clock menu more-arrow no yes

Filed under:

Colts vs Chiefs: What Are the Odds?

New, comments
NFL: AFC Wild Card-Indianapolis Colts at Houston Texans Thomas B. Shea-USA TODAY Sports

Heading into the divisional round playoff game, the Colts are underdogs to the Kansas City Chiefs with the current point spread at about +5-1/2. For you non-bettors, that means the Colts are expected to lose by 5.5 points.

But what does that say about the chance of a Colts win? Can that be converted into a probability? Clearly it’s less than a 50% chance, but how much less? If you listen to Chiefs fans, it’s about 0%, but I think I’ll attempt a more scientific approach rather than simply trusting the opinions of Camaro-heads.

Looking at all games since 1978 for teams where the point spread was +5.5, the actual point differentials of the completed games are all over the place.

That’s a big range of outcomes, but Vegas was fairly accurate, as about half of the games are under the spread and half over. Totaling the results up by win-loss outcome gives the following:

Wins

Category L W Total % Win
Category L W Total % Win
Point spread 228 101 329 30.7%

So, about 31% of teams that were 5-1/2 point underdogs ended up winning their game. Wait a minute, the Chiefs fans say. This is an away game for the Colts, that changes things. Yeah, not really. The point spreads already include the impact of home/away. Here is the data with just visiting teams added.

Away Wins

Category L W Total % Win
Category L W Total % Win
Point spread 228 101 329 30.7%
Away 158 71 229 31.0%

31%. Oh yeah? Well, what about outdoor games.

Outdoor Wins

Category L W Total % Win
Category L W Total % Win
Point spread 228 101 329 30.7%
Away 158 71 229 31.0%
Outdoors 162 74 236 31.4%

31%. What about Outdoors AND away?

Away + Outdoor Wins

Category L W Total % Win
Category L W Total % Win
Point spread 228 101 329 30.7%
Away 158 71 229 31.0%
Outdoors 162 74 236 31.4%
Away + Outdoors 112 53 165 32.1%

Okay, 32%.

I tried running the data for games with the temperature under 40 degrees and although there were too few games to be meaningful, the win rate was similar. The point is that no matter how you slice it, a +5.5 point spread historically translates to about a 30% chance of winning.

Of course the upcoming game’s point spread is not yet finalized and could change up until game time. So, it would be handy to have a general formula for any point spread. To create that, I took all of the games and ran a regression of Win% against the point spread. Limiting the data to just point spreads that have occurred at least 30 times since 1978, the plot looks like this.

Using the regression formula, the expected win % on +5/1-2 point games is:

  • 0.4988 - 0.0282 * 5.5 = 34.3%

So the generalized formula gives the Colts a slightly higher chance of winning than I calculated before and is likely a better measure as it is based on a much larger set of data.

The beauty of all of this is that these are just probabilities and not certainties, so no matter what happens tomorrow, I can’t be proven wrong. I love math.