At this point in the Handicapping Tutorial Series, we have covered the basic strategy on how I project a score output for each team on the ground and through the air. These output projections are rarely perfectly round numbers, so what do we do? This is the human element of handicapping, of course aided by statistical data and theories. But first, lets summarize what we’ve covered.

Part 1:

In Part 1 of the series, I provided all of the tools that I use most frequently in capping a college football game. As you create your model, or do any work in excel really, there are many of different shortcuts to make things faster. If you say to yourself, “there must be an easier way to do this,”…I can guarantee you there is.

Part 2:

In Part 2 of the series, I provided some detail of the different sources I use before the season, during the season, and after the season. Part of handicapping is filtering through the smoke and finding what is important, and what isn’t, while always remembering, there is never going to be much that completely deviates the outcome.

Part 3:

In Part 3, we finally got into the team vs team handicapping. I broke down how a football game is a sum of its parts, and how we can analyze those different parts to create an output.

Part 4:

In Part 4, we took the Team vs Team output and took it to the next level, incorporating turnovers and field goals.

---

What does a correlation mean?

In the end of Part 4 of the series, I mentioned the correlation equation I use. If you read Part 1, it will provide access to all the data necessary. In college football, because of the length of time that coaches stay at schools, players stay at schools, and how quick the game changes, I only utilize data for the past 5 years. The correlation helps us realize what components of an offense and defense are relevant, from a statistical standpoint.

When you run the correlation, you will get a number between 1 and -1. A positive one means that the two statistics you are comparing are perfectly correlated, or in layman terms, they move exactly in unison. If you get a negative 1, or a number close to it, the numbers move as opposites. Easy enough right?

One last thing – anything over .3 in sports is “statistically” significant enough to consider it game changing.

What correlations matter?

So I’m not going to do all the work for everyone, because I’m a proponent of you all learning, finding out new things, and sharing the knowledge so we all get better, but I will give you some things. The top 5 stats on offense, in order, that correlate with scoring are:

1. Yards per Pass Attempt 2. 3rd Down Conversion 3. Yards per Carry 4. Total Pass Yards 5. Completion Percentage.

As you can see, 3 of the 5 categories are passing related. This makes sense, the game has moved increasing towards a passing game (if you really want to be successful).

On defense, the Top 5 stats the correlate with preventing scoring are: 1. Yards per play 2. Yards per game 3. Yards per Pass 4. Yards per Rush 5. Total Rushing Yards.

Also makes sense. Don’t let your team move the ball, and they don’t score. Duh.

What does this all mean?

Let’s review the Ole Miss vs. Alabama game we have been using as an example.

 

Our goal is to properly adjust the score projection for each team.

One of the most important things to remember from this model is that it is only as good as the data feeding it. In other terms, if we’re 5 weeks into the season, you only have 5 weeks of a teams data. Maybe they played trash out of conference teams? Maybe their starting QB was hurt? So, keeping that in mind, how do we assess how a team will perform?

We will go through many different aspects to consider in following post, but tonight we will just look at the correlations.

The Application of the Correlation

Where will a team succeed? In this projection, Ole Miss is supposed to score 1.1 rushing TDs and .79 passing. This doesn’t leave us much room to adjust. Both of these numbers are pretty close to an even number.

The real question I ask myself is: Do I see Ole Miss scoring on the ground vs Bama? Do I think they can get one through the air?

The correlations help me answer that question. Because Bama is so damn good, it is nearly a mute point, but at this point in the season, Ole Miss was the best offense that Alabama had seen. Ole Miss ran a read option offense that was fast paced, and moved the ball well. I was pretty confident that Ole Miss would get some points on the board. They were very successful on 3rd downs, they passed well and could run according to the correlations.

On the other hand, how many was Alabama going to score? This was the real question since I assumed Ole Miss would score between 10 and 17.

The Tide were projected to score 1.86 TDs through the air, 1.68 TDs on the ground, and 1.05 field goals. These numbers are more adjustable. Ole miss defense struggled to stop the pass so I obviously adjusted the passing TDs up to 2. On the ground, just looking at the correlations, Ole Miss actually performed pretty well, but Bama is just another beast running the ball according to the correlations (6th in Yards per Rush). I also rounded this TD projection up to 2.

The field goal percentage is pretty obvious to keep at 1 since it is so close to the round number.

So with that, we have a final score of  Ole Miss 10/17ish to Alabama 31.

What’s next?

There is many more things to consider before we finalize our score adjustments. In the next post, I will go through the coaching model as well as other things that play into the projection.