Home underdogs in the NFL

A few years back I wrote an academic paper that set out to resolve a paradox in sports betting: how could it be the case that bookies systematically got the spread wrong in NFL football? In particular, home underdogs win far more games than they should against the spread. Despite the fact that bookies take a healthy cut of every dollar bet (the “vig”), a bettor who bet on every home underdog for the last two decades would show positive profits. The answer, it turns out, is that bettors have a strong bias against home underdogs. Even though the bookies set spreads that work to the advantage of home underdogs, a disproportionate amount of the money is still bet on visiting favorites. By taking advantage of the bettors’ misperception, bookies greatly increase their profits by setting the “wrong” spread.

This year, I’ve had the chance to put theory into practice. I entered a handicapping contest at www.caribsports.com. Each competitor puts up $250 at the beginning of the year. Each week you need to pick the outcome of five NFL games against the spread. The people who pick the most winners get paid a prize. There were about 800 entrants.

My strategy was more or less to pick every home underdog, although, in most weeks there were not five home underdogs and in a few weeks there were more than five. This year home underdogs have been 47-30 against the spread. Not coincidentally, my record in the handicappers challenge is 49-31. That is good enough to put me in a tie for 7th place out of the 800 participants. Most of the other top scores are bettors who also bet a lot of home underdogs.

Unfortunately for me, the payoff structure is highly skewed, with first place paying $70,000 and seventh place paying “only” $2,500. So I need a great last week to propel me up the standings.

Knowing the top bettors all pick lots of home underdogs, it means that this week I have no choice but to bet against home underdogs. I need to make sure my picks are as different as possible from the picks of the people in front of me. I gain a lot in dollars if I do much better than the people in front of me. If I do much worse, it doesn’t cost me much.


mhertz

Just make sure the people behind you don't bet all the home favorites and jump ahead of you in the standings!

jkachner

Steve, I noticed that the 2005-6 NFL season was a horrible one to play the home underdog. Is it possible that a sizable portion of big gamblers read your paper, played the home underdog, and forced the bookmakers to shift the line in the opposite direction? After a year of the strategy not working out, sports bettors went back to their usual tendencies, perhaps even overcompensating by avoiding the home underdogs and propelling you to 7th place. I assume that a single season of games in which the road team was favored is not a large enough sample to draw any conclusions. I'm even curious if 20 years is enough. As of this writing, I haven't read your paper. I'll give it a thorough perusal and get back to you.

Also, how's Pokernomics coming?

EmilyAnabel

I didn't find a way of listing the leaderboard at http://www.caribsports.com/docs/nflchallenge.html without having an account there, which seems to require credit card information. Can someone figure out a way of listing the leaderboard without an account or post the standings here? I'm curious how much of a spread there is from first to seventh.

Steven D. Levitt

EmilyAnabel,

The standings are (wins-losses-ties):

1)52-27-1
2)51-29
3)50-29-1
4)50-30
5)49-30-1
6)48-29-3
7)49-31 (3 tied)

rolub

Fortunately (unfortunately?) I only see 2 home dogs this week, and both are looking to be close to even picks anyways.

I'm not sure how big of an NFL fan/bettor you are, but the only game this week that means a ton to both teams is the JAX/KC tilt. Both are still in the running for the final AFC wildcard spot, and with a 1:00 EST start time, it will still be undecided by kickoff. Despite losing at home vs. BAL two weeks ago, Kansas City is 20-2 at home in the month of December dating back to 1995. Add in the Jaguars' awful road record, including losses at WAS, HOU, and BUF, and you've got a pretty decent (yet likely popular) pick on Kansas City (-1).

Good luck this weekend.

rrob13

49-31 ATS is plenty impressive, regardless of the method. Nice work. Good luck this weekend.

studerby

Your dilemma reminds me a bit of businesses that focus extensively on adopting "best practices"; the better the peer group executes on that strategy, the tighter the companies cluster. Margins thin and products become commodities. A great strategy for "not losing", or in the gambling case "keeping yourself in the running", but not exactly the way to become a market-dominating business. As your gambling situation shows, sometimes you have to break away and ignore "best practices" to get to the pinnacle of success.

snubgodtoh

Arrowhead is brutal, especially this time of year. I would check the KC weather before you bet on the jags. It's nice in STL now, but the nasty crap in Denver is surely headed for us. If they pulled it off it would be huge for your standings assuming you are pursuing a polar strategy and others go with the Chiefs. How many places are paying, and is the utility that you will gain from idyllic Sunday spent rooting for underdogs and consuming beer and adrenaline (from taking risky picks) worth $2500? I think so. Betting on games that you could give a crap about makes them instantly interesting, it's such a great tool, especially when my beloved Rams are crapping out amid so much talent.

Jmac26

I was very intrigured when I discovered your paper a year ago. Since I am very close to sports gambling I really like looking for inefficiencies. My only problem is that if I can't find an inefficiency I still bet. I truly believe last year was an oddity for home underdogs not beating the spread. I know a few bookies and they said a lot of the small online sites that didn't have proper reserves went broke last year. That never happens so I think last year needs to be chalked up as an outlier. As for this year you can bet on home or away underdogs and your record would be great. Bill Simmons has the underdogs at
136-97-6. This can very on what type of spread you are using. Any thoughts on the crazy underdog year Steven?

Holden

I really don't understand why you think it's wise to deviate from your strategy now. Sure, the people ahead of you picked a lot of home dogs--but if home dogs have a good record, and the people ahead of you have good records, you'd expect the two sets of picks to have more in common with each other than random, right?

The question is whether they're _systematically_ picking home dogs. You could easily get a better idea of this by looking at whether their outperformance came from the weeks with low #'s of home dogs, when you were forced to make random other picks (and other people systematically picking home dogs also would have been).

DanTilkin

So, how did you do?

nphebel

To me, this sounds like a mixed-strategy matching pennies type of equilibrium. You want to pick home dogs (best prob) but also separate yourself from other competitors. If they were rational, players 2-6 would be doing the same thing. So wouldn't you want to randomize your strategy?

Kent

Gene Fama would say that home underdogs win far more games against the spread because of priced risk factors. Home teams and underdogs are riskier just like small cap and value stocks.

dylan

peyton manning should hang himself... he lost the game for the colts... i think they should bench his dumb ass

Eddie Getz

Hi Steven, any update on this strategy? I have to say I'm pretty confused by this entry. Not by the idea itself -- bet home underdogs -- that's straightforward enough. I'm confused by your claim -- "home underdogs win far more games than they should against the spread...a bettor who bet on every home underdog for the last two decades would show positive profits."

A quick query returns the following results, please let me know what I'm doing wrong here. This info really seems to go against what you claim. Over the past two decades a gambler would show a minuscule profit.

1987-2007
- 1571 games featuring home underdog
- 804 wins ATS, 728 losses ATS, 39 push ATS
- 51% wins ATS by home underdog

2000-2007
- 651 games featuring home underdog
- 321 wins ATS, 314 losses ATS, 16 push ATS
- 49% wins ATS by home underdog

2007
- 92 games featuring home underdog
- 45 wins ATS, 47 losses ATS, 0 push ATS
- 49% wins ATS by home underdog

2006
- 80 games featuring home underdog
- 46 wins ATS, 32 losses ATS, 2 push ATS
- 58% wins ATS by home underdog

2005
- 78 games featuring home underdog
- 27 wins ATS, 48 losses ATS, 3 push ATS
- 35% wins ATS by home underdog

2004
- 81 games featuring home underdog
- 36 wins ATS, 43 losses ATS, 2 push ATS
- 44% wins ATS by home underdog

2003
- 76 games featuring home underdog
- 38 wins ATS, 36 losses ATS, 2 push ATS
- 50% wins ATS by home underdog

2002
- 90 games featuring home underdog
- 52 wins ATS, 36 losses ATS, 2 push ATS
- 58% wins ATS by home underdog

2001
- 82 games featuring home underdog
- 42 wins ATS, 35 losses ATS, 5 push ATS
- 51% wins ATS by home underdog

2000
- 72 games featuring home underdog
- 35 wins ATS, 37 losses ATS, 0 push ATS
- 49% wins ATS by home underdog

Read more...

Trevor

The best way to win 60% of your bets is to bet home underdogs +7 or greater when the visiting team is playing on a surface different from their home stadium.
Ex: NE Pats (field turf) -10
Cle Browns (grass) +10
bet Cleveland

mhertz

Just make sure the people behind you don't bet all the home favorites and jump ahead of you in the standings!

jkachner

Steve, I noticed that the 2005-6 NFL season was a horrible one to play the home underdog. Is it possible that a sizable portion of big gamblers read your paper, played the home underdog, and forced the bookmakers to shift the line in the opposite direction? After a year of the strategy not working out, sports bettors went back to their usual tendencies, perhaps even overcompensating by avoiding the home underdogs and propelling you to 7th place. I assume that a single season of games in which the road team was favored is not a large enough sample to draw any conclusions. I'm even curious if 20 years is enough. As of this writing, I haven't read your paper. I'll give it a thorough perusal and get back to you.

Also, how's Pokernomics coming?

EmilyAnabel

I didn't find a way of listing the leaderboard at http://www.caribsports.com/docs/nflchallenge.html without having an account there, which seems to require credit card information. Can someone figure out a way of listing the leaderboard without an account or post the standings here? I'm curious how much of a spread there is from first to seventh.

Steven D. Levitt

EmilyAnabel,

The standings are (wins-losses-ties):

1)52-27-1
2)51-29
3)50-29-1
4)50-30
5)49-30-1
6)48-29-3
7)49-31 (3 tied)