Introduction

This post analyzes the historical moneyline odds offered on popular professional and collegiate sports. Odds were gathered from the NBA, NFL and NHL, and college football and basketball. The broader question addressed is whether the moneyline odds offered by the sportsbooks are better or worse for the individual bettor than the actual results. My hypothesis is that the moneyline odds are worse for the bettor, meaning that the payout offered by the sportsbook does not correspond to to the actual results of the game.

As the saying goes Vegas wasn’t built on winners.

NOTICE: Nothing in this post or anywhere else on the site should be considered gambling, wagering or financial advice and is offered solely for entertainment purposes.

A moneyline bet is simple to understand. It is a wager on the home or visitor team to win the game outright. Odds are the way the payout is controlled. The odds vary greatly for a moneyline bet based on the predicted outcome and the matchup between the two teams. A big favorite will have a significantly lower payout when compared to an underdog. For a defined payout like $100, a bettor will have to wager more for a favorite and less for an underdog. This makes sense intuitively; a longshot should payout more than a sure thing.

Decimal odds represent the total payout a bettor receives (including the initial stake) for a 1-unit bet. For example, a game where the home team has 1.20 decimal odds means that a bettor can receive 0.20 units for every 1.0 units bet. In percentage terms, the bettor receives 0.20/1.0 or a 20% return on their money. A big underdog may have odds of 5.0 meaning that a 1-unit bet would profit by 4 unit meaning a return of 4.0/1.0 = 400% return.

Implied probability, on the other hand, is the percentage chance of the event occurring as “implied”” by the odds. You can solve for implied probability knowing the decimal odds by taking the inverse or 1 over the decimal odds. For example, decimal odds of 1.20 imply a probability of 1/1.20 = 0.833 or 83.33%. For the 5.0 odd underdog, the implied probability is 1/5.0=0.20 or 20%.

Decimal odds and implied probability can be used to factor into a bettor’s expected value (“EV”). The EV equation is generally:

EV = (Probability of Winning x Payout) − (Probability of Losing × Stake)

Where:

The EV equation can also be re-written to include decimal odds and implied probability:

EV = (Implied Probability × Decimal Odds) − 1

Implied probability derived from decimal odds is just the method of setting the EV equation equal to zero. A team that is a favorite with 1.20 odds and a 0.8333 implied probability should win the game 83.33% of the time. If the sportsbook paid out 0.20 units every time they won, then the original EV equation can be used to show that the EV is zero:

EV = (0.8333 X 0.20) - ((1 - 0.8333) X 1) = 0

Or you can use the modified EV to show that EV is zero:

EV = (0.8333 X 1.20) - 1 = 0

Bettors are tasked with finding Positive EV while the sportsbook only want to offer Negative EV bets. These can be classified as follows:

For example, home NBA teams that close at 1.20 decimal odds translate to an implied probability of 83.33% (1/1.20 = 0.8333). If all home NBA teams with decimal odds of 1.20 won at a rate of 83.33%, then the bet would be a Zero EV bet. In reality, NBA home moneyline favorites at 1.20 win actually win at a 77.7% rate. Making this bet repeatedly would lead to a Negative EV of -6.76%.

Moneyline odds are “better” if the offered implied probability is less than actual probability. In other words, teams win more often than what the odds imply. For the home NBA favorite at 1.20 with an expected implied probability of 83.33%, the actual win percentage would need to be greater than 83.33% to make it a positive EV (+EV) bet. Conversely, moneyline odds are “worse” if the offered implied probability is greater than actual probability.

odds grade measurement expected value
better actual > offered positive
same actual = offered zero
worse actual < offered negative

Returning to the original question, I wanted to analyse if the payouts set by the consensus odds were fair (ie had zero expected value) considering the actual outcome of the games. I used the consensus moneyline odds and compared that with the actual outcome of the games. Big favorites are those with decimal odds closer to 1. Big underdogs are those with decimal odds further away from 1.

Historical odds and game outcome data was obtained over the following periods:

To summarize the results set forth below:

Percentage of Games With Positive diff

home visitor both
NBA 0.284 0.326 0.305
NCAAB 0.25 0.15 0.2
NFL 0.3 0.5 0.4
NCAAF 0.42 0.48 0.45
NHL 0.23 0.36 0.29

Home Teams - Average Moneyline Odds with Positive and Negative diff

positive diff average negative diff
NBA 2.18 1,82 1,67
NCAAB 1.95 2.17 2.32
NFL 2.01 1.97 1.94
NCAAF 4.20 2.06 1.77
NHL 1.78 1.88 1.94

Visitor Teams - Average Moneyline Odds with Positive and Negative diff

positive diff average negative diff
NBA 3.30 2.94 2.79
NCAAB 4.02 4.90 5.30
NFL 2.57 2.77 2.93
NCAAF 4.38 3.87 3.47
NHL 2.18 2.25 2.29

NBA Moneyline Analysis

NBA moneyline odds were divided between home and visitor teams and then grouped by moneyline. For example, there were 125 games where home NBA teams had moneyline odds of 3.20. Overall, these home teams had an actual win percentage of 39.2% compared with an expected implied probability of 31.25% which is a diff of 7.90%. Betting all 125 games would have resulted in 0.25 unit advantage per game for a total gain of 31.8 units.

The graphs below show the diff for Home and Visitor NBA teams by moneyline odds. Points in green are where the diff was greater than zero, and points in red are where the diff was less than zero. Only odds where the moneyline was offered a minimum of 100 times is included. This eliminates some of the less common odds so that trends can be identified.

The top ten home NBA teams where the actual win percentage was greater than the expected implied probability and there were at least 100 games are shown in the table below. The diff column shows the difference between the actual home win percentage and the expected implied win probability as determined by the odds. Positive diff means that an edge exists at those odds over the historical dataset. For reasons discussed herein, this is not a “true” edge independent of other corroborating paramaters.

Top Ten Best Home NBA Moneylines - Actual Greater than Expected

ml_home count actual_win_home actual_loss_home actual_win_home_pct expected_ip diff ev pl
3.2000 125 49 76 0.3920 0.3125 0.0795 0.2544 31.8000
1.9524 155 88 67 0.5677 0.5122 0.0555 0.1084 16.8095
2.5500 166 74 92 0.4458 0.3922 0.0536 0.1367 22.7000
2.5000 178 80 98 0.4494 0.4000 0.0494 0.1236 22.0000
2.8000 161 65 96 0.4037 0.3571 0.0466 0.1304 21.0000
1.3571 192 149 43 0.7760 0.7368 0.0392 0.0532 10.2143
3.1000 128 46 82 0.3594 0.3226 0.0368 0.1141 14.6000
2.7500 153 60 93 0.3922 0.3636 0.0285 0.0784 12.0000
1.3636 117 89 28 0.7607 0.7333 0.0274 0.0373 4.3636
2.0500 302 154 148 0.5099 0.4878 0.0221 0.0454 13.7000

The top ten visitor NBA teams where the actual win percentage was greater than the expected implied probability and there were at least 100 games are shown in the table below.

Top Ten Best Visitor NBA Moneylines - Actual Greater than Expected

ml_vis count actual_win_vis actual_loss_vis actual_win_vis_pct expected_ip diff ev pl
2.5500 277 133 144 0.4801 0.3922 0.0880 0.2244 62.1500
2.4000 267 127 140 0.4757 0.4167 0.0590 0.1416 37.8000
1.4348 121 90 31 0.7438 0.6970 0.0468 0.0672 8.1304
1.8333 261 154 107 0.5900 0.5455 0.0446 0.0817 21.3333
5.0000 276 67 209 0.2428 0.2000 0.0428 0.2138 59.0000
2.3500 261 122 139 0.4674 0.4255 0.0419 0.0985 25.7000
3.9000 178 53 125 0.2978 0.2564 0.0413 0.1612 28.7000
2.1500 239 121 118 0.5063 0.4651 0.0412 0.0885 21.1500
6.0000 182 37 145 0.2033 0.1667 0.0366 0.2198 40.0000
2.5000 232 99 133 0.4267 0.4000 0.0267 0.0668 15.5000

Home NBA teams were less likely to have a positive diff as compared with visitor NBA teams. Visitor teams exceeded the moneyline probability 32.6% of the time as compared with home teams at 28.4%. Conversely, that means that home teams had the appropriate (or better) odds around 72% of the time and visitor teams 68% of the time. Bookmakers were correct around 70% of the time.

NBA - Summary of All Games by diff

diff home visitor both
greater than zero 5568 6404 11972
less than zero 14022 13236 27258
total 19590 19640 39230
percent above zero 0.284 0.326 0.305

It was interesting to compare the average moneyline odds for positive and negative diff and compare those to the average for all games. See the table below. Both home and visiting NBA teams tended to outperform (positive diff) the longer the odds were and underperform (negative diff) when the odds were less than the average. Home teams had average moneyline odds of 2.18 when diff was positive compared to 1.67 when diff was negative. Visitors had longer odds overall as compared to home teams, but the trend stayed the same when comparing average odds for positive and negative diff. Visiting teams had average odds of 3.3 when they outperformed and 2.79 when they under performed the odds.

NBA - Mean Moneyline by diff

diff home visitor
greater than zero 2.18 3.30
less than zero 1.67 2.79
all games 1.82 2.94

What is my interpretation of this information? From a betting standpoint, there is not a large edge that can be extracted fronm this information. Recall that only the odds are used to group the games and measure the expected outcome versus the implied outcome. No additional paramaters are considered. For example, team specific information like shooting percentage or defensivensive efficiency is not incorporated to forecast which team will win. It’s tempting to use the above rules like a model; if a visiting NBA team closes at 2.55 (or +155 odds) then bet that game since their is a +0.0880 diff. However, this fact alone should not dictate that an edge is present. These variations of diff - even with 100 occurrences - are not significant in and of themselves. Further, there was no pattern that would indicate that a certain group of odds indicated positive diff. For example, there were 269 times the NBA visitor closed at 2.60 odds, and these games had a negative diff of -0.0575. Likely the instances of positive diff I have identified are variance experienced by the book.

One last thing to note is we can make a rough estimate of the total profit and loss a bettor would have assuming the bettor could do two things:

Recall that the bettor wins the decimal odds minus 1 unit in profit if they win and loses 1-unit when they lose.

If the bettor could do both of those things above - bet 1 unit on both home and visitor - then they would be down 1,466 units (meaning that the book profited 1,466 units). This illuminates the conclusion that the odds are inherently unfair to the bettor absent an identifiable and distinct edge. See also What is the handle in sports betting? What is revenue? - The Athletic - Link - DB.

NBA - Total Profit & Loss

team profit_loss
home -711.36
visitor -754.78
total -1466.14

NCAAB Moneyline Analysis

NCAA Basketball (“NCAAB”) moneyline odds were analyzed to compare the implied probability with the actual probability.

Top Ten Best Home NCAAB Moneylines - Actual Greater than Expected

Home NCAAB teams where the actual win percentage was greater than the expected implied probability.

ml_home count actual_win_home actual_loss_home actual_win_home_pct expected_ip diff ev pl
1.3077 142 116 26 0.8169 0.7647 0.0522 0.0683 9.6923
4.7500 148 38 110 0.2568 0.2105 0.0462 0.2196 32.5000
1.5128 145 100 45 0.6897 0.6610 0.0286 0.0433 6.2821
1.4878 193 135 58 0.6995 0.6721 0.0274 0.0407 7.8537
1.6452 808 511 297 0.6324 0.6078 0.0246 0.0404 32.6774
1.2985 101 80 21 0.7921 0.7701 0.0220 0.0285 2.8806
1.2105 125 106 19 0.8480 0.8261 0.0219 0.0265 3.3158
1.2222 628 527 101 0.8392 0.8182 0.0210 0.0257 16.1111
1.3704 408 306 102 0.7500 0.7297 0.0203 0.0278 11.3333
2.5000 439 184 255 0.4191 0.4000 0.0191 0.0478 21.0000

Top Ten Best Visitor NCAAB Moneylines - Actual Greater than Expected

Visitor NCAAB teams where the actual win percentage was greater than the expected implied probability.

ml_vis count actual_win_vis actual_loss_vis actual_win_vis_pct expected_ip diff ev pl
1.6061 349 239 110 0.6848 0.6226 0.0622 0.0999 34.8485
1.2500 192 165 27 0.8594 0.8000 0.0594 0.0742 14.2500
3.9500 110 33 77 0.3000 0.2532 0.0468 0.1850 20.3500
1.6452 494 319 175 0.6457 0.6078 0.0379 0.0624 30.8065
1.3226 192 151 41 0.7865 0.7561 0.0304 0.0402 7.7097
3.6000 524 161 363 0.3073 0.2778 0.0295 0.1061 55.6000
1.5405 293 197 96 0.6724 0.6491 0.0232 0.0358 10.4865
1.3571 216 164 52 0.7593 0.7368 0.0224 0.0304 6.5714
2.6500 368 146 222 0.3967 0.3774 0.0194 0.0514 18.9000
4.0500 109 29 80 0.2661 0.2469 0.0191 0.0775 8.4500

NCAAB - Summary of All Games by diff

diff home visitor both
greater than zero 15451.00 9336.00 24787.0
less than zero 45934.00 51988.00 97922.0
total 61385.00 61324.00 122709.0
percent above zero 0.25 0.15 0.2

NCAAB - Mean Moneyline by diff

diff home visitor
greater than zero 1.95 4.02
less than zero 2.32 5.30
all games 2.17 4.94

NCAAB - Total Profit & Loss

team profit_loss
home -2162.04
visitor -7395.01
total -9557.06

NFL Moneyline Analysis

National Football League (“NFL”“) moneyline odds were analyzed to compare the implied probability with the actual probability.

Top Ten Best Home NFL Moneylines - Actual Greater than Expected

Home NFL teams where the actual win percentage was greater than the expected implied probability.

ml_home count actual_win_home actual_loss_home actual_win_home_pct expected_ip diff ev pl
1.4255 14 12 2 0.8571 0.7015 0.1557 0.2219 3.1064
1.3030 35 32 3 0.9143 0.7674 0.1468 0.1913 6.6970
1.1429 21 21 0 1.0000 0.8750 0.1250 0.1429 3.0000
5.2500 13 4 9 0.3077 0.1905 0.1172 0.6154 8.0000
3.6000 23 9 14 0.3913 0.2778 0.1135 0.4087 9.4000
1.2353 12 11 1 0.9167 0.8095 0.1071 0.1324 1.5882
2.6500 27 13 14 0.4815 0.3774 0.1041 0.2759 7.4500
1.3279 14 12 2 0.8571 0.7531 0.1041 0.1382 1.9344
2.3500 45 23 22 0.5111 0.4255 0.0856 0.2011 9.0500
2.6000 49 23 26 0.4694 0.3846 0.0848 0.2204 10.8000

Top Ten Best Visitor NFL Moneylines - Actual Greater than Expected

Visitor NFL teams where the actual win percentage was greater than the expected implied probability.

ml_vis count actual_win_vis actual_loss_vis actual_win_vis_pct expected_ip diff ev pl
3.1500 26 14 12 0.5385 0.3175 0.2210 0.6962 18.1000
4.2000 25 11 14 0.4400 0.2381 0.2019 0.8480 21.2000
2.4000 87 48 39 0.5517 0.4167 0.1351 0.3241 28.2000
1.3846 27 23 4 0.8519 0.7222 0.1296 0.1795 4.8462
3.4000 46 19 27 0.4130 0.2941 0.1189 0.4043 18.6000
1.4651 15 12 3 0.8000 0.6825 0.1175 0.1721 2.5814
2.4500 83 41 42 0.4940 0.4082 0.0858 0.2102 17.4500
1.8696 26 16 10 0.6154 0.5349 0.0805 0.1505 3.9130
1.3333 23 19 4 0.8261 0.7500 0.0761 0.1014 2.3333
1.3571 16 13 3 0.8125 0.7368 0.0757 0.1027 1.6429

NFL - Summary of All Games by diff

diff home visitor both
greater than zero 1467.0 2466.0 3933.0
less than zero 3344.0 2447.0 5791.0
total 4811.0 4913.0 9724.0
percent above zero 0.3 0.5 0.4

NFL - Mean Moneyline by diff

diff home visitor
greater than zero 2.01 2.57
less than zero 1.94 2.93
all games 1.97 2.77

NFL - Total Profit & Loss

team profit_loss
home -237.01
visitor -109.69
total -346.69

NCAAF Moneyline Analysis

National Collegiate Athletic Association Football (“NCAAF”) moneyline odds were analyzed to compare the implied probability with the actual probability.

Top Ten Best Home NCAAF Moneylines - Actual Greater than Expected

Home NCAAF teams where the actual win percentage was greater than the expected implied probability.

ml_home count actual_win_home actual_loss_home actual_win_home_pct expected_ip diff ev pl
4.80 23 13 10 0.5652 0.2083 0.3569 1.7130 39.40
3.15 29 17 12 0.5862 0.3175 0.2687 0.8466 24.55
5.05 11 5 6 0.4545 0.1980 0.2565 1.2955 14.25
4.40 32 15 17 0.4688 0.2273 0.2415 1.0625 34.00
8.00 40 14 26 0.3500 0.1250 0.2250 1.8000 72.00
5.10 15 6 9 0.4000 0.1961 0.2039 1.0400 15.60
17.00 16 4 12 0.2500 0.0588 0.1912 3.2500 52.00
10.50 11 3 8 0.2727 0.0952 0.1775 1.8636 20.50
6.40 15 5 10 0.3333 0.1562 0.1771 1.1333 17.00
9.50 25 7 18 0.2800 0.1053 0.1747 1.6600 41.50

Top Ten Best Visitor NCAAF Moneylines - Actual Greater than Expected

Visitor NCAAF teams where the actual win percentage was greater than the expected implied probability.

ml_vis count actual_win_vis actual_loss_vis actual_win_vis_pct expected_ip diff ev pl
9.5000 44 16 28 0.3636 0.1053 0.2584 2.4545 108.0000
7.7500 17 6 11 0.3529 0.1290 0.2239 1.7353 29.5000
1.3077 20 19 1 0.9500 0.7647 0.1853 0.2423 4.8462
1.2273 17 17 0 1.0000 0.8148 0.1852 0.2273 3.8636
5.6000 14 5 9 0.3571 0.1786 0.1786 1.0000 14.0000
1.2128 13 13 0 1.0000 0.8246 0.1754 0.2128 2.7660
4.6000 41 16 25 0.3902 0.2174 0.1729 0.7951 32.6000
5.4000 14 5 9 0.3571 0.1852 0.1720 0.9286 13.0000
4.0500 22 9 13 0.4091 0.2469 0.1622 0.6568 14.4500
4.8000 27 10 17 0.3704 0.2083 0.1620 0.7778 21.0000

NCAAF - Summary of All Games by diff

diff home visitor both
greater than zero 5506.00 6010.00 11516.00
less than zero 7602.00 6632.00 14234.00
total 13108.00 12642.00 25750.00
percent above zero 0.42 0.48 0.45

NCAAF - Mean Moneyline by diff

diff home visitor
greater than zero 4.20 4.38
less than zero 1.77 3.47
all games 2.86 3.87

NCAAF - Total Profit & Loss

team profit_loss
home 1086.30
visitor 301.03
total 1387.33

NHL Moneyline Analysis

National Hockey Leaque (“NHL”) moneyline odds were analyzed to compare the implied probability with the actual probability.

Top Ten Best Home NHL Moneylines - Actual Greater than Expected

Home NHL teams where the actual win percentage was greater than the expected implied probability.

ml_home count actual_win_home actual_loss_home actual_win_home_pct expected_ip diff ev pl
3.3000 13 8 5 0.6154 0.3030 0.3124 1.0308 13.4000
2.5100 22 14 8 0.6364 0.3984 0.2380 0.5973 13.1400
1.7246 45 35 10 0.7778 0.5798 0.1979 0.3414 15.3623
1.7042 12 9 3 0.7500 0.5868 0.1632 0.2782 3.3380
2.8500 14 7 7 0.5000 0.3509 0.1491 0.4250 5.9500
1.9615 11 7 4 0.6364 0.5098 0.1266 0.2483 2.7308
2.1300 42 25 17 0.5952 0.4695 0.1258 0.2679 11.2500
1.6579 22 16 6 0.7273 0.6032 0.1241 0.2057 4.5263
1.2857 16 14 2 0.8750 0.7778 0.0972 0.1250 2.0000
1.9174 18 11 7 0.6111 0.5215 0.0896 0.1718 3.0917

Top Ten Best Visitor NHL Moneylines - Actual Greater than Expected

Visitor NHL teams where the actual win percentage was greater than the expected implied probability.

ml_vis count actual_win_vis actual_loss_vis actual_win_vis_pct expected_ip diff ev pl
2.3700 19 13 6 0.6842 0.4219 0.2623 0.6216 11.8100
1.8197 19 14 5 0.7368 0.5495 0.1873 0.3408 6.4754
2.2700 31 19 12 0.6129 0.4405 0.1724 0.3913 12.1300
1.8929 26 18 8 0.6923 0.5283 0.1640 0.3104 8.0714
1.7874 32 22 10 0.6875 0.5595 0.1280 0.2288 7.3228
1.3922 13 11 2 0.8462 0.7183 0.1278 0.1780 2.3137
1.4762 30 24 6 0.8000 0.6774 0.1226 0.1810 5.4286
2.4800 62 32 30 0.5161 0.4032 0.1129 0.2800 17.3600
2.4400 55 28 27 0.5091 0.4098 0.0993 0.2422 13.3200
2.7600 37 17 20 0.4595 0.3623 0.0971 0.2681 9.9200

NHL - Summary of All Games by diff

diff home visitor both
greater than zero 2392.00 3764.00 6156.00
less than zero 8043.00 6671.00 14714.00
total 10435.00 10435.00 20870.00
percent above zero 0.23 0.36 0.29

NHL - Mean Moneyline by diff

diff home visitor
greater than zero 1.78 2.18
less than zero 1.94 2.29
all games 1.88 2.25

NHL - Total Profit & Loss

team profit_loss
home -485.50
visitor -310.56
total -796.06

Summary

This was an illuminating exercise to examine actual historical performance for professional and collegiate sports teams. With few exceptions, moneyline bets have negative EV and will likely continue that way. Without an independent and identifiable edge based on other parameters, betting teams to win will continue to be a loser. Based on the conclusions in this post, a good place to start to find positive EV would be:

Thanks for reading!