Advanced Statistical Terms Every NBA Prop Bettor Should Know

Why Your Props Keep Falling Short

You’re watching a point guard sprint to the rim, and the line says 7.5 assists. You bet it, you lose. The gap? You’re ignoring the numbers that actually drive those outcomes.

Regression to the Mean: The Silent Killer

Look: a player who’s been dishing out 12 assists for three games straight isn’t a statistical anomaly—he’s a fluke. Over a larger sample, his true assist rate will slide back toward his career average. If you chase the hot streak, you’re selling yourself short.

How to Spot It

Grab the player’s season‑to‑date assist per 48 minutes, compare it to the last five games. If the recent rate is 30% above the season rate, the regression coefficient tells you the next game will likely shave 0.3 assists off.

Expected Value (EV) – The Real Money‑Maker

Here’s the deal: every prop line has an implied probability. Multiply that by the payout, subtract the opposite probability, and you get EV. Positive EV? Bet it. Negative EV? Walk.

Quick EV Formula

EV = (Odds × Probability) – (1 – Probability). Plug in the odds from the bookmaker, the probability you derive from your statistical model, and you’ve got a single‑digit confidence gauge.

Standard Deviation (SD) – Measuring Volatility

And here is why you need SD: two players can have identical averages, but one’s game‑to‑game variance can be twice the other’s. A high SD means the line is a wider playground for profit.

Using SD in Props

Take a player with a 25‑point average and a 5‑point SD versus another with the same average but a 2‑point SD. Target the high‑SD player when the line is under, the low‑SD when it’s over. Simple math, big edge.

Correlation Coefficients – The Hidden Connections

Stop treating rebounds, assists, and points as isolated. They’re often linked. A point guard’s assist total correlates 0.45 with his three‑point attempts in the last 30 games. When the three‑point line is low, expect assists to drop.

Applying Correlation

Check recent team shooting percentages. If a team is shooting 48% from downtown, the guard’s assist line probably inflates. Adjust your prop down accordingly.

Weighted Moving Averages (WMA) – Prioritizing Recent Form

Forget simple averages. A WMA gives more weight to the last few games, smoothing out noise while still respecting the broader trend. It’s the sweet spot between raw streaks and season totals.

Set Up Your WMA

Assign 5% weight to games beyond ten, 15% to games six to ten, and 30% to the last five. Apply it to any prop metric, and you’ll see a clearer projection.

Logistic Regression – Predicting Binary Outcomes

If you’re betting over/under, you’re dealing with a binary outcome: a player either exceeds the line or he doesn’t. Logistic regression transforms your continuous stats (points, minutes, usage rate) into a probability of crossing that line.

Quick Logistic Cheat‑Sheet

Collect data: points per game, usage rate, opponent defensive rating. Run a regression (many free tools exist). The resulting coefficient set gives you a probability, which you then compare to the book’s implied odds.

Actionable Edge

Pull your player’s season assist rate, calculate regression, apply a weighted moving average, adjust for opponent shooting, and feed the numbers into a logistic model. If the resulting EV stays positive, place the bet. Now go.