VantoraBets
Methodology

How VantoraBets works

A plain-English look at where the numbers come from (NFL and college football), how picks are chosen, and how results are graded.

1. Estimating each game

The model starts with past game results and builds a power rating for every team that updates after each game (an Elo-style rating). It also looks at rest days and each team's recent scoring margin. From these it estimates the probability that each team wins.

2. Comparing with the market

Betting odds already contain a lot of information. We convert odds into the market's implied probability and remove the sportsbook's built-in margin. The model's estimate is then blended with the market, with the blend weights fitted only on past seasons. This keeps the model from claiming edges it has not earned.

3. Estimated value

Estimated value (EV) compares our probability with the price. If a team pays +105 and we think it wins 56% of the time, a 1-unit bet is worth about +14.8% on average, if our probability is right.

Example only

Team A vs. Team B · Moneyline

56%

Model estimate

Best available odds: +105
Estimated value: +14.8%

Illustrative numbers, not a real prediction. Actual value depends on the probability model and the odds available.

The math: 56% × 2.05 (the payout multiple at +105) − 1 = +14.8%. If the probability is wrong, the value is wrong too.

4. When a pick is made

A game becomes a pick only if all of these are true:

  • Estimated value is at least +3% for the NFL (+4% for college football).
  • The model is at least 2 percentage points above the market (2.5 for college football).
  • The market doesn't already see the game as nearly decided: the team's market probability is between 10% and 90%.
  • The price is no longer than +250 (we skip longshots, where edges are least reliable).
  • The game has not started.
  • Fewer than 5 picks (NFL) or 8 picks (college football) have been made that week.

If nothing qualifies, there is no pick. We don't force one.

5. Saving and grading

Each pick is written to a permanent log with a timestamp before the game starts. It is never edited or deleted. After the game, it is graded as a win, loss or push at flat 1 unit. A tie is a push.

6. Closing-line value

Closing-line value checks whether the price we saved was better than the market's last price before kickoff. It is a faster signal than win rate, which can take hundreds of games to mean anything.

College football

College football uses the same method with a few changes. There are far more teams and bigger talent gaps, so the model uses stricter limits and only looks at games between top-level (FBS) teams. Teams change a lot every year through the transfer portal and graduation, so ratings are pulled back toward average more at the start of each season. Early-season college picks are the least reliable.

What the model does not know

  • Injuries, inactive players and late lineup changes are not in the model yet.
  • Weather, coaching changes and trades are not in the model.
  • Odds move. By the time you look, the price may be different.

Honest limits

Sportsbooks are hard to beat, and most bettors lose over time. This model may be no better than the market. That's why we publish every result, test on past seasons, and show losses next to wins. Treat this as analysis and entertainment, not advice.

Must be 21 or older. If gambling is a problem, call or text 1-800-GAMBLER (1-800-522-4700).