Saturday, July 23, 2011

Texas A&M Football Coach Extension

Texas A & M has extended head coach Mike Sherman's contact by one year and also increased his salary. So let's take a look at how the NCAA football model evaluates the Aggies football team over the last three years.

In 2010, Texas A & M from the model finished #44 overall and #42 on offense and #47 on defense. Texas A & M had a number of high profile wins over Oklahoma, Nebraska and Texas.

In 2009, Texas A & M from the model finished #46 overall and #12 on offense and #96 on defense.

In 2008, Texas A & M from the model finished #114 overall and #107 on offense and #113 on defense.

The model shows that Texas A & M has made substantial improvement over the last three years on defense, from one of the worst in 2008 to above average in 2010. Overall, the Aggies also improved from one of the worst in 2008 to a team near the top third of all FBS teams over the last two years. Given the amount of quality recruits in Texas, one would expect Texas A & M to be better than their performance in 2008. Given their increase in overall performance, it seems that the administration at Texas A & M is expecting better overall performance.

Thursday, July 21, 2011

NHL Goalies Consistency Part II

Earlier this month I wrote about NHL goalie consistency using our simple measure of player consistency as used in The Wages of Wins. Our player consistency measure calls NHL goalies that are in the top 20% in consecutive seasons as consistent, and likewise players that were in the bottom 20% in consecutive seasons as consistent. Admittedly, being in the bottom 20% in consecutive seasons is not the type of consistency that fans or GM's are looking for. So, how many NHL goalies are consistent in each 20% "bracket" and who are the NHL goalies that are the most consistent at the top?

Remember that only 26% of NHL goalies kept their same grade in consecutive seasons from 1997/98 to 2010/11. This amounts to a total of 162 of the 616 observations. Of those 162 NHL goalies that kept the same grade in consecutive seasons, 59 of the 162 (or 36%) were in the top 20% ("A" grade) from one season to the next, followed by 12% keeping the same "B" grade, 16% staying in the same "C" grade, with only 14% keeping the same "D" grade, and the final 21% in the "F" grade.

So of those 59 NHL goalie observations that were in the top 20% consecutive seasons from 1997/98 to 2010/11, only a few NHL goalies were consistent more than one time (from one season to the next). Roberto Luongo heads the list with being in the top 20% of NHL goalies seven times, followed by Patrick Roy and Tomas Vokoun five times, Dominik Hasek four times, Ed Belfour, Henrik Lundqvist, Jean-Sebastien Giguere, Martin Brodeur and Sean Burke three times, Cristobal Huet, Jonas Hiller, Miikka Kiprusoff, Niklas Backstrom and Roman Cechmanek two times.

Of course, the time period that I evaluated covers most of Roberto Luongo's career while it covers only a small part of Patrick Roy's career; so drawing any conclusions about NHL goalies career is problematic. Yet if we keep in mind the limits of the time period (1997/98 to 2010/11), we see that Roberto Luongo was the most consistent NHL goalie during this time period.

This got me to thinking, if NHL goalies are as inconsistent as we have demonstrated, then acquiring a goalie that is consistently (say in the top 20%) is rather rare and would make an NHL goalie very valuable. Problem is that so few NHL goalies are consistent over a long period of time. Of those NHL goalies listed above, only Roberto Luongo, Patrick Roy, Henrik Lundqvist and Jonas Hiller were in the top 20% for at least 50% of the observations. That's it; four NHL goalies and Jonas Hiller had two consecutive top 20% performances out of four possible observations. Yet NHL GM's pay goalies like they can predict their performance in the future, which we show is not accurate.

Saturday, July 9, 2011

NHL Goalie Consistency

In our book, The Wages of Wins, we look at how consistent players are from one season to another. Specifically we look at NBA players, MLB hitters and NFL quarterbacks. Unfortunately we skipped NHL players, a point I want to rectify.

So in order to figure out how consistent NHL goalies are from one season to another, I have collected NHL goalie regular season performance data for thirteen seasons (1997/98 to 2010/11) with the lockout season missing for obvious reasons. Then for each season, I have evaluated each NHL goalies regular season performance exactly as I did for the 2010/11 season earlier this year. For each season, I also eliminated from evaluation any goalie that started less than 10% of regular season games (i.e. less than 8 game starts), since these goalies are most likely emergency call ups or players that were injured for a significant part of the season. Including players with limited playing starts (and playing time) are likely to result in even greater performance variance, so including these goalies will make the consistency results poorer. So for that reason, I have limited the number of NHL goalies that had eight or more game starts in consecutive seasons.

Next, I "graded" all goalies in an individual season as those that are in the top 20% of WAA receive an A, those in the next highest 20% WAA receive a B all the way down to the bottom 20% of WAA who get an F. (Not that is a grading scale I would use in class). Then I looked at how many goalies over those thirteen seasons kept the same "grade" from one season to another. Those that keep the same "grade" (i.e. are in the same quantile) are considered consistent; those that move up or down only one quantile are considered nearly consistent and those that move up or down more than two quantiles are considered inconsistent. Some may argue that this is either too restrictive a measure for measuring consistency and others may feel this is too lenient.

For example, from 1997/98 to 2010/11 if a goalie started atleast 8 games each season they would have 12 observations of consecutive seasons. Martin Broduer has played during this entire time period. During this time period, Broduer has keep his same grade (top 20% each time) four times, and moved up or down one grade five times, meaning that is "grade" has changed by more than one "grade" three times.

Over the entire time period, for goalies that started at least eight games in consecutive seasons (including 2003/04 and 2005/06 as a consecutive season due to the lockout), NHL goalies as a whole maintained their "grade" - stayed in the same quantile - 26% of the time, and moved either up or down one "grade" 34% of the time; thus about 40% of the time NHL goalies moved up or down more than one "grade".

Thus, NHL goalies are about as consistent as NFL QB's, and are less consistent than MLB batters and much less consistent than NBA players.