Someone hinted at this in the forums several weeks ago...

Is there some sort of an HD factor that limits the score from being a complete blowout in the first half? Except in only a couple of the 300+ games I've played, whenever I am down big early in the 1st, I am always assured of slicing into the lead significantly. This also happens when I start a game up big. Almost always, games are single digits with around 12 to 14 minutes left to play in the game. At that point it seems the real meaningful lasting action begins.

Is there some auto-factor that pulls scores more even?

I've heard rumors of this myself, but I honestly don't know the answer - so if someone does, I'm interested in learning if there is any truth to this as well.

Posted by tkimble on 4/20/2015 10:38:00 PM (view original):
Regression to the mean

Right, that is a part of it.
But someone here mentioned once that there was an HD auto factor that keeps scores semi-close (in addition to the regression to the mean)..

There is now an additional bit of code in HD that causes overperforming or underperforming players to regress toward the mean somewhat faster than would be dictated by normal statistical behavior - that is, if you're overperforming, you get downgraded a bit, and vice versa.  But that would only be significant if the expectation values dictate that the game should be close, and there's nothing on a team basis, only at the player level.  So if you have fairly evenly-matched teams get off to a very uneven start, yeah, odds are, it's going to come back.  And that's no longer governed exclusively by statistics, but also by some HD code.  But if your team is much better, you can just keep pulling away.

Well, the mechanism that exists (to my knowledge) is not to keep scores close, but to strengthen the regression back towards the mean if there are prolonged extreme results for players.

ninja edit: dahs beat me to it by a few seconds.  

I'm not sure I'm a big fan of HD artificially strengthening the regression faster. Let it happen naturally, otherwise teams almost have to beat teams twice... get off to a great start, have the game artificially pull the game closer, then try to maintain your not-as-large-as-it-should-be lead...

Posted by npb7768 on 4/20/2015 9:20:00 PM (view original):
Someone hinted at this in the forums several weeks ago...

Is there some sort of an HD factor that limits the score from being a complete blowout in the first half? Except in only a couple of the 300+ games I've played, whenever I am down big early in the 1st, I am always assured of slicing into the lead significantly. This also happens when I start a game up big. Almost always, games are single digits with around 12 to 14 minutes left to play in the game. At that point it seems the real meaningful lasting action begins.

Is there some auto-factor that pulls scores more even?

I was the one that mentioned it (unless someone else also did).  Llamanunts said he had noticed the same thing.  Don't care what anyone else says I've seen it happen way, way too many times to just have been a fluke game here or there.  It's in the code and it shouldn't be.  Let the ratings dictate what happens, for the ENTIRE game.  Don't artificially manipulate just too keep scores close or keep numbers confined to certain ranges.  If I'm playing a bad Sim team and his guys aren't good enough to stop me from scoring or to stop fouling then let me run up the score and foul every one of his players out.  The last five that were still playing will come back anyway and for every foul after that, it'll be shots AND a technical, but you know what?  So what.  If it's that big of a mismatch, so be it!

I'm pretty sure it used to be like that a long time ago (well before I joined), and it was changed because there were too many bizarre upsets. 

It is kinda frustrating though in some cases. I feel like there are some teams where every time I play them, I have a double-digit lead at halftime and then get outscored by 20 in the second half. The sort of games where your opponent is really strong but you have the better team. I feel like in those games, if I do too well in the first half, I'm doomed to get crushed in the second. Especially when I have a pressing team (because there are enough possessions for that quicker regression to mean a lot). 

Posted by emy1013 on 4/21/2015 4:55:00 AM (view original):

Posted by npb7768 on 4/20/2015 9:20:00 PM (view original):
Someone hinted at this in the forums several weeks ago...

Is there some sort of an HD factor that limits the score from being a complete blowout in the first half? Except in only a couple of the 300+ games I've played, whenever I am down big early in the 1st, I am always assured of slicing into the lead significantly. This also happens when I start a game up big. Almost always, games are single digits with around 12 to 14 minutes left to play in the game. At that point it seems the real meaningful lasting action begins.

Is there some auto-factor that pulls scores more even?

I was the one that mentioned it (unless someone else also did).  Llamanunts said he had noticed the same thing.  Don't care what anyone else says I've seen it happen way, way too many times to just have been a fluke game here or there.  It's in the code and it shouldn't be.  Let the ratings dictate what happens, for the ENTIRE game.  Don't artificially manipulate just too keep scores close or keep numbers confined to certain ranges.  If I'm playing a bad Sim team and his guys aren't good enough to stop me from scoring or to stop fouling then let me run up the score and foul every one of his players out.  The last five that were still playing will come back anyway and for every foul after that, it'll be shots AND a technical, but you know what?  So what.  If it's that big of a mismatch, so be it!

it does emy... if you are playing a bad sim team and his guys are awful, it does let you run up the score. that's not what seble changed. there is nothing in the game that pulls the score closer together. there is something in the game that pulls towards the mean result. this works for you just as often as it works against you - actually, in your case, it works for you a lot more than it works against you. the worse team is less likely to pull upsets because of the change, that is the consequence, not to keep the score close.

now, when teams are evenly matched, the indirect effect is to keep the score close, generally speaking. i suspect the reality is that people are watching the games more closely when its a close game, not over analyzing the results when they are crushing crap sims or getting beat down by killer teams when they have down teams. i think this is causing a perception problem, where folks are mostly watching the close games - and thus, are mistaking the reversion to the mean effect as a "keep the scores close" effect, missing the cases when playing a 40 pt underdog sim who is even early in the game, and then quickly gets to down 20 because that is what the reversion to the mean is pushing for. if the mean is that you will foul out the entire other team (it virtually never is), then that is even more likely to happen today, than before seble's change.

so, im not saying you are seeing a ghost, seeing something that isn't there. but you are seeing something and taking it for something it isn't. im not suggesting i support the regression to mean (feedback) code either, i'm just trying to make sure its clear what the real effect is, so people can rail against the appropriate issue :)

The piece of propped-up artificial performance in the engine that I would like to get rid of is the increased likelihood that a losing team will complete 3pt shots in the last minute of the game. I cannot prove that this is empirically the case, but I have utterly convinced myself that it is a real effect.