So far, all I have updated is the NFL. I found a free source of data going back to 2006 here. That said, here is what I have found:
Bets Against Spread
Bet | Win | Loss | Push | Win rate | Std dev | Num std dev | Return |
---|---|---|---|---|---|---|---|
All home | 1540 | 1602 | 78 | 49.01% | 1.26% | -0.78 | -6.27% |
All away | 1602 | 1540 | 78 | 50.99% | 1.26% | 0.78 | -2.60% |
All favorite | 1499 | 1584 | 78 | 48.62% | 1.27% | -1.08 | -7.00% |
All underdog | 1584 | 1499 | 78 | 51.38% | 1.27% | 1.08 | -1.87% |
Home underdog | 512 | 495 | 25 | 50.84% | 2.23% | 0.38 | -2.86% |
Home favorite | 1004 | 1072 | 53 | 48.36% | 1.55% | -1.06 | -7.48% |
Away underdog | 1072 | 1004 | 53 | 51.64% | 1.55% | 1.06 | -1.38% |
Away favorite | 495 | 512 | 25 | 49.16% | 2.23% | -0.38 | -6.01% |
Home pick | 24 | 35 | 0 | 40.68% | 9.21% | -1.01 | -22.34% |
Away pick | 35 | 24 | 0 | 59.32% | 9.21% | 1.01 | 13.25% |
All | 3142 | 3142 | 156 | 50.00% | 0.89% | 0.00 | -4.44% |
Not surprisingly, underdogs do better than favorites. What is a little surprising is away teams to better than home teams. I also indicate a standard deviation of the win rate. The only categories that are one standard deviation up, assuming every bet had a 50% chance of winning, is all underdogs and away underdogs.
Over/Under
Here is a summary of over/under bets:
Bet | Win | Loss | Push | Win rate | Std dev | Num std dev | Return |
---|---|---|---|---|---|---|---|
Over | 1586 | 1576 | 58 | 50.16% | 1.26% | 0.13 | -4.17% |
Under | 1576 | 1586 | 58 | 49.84% | 1.26% | -0.13 | -4.76% |
I am surprised that over bets very slightly outperformed under bets. I had expected unders to do a little better. Both are very close to the 50% win rate expectation.
Money Lines
With money line bets, not surprisingly underdogs did better than favorites. The following table shows the rate of return on money line bets.
Bet | Return |
---|---|
Underdog | -3.91% |
Favorite | -6.24% |
All | -5.08% |
Next is an updated look at my Probability of underdog winning by spread. The "Prob. win" column is the actual probability an underdog with this spread won. The "Estimated" column attempts to tease out the variance by smoothing out the ups and downs.
spread | spread | games | wins | Prob. Win | Estimated | Fair |
---|---|---|---|---|---|---|
0 | 0 | 154 | 77 | 50.0% | 50.0% | 100 |
1 | 1 | 139 | 69 | 49.6% | 46.4% | 115 |
1.5 | 1.5 | 88 | 43 | 48.9% | 44.6% | 124 |
2 | 2 | 126 | 51 | 40.5% | 42.9% | 133 |
2.5 | 2.5 | 224 | 98 | 43.8% | 41.1% | 143 |
3 | 3 | 517 | 235 | 45.5% | 39.4% | 154 |
3.5 | 3.5 | 279 | 107 | 38.4% | 37.7% | 165 |
4 | 4 | 157 | 59 | 37.6% | 36.1% | 177 |
4.5 | 4.5 | 128 | 47 | 36.7% | 34.4% | 191 |
5 | 5 | 89 | 23 | 25.8% | 32.8% | 205 |
5.5 | 5.5 | 118 | 38 | 32.2% | 31.3% | 220 |
6 | 6 | 133 | 45 | 33.8% | 29.7% | 236 |
6.5 | 6.5 | 147 | 42 | 28.6% | 28.3% | 254 |
7 | 7 | 220 | 51 | 23.2% | 26.8% | 273 |
7.5 | 7.5 | 146 | 36 | 24.7% | 25.5% | 293 |
8 | 8 | 65 | 14 | 21.5% | 24.1% | 315 |
8.5 | 8.5 | 52 | 16 | 30.8% | 22.8% | 338 |
9 | 9 | 58 | 13 | 22.4% | 21.6% | 363 |
9.5 | 9.5 | 48 | 6 | 12.5% | 20.4% | 390 |
10 | 10 | 104 | 19 | 18.3% | 19.3% | 419 |
10.5 | 10.5 | 61 | 15 | 24.6% | 18.2% | 450 |
11 | 11 | 41 | 4 | 9.8% | 17.1% | 483 |
11.5 | 11.5 | 22 | 4 | 18.2% | 16.1% | 519 |
12 | 12 | 13 | 3 | 23.1% | 15.2% | 558 |
12.5 | 12.5 | 25 | 4 | 16.0% | 14.3% | 599 |
13 | 13 | 34 | 6 | 17.6% | 13.4% | 644 |
13.5 | 13.5 | 36 | 4 | 11.1% | 12.6% | 692 |
14 | 14 | 36 | 3 | 8.3% | 11.9% | 743 |
14.5 | 14.5 | 12 | 1 | 8.3% | 11.1% | 798 |
This differs quite a bit from my previous look at this, as seen from the link above. The old data was 1994 to 2013. Again, the new data is 2006 to 2017. For example, I previously said a 7-point underdog had a fair money line at +253. Now I say it is +273. No wonder I haven't been doing well on underdog money line bets (this is not a joke).
Any questions on the NFL. I also want to look at the NBA, NHL, MLB, and NCAAF. I need data for these sports, if anyone has anything up to date.
Sport Name. Alternative Names.
Season start.
Season end.
Differences between conferences, leagues, teams, etc.
General reputation for honesty in the sport.
General reputation for honesty in the officiating.
Sites for further information.
Sites for last minute information.
Quote: FleaStiffFor the revised edition can we have a general introduction:
Sport Name. Alternative Names.
Season start.
Season end.
Differences between conferences, leagues, teams, etc.
General reputation for honesty in the sport.
General reputation for honesty in the officiating.
Sites for further information.
Sites for last minute information.
Why don't you just have him settle the lawsuits now?
Keep in mind that, technically, the book can't use the trademarks NFL, NBA, Major League Baseball, NHL, or NCAA, or any of the professional team or athletic conference names without permission, and while most of the time, the leagues don't really care (the only two phrases I have seen set off alarm bells are "Olympics" and "Super Bowl"), they can play the trademark card if necessary for some other purpose.
A few questions you might want to consider for any chapters sports:
-probability for each margin of victory / push rates / value of half-point on or off each number
-easy to find correlated parlays. Baseball home-and-under, road-and-over, etc.
-How the leagues changes over time can effect databases. Larger sample is obviously better, but having to go back over 10 years means looking at a slightly different game. NFL scoring jumped slightly right after 2006, which I think is the reason for more overs during this specific 12 year stretch
Quote: TomG-probability for each margin of victory / push rates / value of half-point on or off each number
Good suggestion. Previous analysis has shown paying 10 basis point for an extra half point is only worth it off of 3 and 7. Usually, they don't let you buy it off 3.
Quote:-easy to find correlated parlays. Baseball home-and-under, road-and-over, etc.
Yes, I love those baseball correlated parlays. Good suggestion.
Quote:-How the leagues changes over time can effect databases. Larger sample is obviously better, but having to go back over 10 years means looking at a slightly different game. NFL scoring jumped slightly right after 2006, which I think is the reason for more overs during this specific 12 year stretch
A very good question is how far back should you go, assuming time was no issue. As you said, more data is always better, but what if it is stale data? The NFL is not the same as it was before rules were put in place to put a ball and chain on defenses, in an effort to reduce head trauma. Field goal kickers are also much better than they were ten years ago.
The basic info on NFL sports betting that I would like to see is percentage of games decided by 1 point, 2 points, 3points, etc. In other words, as the line shifts from +3.0 points to +3.5 points how much of a percentage advantage is that?
The information that you provide in the tables leads me to conclude that almost none of those win percentages is any different than 50%, and the variations you discern are not distinguishable from statistical noise.
When I sit around with buddies, these are the type of comments that I hear about picking NFL games:
Denver is known to have a larger home field advantage than most teams, as is Miami during the first month of the season (because of the heat).
Home field advantage is arguably less important in the first week of the season when your starters are not recovering from a physically punishing game.
High game temperatures have recently been shown to be slightly correlated with winning UNDER bets, presumably because the heat makes the offensive linemen wilt as the game goes along.
Traveling across multiple time zones leads to a larger disadvantage for the traveling team, particularly for
- East coast teams traveling to the west coast for afternoon or evening games
- West coast teams traveling to the east coast for 1:00 pm games.
Will later chapters in your book address issues like these?
Quote: WizardGood suggestion. Previous analysis has shown paying 10 basis point for an extra half point is only worth it off of 3 and 7. Usually, they don't let you buy it off 3.
Can buy on or off the 3 for 20-cents some places. Seems like that must be a good bet if either the -2.5, +3.5 or +-3 are also good. Most of Las Vegas is 30-cents for the places that allow it.
Any idea what the other numbers are worth?
Home field advantage: Can always take home team winning percentage (58% in NFL, I think) and convert it into a no-vig moneyline, somewhere around -135 to -140. That would be a 2.5 spread (according to wizard of odds). So home field in the NFL *should* be worth 2.5 points based on that. Most people usually say it's worth 3. That would explain why road teams have done slightly better overall since 2006.
But why do home teams have an advantage? Travel and time-zone changes if often cited as one reason. That would mean out-of-division and out-of-conference games would see a bigger home field advantage than games within a division or conference. Is this true? To what extent.
Can also spend some words talking about prop bets. I often like to use the sportsbooks themselves to handicap bets. Always do great on the baseball first inning run. Wait until close to game time, sometimes the total will move quite a bit from the morning, but the score in the first inning prop won't. There should be a good bet to be made. Always a lot of great opportunities for that in the Super Bowl.
Quote: TomGCan buy on or off the 3 for 20-cents some places. Seems like that must be a good bet if either the -2.5, +3.5 or +-3 are also good. Most of Las Vegas is 30-cents for the places that allow it.
Any idea what the other numbers are worth?
Here are the results of a quite a bit of number crunching -- the fair price to pay for a half point off of certain key spreads:
Spread | Fair price |
---|---|
3 | 128.1 |
4 | 116.9 |
7 | 121.9 |
10 | 117.9 |
14 | 117.0 |
For example, to buy the half point off of three (meaning -3.5, -3, +2.5, or +3), a fair price to lay would be 128.1 to win 100, to get you the same return as a random picker without the extra half point, who gets back 21/22 = 95.45%.
The bottom line is that it is only worth it to lay the 120 off of 3 and 7. Definitely do it off of 3 and there it is slightly worth it off of 7.
Quote:But why do home teams have an advantage?
If there are more than a few posts about this, I'll split if off. The book Scorecasting spends many pages on this question. The bottom line is money -- sports makes more money if the HOME team wins. People are willing to pay more for tickets for a winning team. While I'm sure every referee would vehemently deny any bias, they present a very good case the referees call many more penalties against visiting teams and shrink the strike zone for them in baseball, and expand it for the home team. They convinced me.
Quote: WizardHere are the results of a quite a bit of number crunching -- the fair price to pay for a half point off of certain key spreads:
Spread Fair price 3 128.1 4 116.9 7 121.9 10 117.9 14 117.0
For example, to buy the half point off of three (meaning -3.5, -3, +2.5, or +3), a fair price to lay would be 128.1 to win 100, to get you the same return as a random picker without the extra half point, who gets back 21/22 = 95.45%.
The bottom line is that it is only worth it to lay the 120 off of 3 and 7. Definitely do it off of 3 and there it is slightly worth it off of 7.
If there are more than a few posts about this, I'll split if off. The book Scorecasting spends many pages on this question. The bottom line is money -- sports makes more money if the HOME team wins. People are willing to pay more for tickets for a winning team. While I'm sure every referee would vehemently deny any bias, they present a very good case the referees call many more penalties against visiting teams and shrink the strike zone for them in baseball, and expand it for the home team. They convinced me.
That’s a cool chart! I’d love to see a database that has the opening and closing lines so you could analyze what line drift is worth, and test the old adage to bet against the trend if the line drifts by two points.
Quote: Wizard
Not surprisingly, underdogs do better than favorites. What is a little surprising is away teams to better than home teams. I also indicate a standard deviation of the win rate. The only categories that are one standard deviation up, assuming every bet had a 50% chance of winning, is all underdogs and away underdogs.
This one is simple, the public (or books) simply over favor home teams by too much. While it jumps out significantly in the PK category (limited sample size) Away performs better than Home in every other category, as well.