What's With the Home Underdogs in the N.F.L.?

In recent years, the federal government has taken various actions to make it harder to bet on sports over the internet.

That’s lucky for me, because when I used to bet on football, one of the key pieces of information I used was whether or not a team was a home underdog. For whatever reason, bettors don’t like to bet on home underdogs.

Recognizing this, bookies shade the point spreads toward the home underdogs. Because of the bettor bias, the bookies can set spreads so that home underdogs cover the spread more than half the time, but attract less than half the betting dollars. According to my estimates, home underdogs covered the spread almost 54 percent of the time over a 20-year period, but they still only attracted about 40 percent of the betting dollars in those games. As a consequence, these games are extremely profitable for the bookies: most of the money goes on the visiting favorite, who manages to cover the spread only 46 percent of the time. (For more on this, see an academic article I wrote on the subject.)

Even after I published that paper, the home-underdog bias remained alive and well. I nearly won an 800-person handicapping contest doing little more than picking home underdogs.

The last two years, however, have not been so kind to home underdogs. According to calculations made by Charles Monteleone, a graduate student at the Booth School of Business here at the University Chicago, home underdogs had a 44-45-1 record against the spread in 2007. That is not so bad, but it’s far from the historic performance. The real outlier was the 2008 season, during which home underdogs had a record against the spread of only 32-45-2!

Why this radical reversal? One possibility is that bettors finally caught on to the fact that home underdogs were a bargain, which in turn led the bookies to alter spreads so that they no longer favor the home underdogs. I think there is almost no chance that this is the correct explanation. Bettors have failed to recognize this bias for 20 years; there is no way gambling patterns could suddenly change this quickly.

The real explanation, I suspect, is that it was just bad luck — bad luck for the minority of bettors who play the home underdogs, but especially bad luck for the bookies. I wouldn’t be surprised if the poor performance of home underdogs destroyed a large chunk of the total profits for the bookmakers on the N.F.L. this year.

Home underdogs are of particular interest right now because, remarkably, in all four playoff games this weekend, the home team is the underdog. If I were a betting man (or more accurately, if I had an account I could bet on), I would be hammering the home underdogs this weekend.


Why this radical reversal? One possibility is that bettors finally caught on to the fact that home underdogs were a bargain, which in turn led the bookies to alter spreads so that they no longer favor the home underdogs. I think there is almost no chance that this is the correct explanation. Bettors have failed to recognize this bias for 20 years; there is no way gambling patterns could suddenly change this quickly.

Going to disagree - I'm not a bettor myself (though I join pools), but a couple of years ago there were numerous columns that mentioned this home underdog point. I believe Bill Simmons of ESPN - one of the most widely read sportswriters - mentioned this very point in a couple of NFL prediction pieces. This could have spurred a shift in awareness that in turn was matched by the lines. You'd also have to factor in the far more easily attainable information on such patterns now readily available on the internet, which increases each year, particularly in the last couple of years.



In that article Simmons also suggested that the new stadiums being built to cater to club seats and luxury boxes didn't provide the same advantage home teams once enjoyed.

Michael "mickeyrad" Radosevich

Great post - & your "academic article" is a fine use of analysis in a practical forum. I advise everyone to click on the link.
While the home underdog phenomena is not a secret, it is, as this post points out, underutilized. I'm made money for years betting on the home 'dogs, & done well in pools for the same reason. Of course it helps to be a little selective on choosing which home 'dog to pick - e.g., the Detroit Lions were an awful team & a bad pick the whole year - but on the whole, the home 'dog is a good "horse to ride" mixed metaphor intended :)


Betting patterns haven't changed significantly, if at all. There are websites that show breakdowns of online wagers. People still love taking road favorites and as the post suggests, its been a tough year for the sportsbooks.

While he has a large audience, Bill Simmons has contributed nothing useful to gambling theory (theories based on anecdotal evidence are not useful) and his handicapping skills would hardly inspire a sportsbook to start shifting its lines.


Link to the Simmons article. Doesn't explain why so much changed between 2006 and 2008 as I don't think that many new stadiums had opened.


Bill Gattis

I do not understand how Mr. Levitt lacks a bank account with which to bet. 20 years of knowledge regarding the undervaluation of home underdogs surely should have reaped him a fortune. Am I missing something? Has he given it all to the Gates Foundation?


It's because the divisional winners are so terrible when stacked next to the wild cards. Minnesota (my home team) barely beat the hapless 0-16 Detroit Lions (both times). The Cardinals won the weakest division in the NFL. In the AFC, an 11-5 Patriots couldn't make the playoffs, but an 8-8 San Diego Chargers did.

Just because the NFL uses the seeding method it does, that doesn't change a mediocre team into a great one.

caveat bettor

A large factor is the current NFL playoff format. Assuming large variances in different divisions (i.e. some divisions are much stronger than others), underdogs should be favored. Byes are awarded to the top 2 conference winners, which would likely play in the stronger divisions. Home field advantage in the wildcard round is awarded to the weaker division winners, who then host wildcard teams from those stronger divisions.

How many teams this year would want to host the 11-5 New England Patriots this weekend? They didn't even make the playoffs! I doubt that either the 8-8 San Diego Chargers nor the 9-7 Arizona Cardinals would want to host them any more than the 12-4 Indianapolis Colts or the 11-5 Atlanta Falcons whom they are facing.

John M.

The academic article that Levitt published and cited is absurd. The data set is from a promotional contest in which all of the entry fees went back in prizes. It was essentially just a big office pool. That is not bookmaking at all. The bookmaker could have assigned spreads randomly without any cost.

Brian R. Murphy

@Bill Gattis -- It is very hard to repatriate sports gambling winnings these days.

I work at the Board of Trade and I'll tell you that the guys who used to run books in the 80's and early 90's are back making a very sizable market. Make your way down to LaSalle St. Mr. Levitt.


"for whatever reason, bettors don't like to bet on home underdogs"- well, why the distinction?- do bettors like to bet on away underdogs?!- I doubt it- you may coldly and rationally decide that the spread against the lions is too large, but as mickeyrad alludes to, actually laying a bet on the detroit lions is going to well up a bitter taste in most mouths, home or away


A lot of what we see is just luck or an oddity. People seem to think that the natural consequence is always symmetry. But symmetry is only possibility, and the odds of symmetry always happening are very low.

Like you showed with Kansas City that year, if a sports team goes fifteen games without a win, it's probably just a below average .300 typical sports team hitting a bad luck spell.


John M.-

Great point. The setup described in the article does not fit the traditional pattern of how and why lines are set and the line-setter's role and function in the process. What is even more ironic is that it is described in detail earlier in the article and then completely disregarded.

Also, let's look at sample-size. Levitt sites 20 years of data demonstrating this phenomenon and than points to two years in which it does not hold true as evidence that the pattern does not hold. Who's say to this isn't just a minor blip in the data based on a small sample-size?

Jerry Tsai

While Levitt's observation that home underdogs are underbet is interesting and perhaps useful in sports betting, his contention that, absent other information, all four NFL games this weekend feature home underdogs and therefore represent good betting opportunities is flawed.

Let us grant him the long-term finding that bettors inefficiently select visiting favorites-- does it apply in these four games? Aaron Schatz's useful DVOA statistics found at Football Outsiders illustrates that the home teams deserves their underdog status. Let's break down some simple stats, however, that illustrate that these home teams have EARNED their 'dog status.

ATL (-2) at ARI:
ATL: 11-5; ARI: 9-7.

ATL won two more games than ARI.

ATL played in the tough NFL South, where no team finished with a losing record. The teams in that division won 70% of their non-divisional games.

ARI played in the absurdly weak NFC West, where the other three teams won just one quarter of their non-divisional games. To give ARI some credit, however, they do play better at home.

IND (-1) AT SD:
IND: 12-4; SD: 8-8.

IND played in the strong AFC South, which won 65% of their non-divisional games. They won 4 more games than SD, which played in the laughable AFC West, in which none of the teams finished with a winning record. The AFC West collectively won 28% of their non-divisional games.

BAL (-3.5) at MIA:
BAL: 11-5; MIA: 11-5.

BAL played in the middling AFC North, which won about 50% of their non-divisional games. However, the divisions that they played this year (outside of their own) were the AFC South and the NFC East, two difficult divisions this year. As stated before, the AFC South won 65% of their non-divisional games, and the NFC East too won 65% of their non-divisional games. The NFC East also had no team in their division finish with a losing record for the second consecutive year. This is a tough division to play.

MIA and the rest of the AFC East, however, played the two weakest divisions this year: the AFC West and the NFC West. So while they finished with identical records, MIA and BAL had extremely different paths. It is obvious that BAL is the stronger of the two teams.

PHI (-3.0) at MIN:
PHI: 9-6-1; MIN: 10-6.

PHI is probably the least deserving favorite. Not only did MIN win more games, but they are playing at home. MIN, however, played in the NFC North, which, led by Detroit, won less than one-third of their non-divisional games. As stated earlier, PHI's division, the NFC East, won almost two-thirds of their non-divisional games. Nevertheless, I suspect the line on Philadelphia is more due to bettor enthusiasm for the team rather than objective analysis.

Cogent football analysis: saframetrx.com.



How can so many people have written so many comments about this year's NFL playoffs and NOT have called out the oddity that five of the six playoff teams in the NFC have an "a" at the end of their city/state names?!? (Atlanta, Arizona, Carolina, Minnesota, and Philadelphia, leaving only New York on the outside, looking in). How crazy is that?! You may be saying, "So what?" but consider this: of the 16 teams in the NFC, only 5 of them are so endowed! (Green Bay and Tampa Bay almost make it in, on a technicality, but no!) In other words, this year we MAXIMIZED the number of "a-ending" teams in the postseason, despite a set of odds being heavilty stacked against them!

Who'd have bet on THAT happening, I wonder?



Another important note regarding the setting of lines:

Americans (and maybe everyone) loves betting on the favorite. This is known by oddsmakers and, therefore, spreads are usually bumped up .5 to 1 points when they are opened because they know that most casual and/or uninformed betters will take the favorite.

By the way, even if you were to hit 54% of the time by picking home dogs, you'd still break even. Bookies and Vegas generally take 10% of all bets, so you usually need to hit at least 55% of bets to break even.


Good points but what many are missing is that bookies and sports books don't care so much who wins and don't pay attention so much to home dogs, etc. The business model is to set a line that will maximize betting on both sides of the game so they can collect as much vigorish (vig- their take) as possible. Having the advantage of living in Vegas and watching early lines move as much as two or three points before the game is played showed us that there was too much money going to one side and they had to push the other. It's important however to get as close as possible to the right line in the beginning to prevent middling. Middling is when a line moves so much due to uneven betting patterns that some savvy bettors in the past could get both sides of the bet (one team -3 and the other +3 for example) and if it fell in the middle as often happened with close games, we could collect both sides. So dozens of experts get together as soon as a game ends and the next opponent is identified and post a line that they believe will generate enough action on both sides. A game that is bet heavily on one side or the other is just plain bad business but sometimes unavoidable. A bad line can kill a sportsbook, so much care is put into the opening line. Once the line is set by these experts, the rest follow suit and they live or die with the result.



Jerry Tsai, Good analysis but as far as the Vikings weak division it might be more accurate to throw out an outlier like the Lions. It's still not an intimidating picture, but perhaps clearer.

John M.

BSK: The break even point is actually 52.4% (11/21) if the vig is 10%.

Something that is missing from all of this discussion is that the spread on any particular game is not especially important to a large book. There are not thirteen to sixteen spreads per week. There are many spreads on many sports every day, plus parlays and a variety of other sucker bets that have much worse payoffs than straight bets. Even if the "market" consisted entirely of straight bets, making separate estimates of the many spreads means that "errors", that is, spreads that are too high or too low, average out. So long as the book can keep the average number of bets for all the favorites and dogs near 50% over time, it will make money.


So if you exclude the 'atrocious' teams - those with minimal chance to win at all - Detroit, St. Louis, Kansas City - where do the home underdog stats lie? My guess is a little better.

Even though the bet is covering the spread and not winning, there have been teams in the last two years so bad that they should not even compete. Realistically, the bookies should be good enough to make the spread large enough so that Detroit could at least cover (win? Hah!) but perhaps no one could properly gauge this level of futility.