the right move is to pass.

Bankroll math beats "trust the edge"

Even a +10% edge will lose 35-40% of the time. The only thing protecting you from a bad run is unit sizing. We default to quarter-Kelly with a 5% bankroll cap because full-Kelly variance will rip a normal-sized bankroll to shreds inside a month.

The sharpest bettors aren't right more often than the model. They're right enough that disciplined sizing turns small edges into a sustainable curve.

If you take one thing from this post: **edge percentage is a starting point, not a verdict.** Pair it with sample size, line confirmation, and Kelly sizing, and you're playing the game the way it's supposed to be played.