"You are educated when you have the ability to listen to almost anything
without losing your temper or self-confidence." - Robert Frost

Wednesday, September 7, 2011

How to Measure Expectations

How to Measure Expectations:

For many years I wanted to try to statistically measure a “successful season”.  In chat rooms across the country, there is often the hard-core fan that insists every year is Super Bowl champions or bust.  That anything but winning is a disappointment.  But those same fans often defend players and coaches for jobs well done even during non-championship seasons.

A decade ago, I used to define a successful football season as “reaching the AFC Championship game”.  This definition worked well for the Steelers (the Cowher-led Steelers went to 6 AFC championship games in 15 seasons).  And since a majority of these were even without a dominant QB, it always seemed an acceptable goal heading into new seasons.  No excuses, right?

But I came to realize that what a Steelers fan might consider “successful” is not the same as what a Raiders fan would, or an Eagles fan, or a Lions fan.  The truth is, each team enters the season with an “expected” amount of success, often defined by the media pundits and analysts. And often, exceeding those expectations can satisfy many fan bases.  Even today, you hear about Tampa Bay’s “successful” season of 2010, even though they failed to make the playoffs.

The obvious difficulty is trying to measure the media/fan base expectations of a game or season.  Pre-season predictions hardly seem “scientific” enough for accurate analysis.

That caused me to turn to another neutral evaluator of teams:  Las Vegas.

Las Vegas predicts the likelihood of victory every week using odds and point spreads and vigorish.   And these odds are surprisingly unbiased in their analysis.

What if I assign an “expected win %” to each game a team plays (using Vegas odds) and then analyze how that team fairs compared to this baseline?

After experimentation on a number of teams, this became a surprising easy statistical measure of over- or underachieving.  And in every case, matched exactly the mood of the corresponding fan base.  Especially if given time to reflect after an emotional playoff loss.

Periodically throughout this season, I will look at how the Steelers are performing based on this analysis and I promise it will correctly match the current mood of Steelers Nation.  In fact, I have been doing this analysis over the past five seasons (since 2006).  The results are listed below:

YEAR         ACTUAL W-L       EXPECTED W-L            PLAYOFFS               Over/Underachieved

2006                8-8                       9.6 - 6.4                         X                               Underachieved
2007              10-6                     11.4 - 4.6        Lost (H) Wildcard               Underachieved
2008              12-4                      9.7 - 6.3                  Won SB                        Overachieved
2009                9-7                      11.0 - 5.0                        X                               Underachieved
2010              12-4                     10.1 - 5.9                  Lost SB                         Overachieved

Other websites construct expected win totals using complex formulas and point differential.  This analysis uses real-time expectations based on injuries, weather, streaks and the million other factors used to create Las Vegas betting lines each week.

If you would like me to analyze another team or two this year, I will be glad to “run the numbers” or share my spreadsheet and calculations with you.  It is a fun exercise that hasn’t failed to predict how a fair and unbiased fan should evaluate his/her own team.

Enjoy your seasons everyone.  I’m looking forward to 2011 and as always, thanks for reading.

No comments:

Post a Comment