Short version: Previous versions of Football Mogul would cause top players (and top draft picks) to tank in the off-season at an alarming rate. This effect has been reduced by about 85% in Football Mogul 19, making the game both more realistic and less frustrating.
Long version: Both Baseball Mogul and Football Mogul include code that models the real-world phenomenon generally refered to as regression to the mean: if a player puts up incredible numbers in one season, he is extremely likely to post a performance in the following season that is closer to the league average (aka "the mean"). One example: the #1 wide receiver (in terms of fantasy points) will lose, on average, more than 30% of their value in the following season. Some of this comes down to luck: bad weather, lucky bounces, missed tackles. But regression to the mean also occurs at the underlying level of raw ability. The league MVP is more likely to get worse than to get better, and this applies to a lesser extent to all superstar players. However ... I'm writing this post to admit that I made a mistake when implementing this phenomenon in previous versions of Football Mogul, and that I was finally able to fix it for Football Mogul 19. (FYI: it was a simple math error/typo that I didn't notice until I stepped through the code and realized that it wasn't performing the calculation I'd described in the comments). You don't have to take my word for it. This year's demo version lets you access "Commissioner Mode" so you can see how player ratings progress in the off-season. And if you see results that you still find unrealistic, please let me know. I appreciate constructive feedback and take seriously any suggestions or criticism backed up by data and a logical argument. Thanks for reading! Clay
2 Comments
We received the following post from a frustrated Football Mogul player this week: I understand his frustration. Compared to baseball, football players are notoriously difficult to evaluate because there aren't enough stats. EA has a full-time expert dedicated to the job, and that's just for the current season's players.
I've been working for years on a new method to rate historical players, and I finally got it working for Football Mogul 19. In a nutshell, team performance is divided among the players on offense or defense for any given game, not unlike the work done in this paper. In previous versions of Football Mogul, Deion's Pass Coverage rating in 1996 would be brought down by the fact that he had only 2 interceptions in a full 16 starts. However, he accrued these starts with the '96 Cowboys – one of the best defensive teams in the league. Thanks to the magic of computers, I wrote code that loops back to evaluate Sanders according to a combination of his stats (such an INTs in other seasons) and the defensive success of the teams he played on, then folding the player ratings into the evaluation of each team and so on – until the ratings stabilize. In the final database, I ran this loop overnight (more than a thousand times) until every rating was stable within 0.06 points. Anyway, I'm attaching Deion's 1995 Scouting Report from the Football Mogul 19 beta. I think it's a LOT better than last year's version, and I'm really excited about these new historical seasons. |
Archives
September 2023
Categories
All
|