I'm sure you all have found some sort of answer before that has left you more frustrated then when you started. This is certainly the case here.
[quote]
Hello Noah,
From investigating the game it looks like William Patterson had some pretty long plays that kind of inflated their stats. I know that some game results should to be more lopsided than they appear, but the flow of a game dictates much of a single games results.
I can run some test simulations to find out what the aggregate results would be from these two teams if think that would help explain the results better for you. Beyond that, it's difficult to find one attribute or one situation that made the game close. I'm guessing that the simulations would show the standard victory is much more substantial, but there are games that live outside of the standard deviation.
Also, it appears that your Running back had 153 yards, which doesn't seem like a pedestrian result. I'm sorry that the score of the game has changed your opinion of the new engine, but the changes have been largely bug related recently.
Please let me know if there is anything I can show you that would help alleviate the doubts you have.
Thanks.
[/quote]
153 yards on 36 attempts is pedestrian, based on the ratings I was going by, however his next ticket update (and I absolutely believe this is oriole himself responding) gives me an idea of just how off my (and frankly nearly everyones) idea of what is important is significantly off. I'm going to do some tinkering to see if I can reproduce the numbers in this ticket to confirm or deny a suspicion I have about ratings importance and 2.0 vs 3.0.
[quote]
Noah, I investigated this problem that you are seeing. I created a brand new test to allow me to input the exact teams into the system. So discounting game planning and formation IQ, I can run 10 tests games to see what the average score is between the two teams (given standard playcalling)
Looks like you win most of the time.
In the ten cases I ran you had a score of 39.15 (with a 7.86 stdDev) and he had a 14.15 (with a 8.6 stdDev).
Here are the scores for 10 games.
HomeTeam 48 41 40 27 43 52 30 41 26 40
AwayTeam 16 20 17 20 14 9 3 30 10 34
Also, some of his prowess can be associated with his relatively deep team
Here are the average rating differences by position.
QB RB WR TE OL DL LB DB
HomeTeam 39.29 45.87 43.67 45.21 45.29 42.37 42.17 47.69
AwayTeam 41.04 45.30 41.10 41.21 40.26 43.18 40.40 44.57
You have the better team, and it shows in the results, but he has a good enough team that once in a while he will give you a close game.
Hope that helps.
[/quote]
And all I will say is that those numbers are gamechangingly different then any number I have come up with in my testing. Any number that is in GUESS. Even if you are talking about the whole group of players those numbers (with a notable exception of DB's, which is HIGHER then any formula would show me) are lower then the lowest rated player in that group.
So now I know. I know that I have no freaking idea what goes into the engine. I guess I already had deduced that. When I was in beta my numbers had little difference to the engine numbers (and for those that didn't know we could see those in test game), and now there is a huge disparity.