The Numbers Predict The Season

Last May I reviewed two preseason projections for the Kentucky football season. I focused on Football Outsiders' S&P+ and ESPN's Football Power Index (FPI). Projecting outcomes in football using statistics is notoriously wonky given the reality of a small sample size relative to something like baseball (12 games versus 180 games). Nonetheless, the recent advancements have helped close the gap between a sea of numbers, and comfortably reliable narratives.

I thought it instructive to highlight what S&P+ and FPI are saying about UK  this coming season, while reviewing how accurate last season's preseason projections turned out. That information will help to put the 2015 preseason projections in perspective. Finally, I'm going to dust off the ol' TI-82 calculator and try my hand at Pythagorean projection to see if UK under- or overachieved last season, because underachieving teams tend to improve the following season.

S&P+

S&P+ was created by Bill Connelly who runs the Missouri blog Rock M Nation. This system is one half of the F+ advanced statistical rankings I use during the season for the game previews. This time last year, S&P+ projected UK to have the 75th best team in the country. The team ended up finishing ranked 58th overall, but S&P+ got pretty close to to aspects of UK's season (specifically the offense's rating) . So, this model ultimately underestimated UK a bit last season. What does it say about UK this season?

This preseason, the model is predicting UK will finish ranked 52nd (but these rankings will be updated at the end of the summer and so is subject to change). Looking at the teams that finished last season ranked between 45th and 55th, the median number of wins was a little over 6 total victories. For an example of the nearest SEC East rival, South Carolina finished ranked 43rd, and they finished the regular season 6-6.

S&P+ seems to suggest UK will win 6 games. Let's hope it underestimates the team as drastically as last season for the second year in a row. If that happens, it would place UK within reach of 7 regular season wins.

FPI

FPI currently projects UK to be the 48th best team in the country which is a step up from last May when it ranked UK 54th overall. This year, it projects against a perfectly average defense, UK's offense will add 2.5 points per game, while the defense will add 3.1 points per game. Compared to last year, UK's offense is expected to do worse (3.6 points per game last season), while the defense is expected to improve (0.8 points per game last season). I would have thought the offense would be better or at least hold serve.

A ranking of 48th at the end of last season would see UK competing for going 6-6 in the SEC most likely. I questioned the opaqueness of FPI in last year's post, but the index did the best job predicting where UK would finished ranked. UK finished 60th which was close to the preseason  ranking of 54th. A ranking of 48th puts the team within spitting distance of UofL (44th), South Carolina (40th), and Vanderbilt (51st).

Pythagorean Projection

The Pythagorean projection formula suggests UK is poised to win more games this season than in 2014. Last season, UK finished the season 5-7, but the projection suggests UK played well enough over the course of the season to win 5.5 games. Teams that win one less game per season than their projection tend to improve the following year in the NFL -- especially if their record was near .500.

UK, while not a professional team, was both borderline .500 last season, and lost roughly one more game than the projection says it should have. Now would be a time to recall UK's 0-2 record in one-score games (*cough* Florida *cough* Louisville). A team can't always lose one-score games as life eventually reverts to the mean.

Conclusion

S&P+ believes UK will be better this season, and in fact, if UK finishes the season in exactly the spot S&P+ projects it will fall in 6-6 territory based on last season's results.  The FPI also expects UK to be better, and it was the most accurate forecaster last preseason. Finally, the Pythagorean projection formula also seems to suggest UK will surpass its win total from last season. All three forecasts suggest UK will be improved, but don't expect a 7 win season based on these numbers.

What specifically could keep UK from reaching more than 6 wins aside from strength of schedule? The few things that went right last season probably won't continue as all things tend to eventually revert to the mean. For example, UK had a strong turnover margin last season causing 23 turnovers while only committing 15 for a turnover margin of +8. UK had 15 interceptions last season, and recall just the year prior UK's secondary only had one interception the entire season. That's how quickly things can change - and dramatically.

UK actually scored 6(!) defensive touchdowns last year, while only allowing 2 themselves, per ESPN game play-by-plays. That good fortune is as unlikely to continue as the misfortune of UK giving up 4 special teams touchdowns again this season. These things shift back-and-forth. It would require a good deal of luck to repeat 2014's successes while also not experiencing that campaign's drawbacks. There's just too much randomness.

If UK is to ascend through the SEC's ranks then it'll have to do so by crawling over other teams in the race to the summit. If UK is going to go 6-6 and challenge for 7-5, it will have to come at the expense of other teams. Which teams are the most vulnerable to trading places with UK this season is a topic I intend to explore in a future post.