Quote:
Originally Posted by USF
Doesn't the AI player rating evaluation in the league settings dictate this in some way. If it is skewed to stats then that will apply however, if it is skewed to ratings then that will apply if it is 50/50 stats/pure ratings then its even. ??
|
as far as i can tell, ai eval only affects Overall, not the individual ratings. easy to verify, load a throw-away league with 100% accurate ratings and set up some extreme AI evaluation including lots of stats... compare contact (etc) to the profile editr value. convert to your scale, it should never deviate even when overall starts adjusting due to recent results.
if you are talking about splits and the pow example i gave... yeah it's supposed to give you the 'average' of l/r power... but it's just one more way that scouting inaccuracy works with a trend.
e.g. when a player has a HUGE split and the higher split is vs righties, i think they are more often to show a contact that's closer to the upper split rating as opposed to a proper 'average' releative to proportion of r/l etc... (in other words: as it normally would with 100% accuracy) edit: i just had a guy with a split of 100 / 180-some for power... it showed he was 80/80 power even though he's closer to 70/80 due to how bad he is wiht lefties or whatever it was. it basically ignored the left split (this can work with a low value too of opposite split.. i think it's beacuse it's 75% of what the scout sees, so that's the more likely scouting error - as in they use the righty and not understand teh vs lefty portion built into scouting incaccuracy and how it is alotted)
relative to hotstreaks and such: if you like tinkering, i'm not telling you to stop. maybe somethign is in the game that's not advertised, but if they use the laws of probability for the basis of their logic, it very likely is not helping or hurting with any real correlation, because it's completely unpredictable when they start or finish - let alone knowing when they are happening in RL (in the video game they give the little fire icon, but again that is due to recent results not that they are going to continue it.
the ratings don't chagne, so i doubt they are actually 'different' during that time period. (beyond the influence of modeled things like morale and personality and their small effects built-in. these are subtle effects not big enough to cause somethign on their own)