Pythagorean Wins: Looking Forward and Back in the AFC North
Last year, I posted a story explaining Pythagorean expectation, and what it predicted for the AFC North. Basically, you take the number of points a team scores and allows, plug it into a formula, and calculate the "ideal" number of games they should have won that season. It can also be used to predict, with fair accuracy, how a team will do the next season.
Two of the four predictions were spot-on, as the Steelers were forecasted to improve, and the Browns to decline, as they respectively under and overperformed their Pyth. expectation. The Bengals were also forecasted to improve - as we know, that didn't work out.
I'm going to take a look at the numbers again, for a few select teams, and see what we're looking at this season.
(As always, take this with a grain of salt - it's not perfect, and it can't predict injuries and performances from new personnel.)
We'll start with the Super Bowl Champion Pittsburgh Steelers, who had a point differential of +124 points, which equates to 11.84 Pyth. wins. Sound familiar? When a team closely matches their ideal wins, their following season projection is hard to gauge. With a more favorable schedule, I'm willing to bet on 11-13 wins in 2009.
(Yes, feel free to dig this story up and mock me in nine months.)
How about the team that we completed a trifecta of futility by losing to the Steelers not once, but twice, but thrice last season, a distinction previously held only by the esteemed Cleveland Browns - the Baltimore Ravens? They actually had a better projection, with a point differential of +141. That predicts 12 Pyth. wins, which they fell short of by one game. Sigh - an underperformance of one full game or more usually predicts a better team the next season. Let's hope for a Flacco sophomore slump, or for that aging defense to finally f***ing age. I'll match them with the Steelers at 11-13 wins, hopefully none against us.
And those aforementioned Cleveland Browns? The preseason AFC North favorites were badly outscored, at -118 points. This factors to 4.5 Pyth. wins, as compared to an actual total of four. Things aren't looking too hot on the Cuyahoga. But don't worry, they have a secret weapon that's sure to turn things around: The Mangenius! And, uh, Brady Quinn, who has a completion percentage roughly equivalent to the staph infection rate in the Browns locker room. And that other guy, the one who couldn't beat Charlie Frye for a starting job. Yeah, they're finally turning things around in Cleveland. 5-7 wins.
And we come to the only member of the division to never win a league title of any kind. The Cincinnati Bengals. Getting sleepy...Let's make this quick before I fall asleep on my keyboard. They were outscored by a whopping -160 points, or 10 per game. According to the formula, they had 3 Pyth. wins. They won four games, and tied in another. This sad sack team actually significantly over-performed last season. I'll spot them a better defense and a competent QB, and predict that they'll beat last year's totals easily, despite the formula. 6-8 wins, and high hopes for another 8-8 season for the most average franchise ever.
Coming soon, more AFC notables, and discover why Tony Romo and the Cowboys will go playoff winless for yet another season.
0 recs |
13 comments
| Add comment
Comments
Will you share the formula with us?
Only a life lived for others is a life worthwhile - Albert Einstein.
by fanofsteel on Jun 25, 2009 11:21 AM EDT reply actions 0 recs
Found it
Only a life lived for others is a life worthwhile - Albert Einstein.
by fanofsteel on Jun 25, 2009 11:36 AM EDT up reply actions 0 recs
Here's a little setup for an Excel Spreadsheet you can use to figure your own Pyth. Wins
A1 Points Scored
A2 Points Allowed
A3 Games Played
A4 =SUM/((A1)^2.37+(A2)^2.37)))
Just paste the formula in A4, enter the numbers in the appropriate cells and you can figure the Pyth. Wins for any matchup. I hope this helps you all win tons of cash in the office pools this season.
Only a life lived for others is a life worthwhile - Albert Einstein.
by fanofsteel on Jun 25, 2009 11:52 AM EDT reply actions 0 recs
OOPS. The formula for A4 is actually:
=SUM*(A1)^2.37/((A1)^2.37+(A2)^2.37))
It didn’t copy the whole thing.
Only a life lived for others is a life worthwhile - Albert Einstein.
by fanofsteel on Jun 25, 2009 12:16 PM EDT reply actions 0 recs
A bit more explanation
If it’s not clear, a team that underperforms or overperforms its Pyth. wins by a full game or more usually posts a better record the next season. If the difference is less than a full game, it tends to swing either way.
I’m going to go trhough the 2008 projections soon and see how the formula did, but it’s generally accurate – though less so in football than in sports with more games on the schedule.
There’s also a concept called second-order Pythagorean wins in baseball. Basically, you use a runs created formula to calculate what a team’s run differential should have been that season, plug that into the Pythagorean formula, and get an even more accurate number of expected wins. Second-order expected wins are very accurate in predicting actual future wins in baseball, so if someone can figure out a way to translate it to football, I’m interested.
by Desroko on Jun 25, 2009 12:17 PM EDT reply actions 0 recs
I'm thinking that one approach to a more accurate system could be
to take the Pyth wins for each player and then average they Pyth wins for the roster and then go heads up with the other teams roster.?
For Example the Pyth. Wins for the lineup when one set of linemen and FWP vs. other Linemen and Moore in the lineup could be different. Or in Cinci’s case the Pyth Wins with Palmer vs. the backup.
Only a life lived for others is a life worthwhile - Albert Einstein.
by fanofsteel on Jun 25, 2009 12:22 PM EDT up reply actions 0 recs
weird. The post editor is cutting part of the formula after =SUM it should be
((A3)*(A1)^2.37/((A1)^2.37+(A2)^2.37))
Only a life lived for others is a life worthwhile - Albert Einstein.
by fanofsteel on Jun 25, 2009 12:18 PM EDT reply actions 0 recs
Hmmm, that's weird,
if anybody wants to try on their own, it’s Points Scored ^ 2.37, divided by (Points Scored ^2.37 +Points Allowed ^ 2.37)
by Desroko on Jun 25, 2009 12:21 PM EDT up reply actions 0 recs
you have to take the above number and times it by the number of games played
Only a life lived for others is a life worthwhile - Albert Einstein.
by fanofsteel on Jun 25, 2009 12:25 PM EDT up reply actions 0 recs
Another approach
Although not more accurate, could give a range of projected wins would be to use a range for the factor (in the above examples 2.37). You could use 2 for the factor and the run it again with 3 as the factor which would give a projected low and high number of projected wins. Less accurate but I like having a range of numbers to work with.
Only a life lived for others is a life worthwhile - Albert Einstein.
by fanofsteel on Jun 25, 2009 12:36 PM EDT reply actions 0 recs
Actually I'd use
1.75 and 3 for the factors. That gives about the same cushion on either side fo 2.37.
Only a life lived for others is a life worthwhile - Albert Einstein.
by fanofsteel on Jun 25, 2009 12:41 PM EDT reply actions 0 recs
using 1.75 and 3
Gives a range of 10.94 to 12.64 for the Steelers going into the season. This gives statistical support to your projection of 11 to 13 wins.
This stuff gets me going!
Only a life lived for others is a life worthwhile - Albert Einstein.
by fanofsteel on Jun 25, 2009 12:44 PM EDT up reply actions 0 recs
The Standard Deviation for Pyth. Wins is 2.5
This means that a team could win/lose 2.5 games on either side of the projected Pyth. Wins and still not be a “fluke”. The Steelers could win as many as 14.34 or as few as 9.34.
Only a life lived for others is a life worthwhile - Albert Einstein.
by fanofsteel on Jun 26, 2009 6:19 PM EDT reply actions 0 recs

by 


















