Flaws Found When Relying on Game Simulators When Making College Football Picks

David Lawrence

Wednesday, August 26, 2015 11:17 PM GMT

Wednesday, Aug. 26, 2015 11:17 PM GMT

The Predicalator sounds great, doesn’t it? Well, not the cheesy name but the fact that it can run 50,000 simulations for each game and season, and accurately predicted the final four BCS Playoff teams at the beginning of last season. And the fact that it has hit 69% winners on it’s Lock Of The Week and went 24-8 in the last two seasons. However, at a closer look, there are some serious flaws with the system that you’ll want to take into account.

A Sports Betting Algorithm: This Isn’t A New Thing
Let’s face it: betting systems, algorithms and projections are not a new thing. Schools like MIT have been around for a long time and math geniuses have been, well, geniuses for quite some time too. To think that somebody has finally cracked the code would probably be a mistake. Has technology gotten better? Yes. Are we in a new age of sharing on the internet? Yes. But we’ve seen things like this before with WhatIfSports, AccuScore and a whole host of other algorithms. They are usually just another tool you want to incorporate in your handicapping but not something you’ll rely upon fully when it comes time to place your College Football picks.


Systems Don’t Take Reality Into Account
For the most part, when we’re referring to “reality” here, we really mean injuries. If you look at last year’s Predicalator projection, it projected that Ohio State, Florida State, Alabama and Oregon would be the last four teams standing. However, three of those teams didn’t really have to deal with significant injuries and the one that did – Ohio State – turned out to have unheard of depth at the quarterback position. It’s a system and an algorithm, so it’s not going to be able to adjust for injuries very well. On top of that, since it relies on stats a lot, it will be hard for it to factor in freshmen’s who emerge on the scene since they don’t have a track record.

While the Predicalator is quick to sing it’s praises on the four finalists it had right, keep in mind it was way off on some of its other season-long predictions. For example, the Memphis Tigers was 10-3 overall even though the machine predicted just 4.5 wins. They also predicted that Houston would win the AAC but they were fifth. And they also had South Carolina as the second-best team in the SEC yet they were fifth-best in the SEC East alone. Here’s another area where they were way off:

PredictionMachine Projected Big 12 Results

Actual 2014 Big 12 Result

1. Baylor

1. Baylor

2. Oklahoma

2. TCU

3. Texas Tech

3. Kansas State

4. Kansas State

4. Oklahoma

5. Texas

5. Texas

6. TCU

6. West Virginia

7. Oklahoma State

7. Texas Tech


Stats Can Be Presented In Many Ways
And here in lies the problem with any type of tout, service or system: they always seem to highlight their wins and hide their losses. Yes, Locks Of The Week look pretty good (hitting at 69% over the last five seasons) and the correct prediction of last year’s BCS Playoff was impressive, but you can see from above that they seem to overly highlight the wins and hide the losses a little bit.


If It Is Accurate, The Value Will Be Gone
The last piece of info that’s important to note here is that if the system is in fact that good, then the sportsbooks will start to adjust. The best value is for the person who first gets to the line but if the books start to shade the lines, then the value will start to drop off. For example: if a certain prediction via the system is a pick at College Football odds of -6.5 but you only got to the line at -7 or -7.5, then the value is gone and your win percentages won’t be as good as the system. That’s a huge problem and it’s something you need to take into account.

comment here