Using Analytics to Inform Gambling Decisions

Yesterday, I did a giant preview of Texas Tech and the Big 12 using multiple analytical models. I mentioned that as a hobby, I track how these models perform relative to the oddsmakers in Las Vegas. Let me first say I don’t gamble. So when I write about picks each week, I’m not actually placing any money on it. I’m simply highlighting what the informed picks are according to the analytical models.


Money Line Betting

The main way to gamble using analytics is by comparing a team’s odds of winning a game to the money line in Vegas. The money line is the payout for a team to simply win the game, regardless of the margin of victory or total number of points scored.

The money line is written as a negative number for the betting favorite, such as -200. This means a bettor must wager $200 in order to profit $100. The moneyline is written as a plus number for the betting underdog, such as +200. This means a $100 bet would profit a bettor $200 if the underdog wins.

We can also use these numbers to tell us how often that particular bet must be successful in order to break even in the long run. To keep it simple, let’s first think of flipping a coin. It has a 50-50 chance of landing on heads each time you flip the coin. So if you are betting $1 to profit $1, and you flip the coin 1,000 times, you will break even by betting heads every single time. This is approximate, of course, since it would be rare that the coin would land on heads exactly 500 times out of 1,000. But give or take, you’ll be roughly even in the long run by winning 490 times or 510 times out of 1,000.

So let’s go back to the money line favorite and money line underdog examples above. A money line favorite of -200 must win the bet two out of three times (66.7%) for you to break even. If you bet $200 on three different -200 favorites, you are leaving $600 at the window when you place your bets.

The first two bets win, so you collect $100 profit for each, or $200 total profit. The third bet loses, so you lose the $200 you bet on it. You profted $200 on the first two bets and lost $200 on the third bet, for a total net of $0. Trust me, you could be doing much worse at this point.

With a +200 underdog, you only need one of the three bets to win for you to break even. You drop off $300 total (three different $100 bets) at the window. The first two lose, so you’re down $200. The third one wins, so you profit $200 on that bet. Overall, you’re back to $0.

-200 and +200 are fairly easy numbers to work with. But how often does a -470 favorite need to win for you to break even in the long run? How often does a +530 underdog need to win for you to break even in the long run? That’s what the spreadsheets are for.

I look at the lines in Vegas, using’s “Vegas consensus” (an average of all the major sports books) and calculate the break even percentage for both the favorite and the underdog in every single college football matchup on any given Saturday.

I compare the break even percentage to the percentages provided by the various models, and look for discrepancies of more than 7 percent. I can use an example from this Saturday’s game between Arizona and Hawaii to show how I do this (and so I don’t have to copy and paste giant spreadsheets into these articles every week).

According to FPI, Arizona has a 61.6 percent chance to beat Hawaii on Saturday. Naturally, this gives Hawaii a 38.4 percent chance to win the game. Arizona is a -400 money line favorite and Hawaii is a +300 money line underdog.

Using some pretty easy math, one can calculate that Arizona needs to win this game 80 percent of the time in the long run for a -400 bet to break even. Hawaii needs to win this game 25 percent of the time for this bet to pay off in the long run at +300 odds.

But as noted, Arizona barely has a 60 percent chance of winning. That’s a terrible bet at -400. However, Hawaii’s odds of winning the game are 13.4 percent better than the break even percentage on their money line. That’s great value, and significantly higher than my 7 percent trigger.

Adam McClintock’s model mostly agrees with FPI. Hawaii’s odds of winning according to McClintock are 16.6 percent higher than their break even percentage.


Against the Spread (ATS) Betting

Another way to bet on college football is against the spread. In this form of betting, one is betting not on a team to win the game, but on a particular margin of victory. Sticking with Arizona vs. Hawaii, Vegas has Arizona as an 11-point favorite. So instead of simply betting on Arizona or Hawaii to win, a bettor must bet on Arizona to “cover the spread”, which means they need to win by more than 11 points. Or they bet on Hawaii to cover the spread by losing by fewer than 11 points (or winning the game).

McClintock’s model is the only one I can find that has predictions for a margin of victory in addition to the odds given to each team to win the game. Since his model is behind a paywall, I want to respect his data by not sharing it publicly, but he indicates that Arizona is not as big of a favorite as Vegas seems to indicate. So if betting against the spread, Hawaii is the pick yet again.

I would be glad to share my calculations with any of you in full if you’re interested. Just let me know if you want access to it or if you want additional info. Otherwise, I’ll just post the picks each week and spare everyone the calculations and spreadsheets. I’ll keep a tally of how I do each week and a season totals to date.

Almost all ATS picks payout -110, so you bet $110 to profit $100. This means you need to do better than winning half your bets to break even in the long run.


Parlay Betting

One more thing. After what I’m calling “Week Zero” (the two games on Saturday), there will obviously be a full slate of football every Saturday. Typically there are at least a handful of high value picks I identify each Saturday. In addition to betting them individually, I also parlay each pair of two games for added fun. This can either pay off big time or lose you a lot of money in a hurry.

To envision this, let’s say I find five outcomes worth betting on next Saturday: Outcomes A, B, C, D, and E. In addition to betting on each individual outcome, I would also bet one unit on Outcome A and B both happening, Outcome A and C both happening, Outcome A and D both happening, etc. So this would look something like betting on Texas to beat Oklahoma AND Texas Tech to beat TCU on the same Saturday, all rolled up into one bet. If either outcome doesn’t occur, I lose the entire bet. But if both occur, I win a lot more than if I had bet on the two outcomes individually.


Week 0 Gambling Picks

Hawaii +300, 2u*

Hawaii +11


  • Last Week Against the Spread (ATS): N/A
  • Last Week Money Line (ML): N/A
  • YTD ATS: N/A
  • YTD ML: N/A


*This means two betting units. Most people use $100 as a standard betting unit, but others are more comfortable betting smaller amounts. Since two different models agree on Hawaii +300 being a high value pick, I am betting two units on this one. If no unit amounts are listed next to the bet, it’s just one unit.

Sorry for the long post, but I wanted to provide a thorough primer before the season got here. Let me know if you have any questions or if I can clarify anything.

To Top