Filed under:

Introduction to H.A.W.K.E.Y.E.S.

Hawkeye Anterior Win Kalculator Engineered to Yield Eventual Scores

Welcome to the first ever installment of the Hawkeye Anterior Win Kalculator Engineered to Yield Eventual Scores. Around these parts we’ll call it H.A.W.K.E.Y.E.S.

This lovely piece of modern Excel engineering is built to do one thing and one thing only: predict the outcome of Iowa football games. Using the outcomes from every Iowa football game from the 2002 season through the 2014 season, I looked at some of the statistics that explained why Iowa won or lost those contests. Then, I used some fancy statistical trickery to see if I could use that data to predict the scores and outcomes from the magical 2015 campaign. Much to my own surprise, H.A.W.K.E.Y.E.S. predicted that the Hawks would win all 12 regular season games and lose to MSU and Stanford. Since we never actually played Stanford (I purged that game from my memory) in a bowl game that is named after a well-known flower, H.A.W.K.E.Y.E.S. correctly predicted Iowa's win/loss record. What's more, it even predicted the scores fairly closely.

The model itself is fairly simple. I identified a handful of statistics that seem to matter in the outcome of a football game, particularly as it relates to scores. Passing yards, rushing yards, turnovers, penalties, tempo, efficiency, and home field advantage. Now, I know I'm not perfect, but neither is the game of football and teams rarely adhere to mathematical models. And the more intricate and complicated the model, the less likely it is to resemble reality. Fumbles bounce weird ways, wind and weather intervene and birds fly in front of Randy Johnson. Random things happen.

By focusing on the more generic and broad statistics mentioned above, a lot of the randomness of the world starts to be absorbed into the prediction. The final step was to collect the above statistics for all of Iowa's opponents in 2016. Obviously, with no games played, it came down to using 2015 statistics. After each week of games, I'll be able to update each team so that their 2016 performance will slowly start to matter more and more. Hopefully by the time conference play begins against Rutgers, there will be enough data on 2016 to start shedding data from 2015.

After a little bit of tweaking here and there, some minor updates under the hood, and some rejiggering to deal with those silly FCS teams, I felt like H.A.W.K.E.Y.E.S. was ready to take on the 2016 season for its first road test. The first hurdle was Miami of Ohio. And while H.A.W.K.E.Y.E.S. is meant to be used in the anterior, it had to be used in the posterior in this case (insert "butt of the joke" joke here). So what did H.A.W.K.E.Y.E.S. say would happen based on their 2015 performances?

So, H.A.W.K.E.Y.E.S. predicted a three-score win for Iowa (17 points) and Iowa delivered a four-score win (24 points). So the computer was a little conservative on the outcome of this game. This is not surprising, as it had no data from the 2016 team with which to work. As the season goes on and I can incorporate data points from this season, I suspect H.A.W.K.E.Y.E.S. will get a little more accurate.

So what did H.A.W.K.E.Y.E.S. have to say about the rest of the season? Well, apparently we'll be going back to Indianapolis. But I'd hold off on buying those plane tickets just yet, since the only 2016 data we have is against arguably Iowa's weakest opponent of the season. As we learn more about this team, so will H.A.W.K.E.Y.E.S., and these predictions will be updated each week as we learn more about our opponents, too. But it does feel good knowing that an inanimate piece of software on my computer likes the Hawkeyes as much as I do.