Posted by bhouska on 3/1/2012 1:49:00 PM (view original):
A lot of feedback we get is that you'd like to know why a certain play happened. You can only go so far with "why?" in answering why a play resulted in a certain outcome. Take, for instance, a play where a RB gets 2 yards on the play and is tackled by a LB. Let's look at an identical play where the RB gets 20 yards because the LB did not tackle him. Why did he tackle him on one play and not the other? Both of their ratings remain the same and both situations are identical. Sometimes the only answer is that in one case he tackled him and in the other he didn't. I don't know how much detail anyone would want so far in wanting to know that he didn't wrap him up properly versus he was just out of reach. Even if you go to that much detail it still boils down to some random chance at some point. I think there is at least more we can give on plays to help evaluate your team and players. If it's something that can help identify weak links on the team or see the effects of your star player, I think that would be great. With the engine changes, we should be able to provide a little more analysis like that.
Personally, I think people are asking for "why" a play turned out as it did because of the apparent utter randomness of the new engine. If I recall correctly, we never had this complaint with GDI, it could be easily gathered by looking at the attributes why something happened the way it did. I don't ever remember being flabbergasted by a game with GDI like I have been many times with GDII. So, people are reaching for answers, hoping that a better breakdown of the play will help them understand why something happened the way it did. From what I am gathering that you're saying is that really can't be done, that more info about a play won't really tell us anything, to which I say more cause/effect in GDIII is what we want/need. Right?
Here's what I believe 95% of us as coaches would like to see in a simulation-based football game....if one team's OL is head and shoulders above their opponent's DL and they have good RB's they should be able to break off chunks of yards at a time. If that game averages over the "average" of 4.321 yards per carry or whatever the number is that GDII engine tries to get each game to, so be it.
If one team's WR's are head and shoulders above their opponent's DB's, they should be open all day. As long as they have a quality QB, they should be able to put up monster numbers against inferior teams.
It all boils down to cause and effect. We want to be able to see it and react to it.
Again, one misconception about the current engine is that it tries to force results to some number. It doesn't. Where the average numbers come in, like 4.321 YPC, is when checking an average team playing an average team 1,000 times. The average results should be average real life results. But in each game, the engine doesn't try to force a team to meet those averages, so if you are trending above 4.321, it doesn't start pulling your rushing down.
Why a team that is head-and-shoulders above another team ends up not running all over the team is because each individual check allows for all possibilities. So let's say the chance the RB is stuffed goes from 10% to 6%, which in real life would be average versus very good. If you look at 10 plays strung together, with a random 6% chance each play, it can still end up that more than 1 play results in a stuff. So the idea that a RB would get over the 4.321 YPC is left up to pure chance. A great RB has a much higher chance to get above the average, but he still has a chance to get below average.
We also can't allow your head-and-shoulders WRs to catch every pass. It would look ridiculous to see a QB hit 50 out of 50. So why would one of these passes be incomplete? We should be able to determine what the expected outcome should be. In this case, it should be a reception. It could be incomplete because the QB overthrows, or the WR runs the wrong route or drops the ball, or maybe the DB steps up and makes a great play. What we wouldn't be able to do is tell you why the WR dropped the ball or why the DB made the great play on that particular play, but we should be able to tell you why the ball was incomplete. But at what point is the number of dropped or overthrown balls unacceptable? Right now, you have to assume when you see a pass was incomplete that one of these things happened. Is it more acceptable to see these incompletions as a mix of these reasons than just a bunch of "incomplete pass" plays?
One of the challenges of the update will be to allow this sort of analysis of plays. The other challenge in trying to rein in the wide range of possible results, especially between a higher rated team playing a lower rated team, is in determining what the limits of the plays should be. Every pass can't be a reception and every run can't be a TD. We should also define what is "head-and-shoulders" above the other team. What rating difference is head-and-shoulders above and what is evenly matched? What is everything in between? I feel like there needs to be some constraints on the current randomness, but there will always need to be some randomness. The problem is how to make it acceptable.