Why Are the Colts 5-Point Favorites?
Posted by Neil Paine on February 7, 2010
Just a quick pre-Super Bowl thought... Given that "Power Rating" style systems like the SRS are designed to retrodict game scores and produce the lowest retrodictive error, you can compare two teams' ratings and derive a decent expectation of how the teams would match up on a neutral field. If we throw out the Week 17 games by all division winners + the Colts-Jets Week 16 game (to only capture games each team was trying to win) and minimize squared errors, we get these ratings:
| Team | Rating |
|---|---|
| atl | 4.4182 |
| buf | -4.4528 |
| car | 2.1842 |
| chi | -3.7367 |
| cin | 2.0606 |
| cle | -8.1995 |
| clt | 10.4767 |
| crd | 0.9833 |
| dal | 5.2163 |
| den | 0.7210 |
| det | -14.1667 |
| gnb | 5.6217 |
| htx | 1.5537 |
| jax | -6.2228 |
| kan | -8.3436 |
| mia | 1.0747 |
| min | 7.4098 |
| nor | 12.0872 |
| nwe | 10.0121 |
| nyg | 1.6399 |
| nyj | 5.6000 |
| oti | -2.4651 |
| phi | 6.0969 |
| pit | 1.9830 |
| rai | -10.2420 |
| ram | -16.9181 |
| rav | 8.5104 |
| sdg | 7.1117 |
| sea | -8.9631 |
| sfo | 0.5119 |
| tam | -6.2509 |
| was | -5.3118 |
| HFA | 2.0733 |
One problem: minimizing squared errors gives undue weight to extreme blowouts, like New England's beatdown of Tennessee (which in turn affects the ratings of both Indy and N'awlins). What if we minimized absolute errors instead?
| Team | Rating |
|---|---|
| atl | 4.3532 |
| buf | -4.3880 |
| car | -1.2291 |
| chi | 1.1098 |
| cin | -0.4821 |
| cle | -7.0644 |
| clt | 7.2744 |
| crd | 1.2745 |
| dal | 6.6062 |
| den | 6.9355 |
| det | -8.3080 |
| gnb | 6.8568 |
| htx | 3.8568 |
| jax | -8.6018 |
| kan | -9.4745 |
| mia | -1.5529 |
| min | 9.3532 |
| nor | 9.9356 |
| nwe | 6.3532 |
| nyg | -1.0644 |
| nyj | 4.0297 |
| oti | -0.4589 |
| phi | 7.5274 |
| pit | 2.1815 |
| rai | -7.7596 |
| ram | -14.6633 |
| rav | 2.7811 |
| sdg | 2.3533 |
| sea | -7.3079 |
| sfo | 3.2745 |
| tam | -8.8115 |
| was | -4.8902 |
| HFA | 2.4177 |
OK, well what if we just isolated the second half of the season, when New Orleans supposedly fell back to earth? Minimizing squared errors, we get:
| Team | Rating |
|---|---|
| atl | 3.2130 |
| buf | -5.3046 |
| car | 7.5346 |
| chi | -6.5979 |
| cin | 0.1855 |
| cle | -2.5621 |
| clt | 9.4075 |
| crd | 0.3849 |
| dal | 4.9098 |
| den | -5.0528 |
| det | -15.3099 |
| gnb | 6.5715 |
| htx | 1.3528 |
| jax | -4.7835 |
| kan | -7.7046 |
| mia | -1.3911 |
| min | 6.4942 |
| nor | 9.8053 |
| nwe | 6.0161 |
| nyg | -0.2253 |
| nyj | 5.8909 |
| oti | 3.8817 |
| phi | 4.2944 |
| pit | 0.5480 |
| rai | -8.1044 |
| ram | -15.6764 |
| rav | 7.1223 |
| sdg | 12.0403 |
| sea | -12.1215 |
| sfo | -0.0113 |
| tam | -1.5260 |
| was | -3.2813 |
| HFA | 2.7451 |
And minimizing absolute errors:
| Team | Rating |
|---|---|
| atl | 3.8829 |
| buf | -3.7508 |
| car | 6.6105 |
| chi | -8.0339 |
| cin | -2.0338 |
| cle | -4.5066 |
| clt | 6.5776 |
| crd | 2.2898 |
| dal | -0.5066 |
| den | -2.1175 |
| det | -11.2685 |
| gnb | 3.6608 |
| htx | -6.1171 |
| jax | -5.8119 |
| kan | -4.3391 |
| mia | 2.8829 |
| min | 13.2713 |
| nor | 11.5774 |
| nwe | 8.1881 |
| nyg | 1.4934 |
| nyj | 6.9440 |
| oti | 1.5030 |
| phi | 1.4931 |
| pit | -1.8119 |
| rai | -12.8119 |
| ram | -13.9465 |
| rav | -0.1171 |
| sdg | 7.1881 |
| sea | -7.1918 |
| sfo | -1.9633 |
| tam | 5.5776 |
| was | 3.1878 |
| HFA | 4.6947 |
Judging from these ratings, New Orleans should be somewhere between 5-point and 1/2-point favorites in the Super Bowl tonight. So why are the Colts 5-point favorites instead, and why have very few people (other than FO's Aaron Schatz) given the Saints a chance to win? Reading and listening to some of the chatter this week, you'd think we'd be looking at an on-paper mismatch like Super Bowl XLII (Pats SRS: 20.1; Giants SRS: 3.3) instead of one of the closest SB matchups in recent memory. In fact, if anything the odds should be leaning towards New Orleans. What gives, Vegas?
Related posts:
ShareThis
February 7th, 2010 at 4:15 pm
I checked some lines, and 5-9 (against) is still being offered on the Saints. Assuming the game is a tossup and you're suggesting it's actually favoring the Saints, then by betting 2/7 of your total net worth (kelly optimal amount), this yields an expected gain of barely over 4%. An overnight gain of 4% is pretty nice. Granted, betting almost 30% of your net worth is scary, but if SRS is accurate, that's the conclusion.
To me saying "What gives, Vegas?" is like asking why a certain stock is valued at a certain price, so I tend to look for discrepancies rather than doubting the accuracy of the price. I'm curious what the missing factor is, but I don't know what it is. That's my excuse for not betting despite the odds being out of whack versus my perceptions and those of everyone else who runs numbers, SRS, FO, advnflstats.
Does any nonsubjective analysis by any site give the Colts nearly a 2/3 chance of winning? (Colts 64.52% vs Saints 35.71% were the implied odds of the lines I saw. Think either team is more likely, and betting is profitable)
February 7th, 2010 at 4:20 pm
The five-point spread has to be a perceptual thing with New Orleans being regarded as the inexperienced team playing in its first Super Bowl against a Colts team that has regularly played in big games - AFC titles, a Super Bowl, and multiple playoffs.
Also, the Saints were outgained substantially in the game against Minnesota, despite prevailing by a field goal in OT. So perhaps there is also a perception that New Orleans really wasn't the best that the NFC had to offer. But I think the experience factor explains most of the five-point differential.
February 7th, 2010 at 4:33 pm
In their last 6 meaningful games the Saints are 4-2, and 3 of those wins were by a FG (two in OT). The Colts are 5-0 in meaningful games since December, and have 4 wins by 10+ pts. So we've seen a bit of a role reversal in the two teams as the season has wore on. The Saints are relying on 4th quarter drives more than Indy now.
Ultimately I see the Colts as a team that should be favored since they haven't lost a game they tried to win yet this year (against a higher SOS), meanwhile the Saints have lost twice (once to a horrible Tampa Bay team), and really should have lost to a bad Washington team (Suisham!). And lots of people see the NFC-C as a fortunate win over Minnesota. Even with all the turnovers, the Vikings were right there with them.
February 7th, 2010 at 4:45 pm
Judging from these ratings, New Orleans should be somewhere between 5-point and 1/2-point favorites in the Super Bowl tonight. So why are the Colts 5-point favorites instead...
Three ideas, FWIW:
1) Bookies setting the spread to equalize what the bettors are betting, nothing to do with which *they* think is actually the better team.
2) No offense, but SRS is not the only rating system. Sagarin has the Colts (modestly) better even after giving away their last two regular season games, etc.
3) The Saints are at the top of the league in interceptions and constantly talk about how they win by making the big takeaway on defense. Lots of analysis on this site and elsewhere says that picks and takeaways are luck. Maybe fans who read this site are knocking the Saints down a few points as a result. It's your fault!
February 7th, 2010 at 4:52 pm
The colts pulled off very close victories the whole season, meaning they get good win loss bad SRS. People say that Manning was responsible for all the close victories going to the colts, but that begs the question, why wasn't he trying to get a lead in the first half, but suddenly tried in the 4th quarter? More likely, they got lucky, and thus they are over rated on their W/L record. Also, everyone thinks drew brees is much worse QB than the stats says he is.
I saw in a documentary once that vegas tries to set the point spread so that approximately half the people bet on either side durring the superbowl, and this choice of safety vs expected value is consistance with what wizard of odds has noted. So if this is true, then maybe they just notice that viewers all overly favour the colts.
February 8th, 2010 at 9:18 am
Checking after the fact, it looks like the books were taking a position on the Saints. even at 5 point favorites, as most of the action was on the Colts. http://thesportseconomist.com/2010/02/super-bowl-wrap-up.htm I'm not sure what to make of that, except that I don't understand the inner workings of a sports book as I've always assumed that the goal was to maximize expected house profits with either outcome.
February 9th, 2010 at 12:30 am
I didn't get it, and I still don't get it, except maybe the betting public overvalues quarterbacks and undervalues everything else -- especially receivers and coaching.
The last time I saw a line this wonky in a SB was Denver - Green Bay, where the Pack was favoured by some huge number - 13 I think. And the game was similar: Denver started out nervous, then realized it was just another game. The rest is history....