Were the Colts the Luckiest Team in the NFL in 2009?

by mgrex03 on Jul 23, 2010 8:31 AM EDT in Winning Stats

Over the course of a season, teams have their share of good luck, and bad luck, and most of the time the two don't line up exactly the same, meaning some teams face more bad luck than others. Sometimes it's pretty obvious which teams experience very good luck (like the Browns from last year), and others it's tough to put an actual value to it. But that doesn't mean we can't do it.

A very common way to estimate Winning Percentage in football is to use the Pythagorean Expectation formula:

Winning % = Points Scored^2.37 / (Points Scored^2.37 + Points Allowed^2.37)

This was originally invented by Bill James for Baseball, but has been modified by several other sports, including football, to the exponents shown above. By doing this, you have an estimated Winning Percentage that you can compare against the actual Winning Percentage, and the difference becomes our "Luck" value. It tells us a little more about the teams season, showing how many wins the should have had, based on Points Scored/Allowed.

While this is a good method, I think we can do better. If you remember correctly, our Winning Stats has both an Offensive and Defensive "Score", so why not use them, as they take into account way more than just Points. Also, I found another method that spits out a Winning Percentage that takes into account the consistency of the team as well. It's called the Correlated Gaussian Method (CGM), and I got the idea from Ken Pomeroy, who runs the awesome stats site for College Basketball. I'm not going to explain the method here, but if you have questions, feel free to shoot me an email, and I'll explain what I did.

OK, enough set-up. Which team was the luckiest in 2009? And what team had the worst luck? Let's find out...

Here's the results from 2009. Numbers in Red are Bad Luck:

Team	Win Percentage					Wins
Team	Actual	CGM	Diff	Pythag	Diff	Actual	CGM	Diff	Pythag	Diff
49ers	0.5000	0.4051	0.0949	0.5941	-0.0941	8	6.5	1.5184	9.5	-1.5056
Bears	0.4375	0.2625	0.1750	0.4196	0.0179	7	4.2	2.8007	6.7	0.2872
Bengals	0.6250	0.6801	-0.0551	0.5278	0.0972	10	10.9	-0.8814	8.4	1.5550
Bills	0.3750	0.3133	0.0617	0.3648	0.0102	6	5.0	0.9871	5.8	0.1626
Broncos	0.5000	0.5654	-0.0654	0.5036	-0.0036	8	9.0	-1.0459	8.1	-0.0583
Browns	0.3125	0.1166	0.1959	0.2672	0.0453	5	1.9	3.1339	4.3	0.7247
Buccaneers	0.1875	0.1459	0.0416	0.2366	-0.0491	3	2.3	0.6660	3.8	-0.7854
Cardinals	0.6250	0.5824	0.0426	0.5840	0.0410	10	9.3	0.6818	9.3	0.6563
Chargers	0.8125	0.6290	0.1835	0.6961	0.1164	13	10.1	2.9356	11.1	1.8618
Chiefs	0.2500	0.1804	0.0696	0.2957	-0.0457	4	2.9	1.1133	4.7	-0.7314
Colts	0.8750	0.7409	0.1341	0.6726	0.2024	14	11.9	2.1451	10.8	3.2380
Cowboys	0.6875	0.7207	-0.0332	0.7049	-0.0174	11	11.5	-0.5319	11.3	-0.2785
Dolphins	0.4375	0.5987	-0.1612	0.4527	-0.0152	7	9.6	-2.5787	7.2	-0.2435
Eagles	0.6875	0.7292	-0.0417	0.6392	0.0483	11	11.7	-0.6669	10.2	0.7722
Falcons	0.5625	0.4565	0.1060	0.5651	-0.0026	9	7.3	1.6963	9.0	-0.0423
Giants	0.5000	0.6180	-0.1180	0.4643	0.0357	8	9.9	-1.8877	7.4	0.5710
Jaguars	0.4375	0.4617	-0.0242	0.3451	0.0924	7	7.4	-0.3865	5.5	1.4782
Jets	0.5625	0.5815	-0.0190	0.7151	-0.1526	9	9.3	-0.3041	11.4	-2.4421
Lions	0.1250	0.0798	0.0452	0.1820	-0.0570	2	1.3	0.7240	2.9	-0.9116
Packers	0.6875	0.7980	-0.1105	0.7392	-0.0517	11	12.8	-1.7683	11.8	-0.8278
Panthers	0.5000	0.3753	0.1247	0.5133	-0.0133	8	6.0	1.9958	8.2	-0.2130
Patriots	0.6250	0.7539	-0.1289	0.7228	-0.0978	10	12.1	-2.0621	11.6	-1.5641
Raiders	0.3125	0.0966	0.2159	0.1750	0.1375	5	1.5	3.4541	2.8	2.2004
Rams	0.0625	0.0951	-0.0326	0.1031	-0.0406	1	1.5	-0.5209	1.6	-0.6493
Ravens	0.5625	0.7046	-0.1421	0.7227	-0.1602	9	11.3	-2.2729	11.6	-2.5633
Redskins	0.2500	0.4906	-0.2406	0.3650	-0.1150	4	7.8	-3.8491	5.8	-1.8402
Saints	0.8125	0.8463	-0.0338	0.7219	0.0906	13	13.5	-0.5406	11.6	1.4493
Seahawks	0.3125	0.3597	-0.0472	0.3132	-0.0007	5	5.8	-0.7547	5.0	-0.0108
Steelers	0.5625	0.7681	-0.2056	0.5749	-0.0124	9	12.3	-3.2903	9.2	-0.1981
Texans	0.5625	0.6637	-0.1012	0.5896	-0.0271	9	10.6	-1.6198	9.4	-0.4335
Titans	0.5000	0.4357	0.0643	0.4252	0.0748	8	7.0	1.0283	6.8	1.1964
Vikings	0.7500	0.7878	-0.0378	0.7253	0.0247	12	12.6	-0.6040	11.6	0.3947

Some thoughts:

The Colts were not the luckiest team in the league, but were the 5th luckiest, accounting for over 2 full games. I'm sure anyone can think of a handful of moments that factor into that, but truth be told, the Colts had some good fortune last year.
The luckiest team? The Raiders, with almost 3.5 of their 5 wins coming from luck. The Browns were also very lucky last year to win 5 games, including their final 4. Can't really argue with this.
The unluckiest team? The Redskins, losing almost 4 games due to bad luck. The New Orleans game clearly jumps out at me, but they outplayed their lousy record last year. The Steelers also underperformed by over 3 games.
There are certainly some discrepancies between the CGM and the Pythagorean formula, but for the most part, they agree with each other. The 49ers seem to be the biggest difference, with a 3 game difference. They scored more points than their actual offensive efficiency, which is the reason for the difference.

Now, I've made a pretty big assumption saying that the entire difference between actual and calculated Winning Percentage is due to Luck. There are clearly other factors that lead to the differences, but it's just easier to tidy it up into one word. Some luck is based on putting yourself in good situations, which many good teams do. I wanted to make sure everyone understood my assumptions.

In my next article, I'll take a look at the luckiest/unluckiest teams since 2001, and whether luck from one year can predict how good/bad a team will be the following year.

Tweet Comment 14 comments | 2 recs |

Do you like this story?

Comments

good stuff

hopefully there is some kind of correlation between year to year figures.

Blogging about my summer at a Chinese law firm.
NEW BLOG, as my original one is blocked by the Great Firewall.
http://ajinshanghai.wordpress.com/

by JustAJ on Jul 23, 2010 1:18 AM EDT reply actions

In my experience with the Pythag, there are certian factors that need to be taken into account

Obvouisly, the Colts Week 16-17 points scored and allowed (22 and 59) is not really indicitive of the team’s true ability. Their PS-PA heading into week 16 was 9.5 (out of 14 games), so a 9.5-4.5, which is better than their final of 10.8-5.2

Another factor is that winning 10-3 is better in pythag than winning 37-30. Pythagorean projection greatly favors defensive teams, just by the way the metric is set up.

And finally, teams that win 13-15 games are almost always outperforming their projections. I think only two or three teams have ever won that many games and in some way underperformed. The only one I can think of off hand was the 1999 Rams, who finished 13-3 scoring 526 points and giving up 242.

EVH+DLR=BFFs........ God I Hope So!!

by dmstorm22 on Jul 23, 2010 8:12 AM EDT reply actions

Question: do these calculations take penalties and bad/questionable calls into account??

by coltsfoot85 on Jul 23, 2010 2:08 PM EDT up reply actions

Sort of

Penalty Yards / Play is one of the factors, so it plays a part, even though it is pretty minor.

Bad/Questionable calls aren’t really quantifiable, so they aren’t included. Like Hochuli’s bad call on the Chargers can’t possible be explicitly taken into account, but things like that will manifest themselves in the full data.

Creator and developer of the Winning Stats.

by mgrex03 on Jul 23, 2010 2:54 PM EDT up reply actions

Interesting stuff. Thanks!

Careful what you wish for... "A government big enough to give you everything you want is a government big enough to take from you everything you have." Gerald Ford, 38th US president

by teej813 on Jul 23, 2010 11:01 AM EDT reply actions

Hello

I’m new here and just wanted to drop by!

That place was for diehard sports fans. I only follow my team when they're in the playoffs" - Homer Simpson
Join the Lacrosse community The Lacrosse Blog

by bestbostonsports on Jul 23, 2010 11:35 AM EDT reply actions

Welcome!

We have a special place for folks from Boston. You get there by clicking that red X in the upper-right. :D

/kidding… glad to have you here.

Careful what you wish for... "A government big enough to give you everything you want is a government big enough to take from you everything you have." Gerald Ford, 38th US president

by teej813 on Jul 23, 2010 1:03 PM EDT up reply actions

haha

Hi, thanks for welcoming me!

That place was for diehard sports fans. I only follow my team when they're in the playoffs" - Homer Simpson
Join the Lacrosse community The Lacrosse Blog

by bestbostonsports on Jul 23, 2010 1:08 PM EDT up reply actions

Thanks for laughing.

I’m just messin’ with you. You’ll find this is a pretty friendly place.

Careful what you wish for... "A government big enough to give you everything you want is a government big enough to take from you everything you have." Gerald Ford, 38th US president

by teej813 on Jul 23, 2010 1:12 PM EDT up reply actions

Yeah

I’ve read stuff here before, but I just became a memeber a few days ago

That place was for diehard sports fans. I only follow my team when they're in the playoffs" - Homer Simpson
Join the Lacrosse community The Lacrosse Blog

by bestbostonsports on Jul 23, 2010 1:18 PM EDT up reply actions

Ok, cool. Glad you decided to join us.

Careful what you wish for... "A government big enough to give you everything you want is a government big enough to take from you everything you have." Gerald Ford, 38th US president

by teej813 on Jul 23, 2010 2:34 PM EDT up reply actions

I was SURE that the Saints would be the "luckiest."

(cough)onsidekick(cough)

"I throw, you catch. It's NOT that hard!"
Peyton Manning, SNL, 2007

by peytonsthebest on Jul 23, 2010 2:53 PM EDT reply actions

haha... they get my vote!

Careful what you wish for... "A government big enough to give you everything you want is a government big enough to take from you everything you have." Gerald Ford, 38th US president

by teej813 on Jul 23, 2010 3:26 PM EDT up reply actions

There's a better word than "luck" for this in statistics...

and it’s “error.” Whenever you make a model to explain an event, the difference between the model’s expected outcome and the actual outcome is its “error.” You can say some of that error is explained by luck (though I view that is mostly a guess with little support, since luck will play into the variables that both the Pythag. and your model use), but realistically, any difference here we find probably says nothing about the team’s luck and everything about the inadequacy of an imperfect statistical model. So, the real question for me, Mgrex, is why do you think this model captures “skill” but not “luck”, so that your error can be mostly attributed to luck?

by dsvirsky on Jul 23, 2010 8:14 PM EDT reply actions