College football teams are volatile as hell. Why don’t we incorporate that into our picks?

You have a different team every week. It is a Lou Holtzian truism that we are reminded of every single week of the college football season. A team beats a great opponent, then lays an egg against a lesser one. Good one week, mediocre the next, great the next, WTF the next.

When you rely on the whims of 20-year old males, you are going to find life rather maddening. Still, some teams are more frustrating and volatile than others. Some are all over the map, while others are reasonably steady. (“Steady” can be “steady good” or “steady bad,” mind you.)

My S&P+ ratings are designed to simplify. They are now presented in the form of an adjusted point total — No. 1 Ohio State’s plus-29.4 rating means the Buckeyes are 29.4 points better than the average team at the moment; No. 128 Charlotte is 21.1 points worse than average. This is designed to tell you that, on a neutral field, Ohio State would be expected to beat Charlotte by 50.5 points. And that’s fine. But the frequency of, say, a closer-than-expected 24-point win or an all-time, 77-point blowout is uncertain.

Because I am an endless tinkerer, I’ve been looking into ways to incorporate volatility (in this case, represented by standard deviation) into my S&P+ picks. I haven’t simulated as much as I need to to know my approach is going to work particularly well, but I thought I’d show my work and open up a conversation.

On the team stat profiles, I share a team’s adjusted scoring margin for a given game. It is intended to express, “Based on this game’s stats, you could have expected to win/lose this game by X points.” It is unadjusted for opponent. But if I actually adjust for opponent, I can come up with a pretty solid basis for a standard deviation figure.

If you’re curious, here are the standard deviations for FBS teams in both 2015 and 2016 (sorted by current 2016 average).

I’m guessing that it’s a bit too early for these to be reliable, and I haven’t finished setting up a simulation for 2015 yet, but like I said — we’re still in experiment stage. So we’ll press forward.

If we have an average (the S&P+ rating) and a standard deviation, we are in position to simulate. So what happens if we take every game on the Week 6 docket and simulate it 10,000 times? In theory, we can come up with different average and median scoring margins. We can also compare these margins to the Vegas spread and take a peek at how frequently Team A covers against Team B in these simulations.

See where I’m going with this?

We’re miles from any sort of “Take THESE Picks to the Bank!!” proclamations, but I’m going to be paying particular attention to the five games at the top (not including Florida-LSU, obviously):

• BYU at Michigan State
• Fresno State at Nevada
• Florida International at UTEP
• Ball State at Central Michigan
• Indiana at Ohio State

This is obviously based on the spread listed. Depending on the volatility of the two teams involved, a 0.5-point change in the spread could result in a change of anything between 0.4% (if both teams are crazy-volatile) and about 1.6% (if they’re not).

One week won’t tell us much, but I’ll keep trying to pursue this and see what I find. Coming up with an approximate “odds of covering” figure would be lovely, especially if Vegas and luck continue to be the jerk that they’ve both been so far this year. And if this does go somewhere, it will be fun to figure out the contributing factors to year-to-year volatility.