The Intuiative side of Regression debate?
homerplayer
Senior Member
Consider myself above average on the math side of things. Lights years ahead of the magical day I found BT oh so many years ago, but still with much to learn and certainly not on the same level of others that think in numbers. Still stray off the reservation now and then.
See myself as a bridge builder, a divide conquerer as it were. So hear me out, because there is often tension between the math and non math guys (non math in the sense of viewing this endeavor as requiring more than just math to win). But the camps aren't that different.
Question: on a rudimentary level, is the phrase "this ________ is due" an intuiative way to say "this _______ has under/over performed and I expect a regression to the mean"?
That is an oversimplified example for illustrative purposes only. I would never endorse that train of thought like that guy that said his system (system sucks people) was due cuz it was never under 33% and it was like 15 games. But on a basic, fundamental level, perhaps the non math guys have an inkling, but not a name. The camps use different terms. Thoughts? Can we hug it out?
See myself as a bridge builder, a divide conquerer as it were. So hear me out, because there is often tension between the math and non math guys (non math in the sense of viewing this endeavor as requiring more than just math to win). But the camps aren't that different.
Question: on a rudimentary level, is the phrase "this ________ is due" an intuiative way to say "this _______ has under/over performed and I expect a regression to the mean"?
That is an oversimplified example for illustrative purposes only. I would never endorse that train of thought like that guy that said his system (system sucks people) was due cuz it was never under 33% and it was like 15 games. But on a basic, fundamental level, perhaps the non math guys have an inkling, but not a name. The camps use different terms. Thoughts? Can we hug it out?
Comments
I'd say they are closer to equivalent statements if you edited it to something like "this ________ is due in the long run" because the common phrasing of something being due usually refers to the short term/immediate future.
Both camps are at fault as well. Math guys sometimes just tear into people. Some non math guys never want to even listen though.
I'd say that this statement shows that the person has no understanding of regression.
It's not that a "correction" is expected, it's that you expect the long term results to be in line with expectations, and if the short term results are skewed one way or the other, they should naturally move back towards the expectation as the sample size increases.
I think the coin flip example often presents the easiest way for people to understand this stuff. Say that you've flipped a coin 10 times and have 10 straight heads. If you're going to flip it 90 more times, the Expected Value of heads over the next 90 flips would be 45. Heads is at 100% right now through 10 flips. It will regress towards 50% with the more flips that you have. If you're going to have 20 total flips, you won't get close to 50% (most likely). If these were the first 10 flips in a series of 1 million flips, chances are that you'll get a lot closer to 50%.
But I don't see anything intuitive about saying that tails is now "due" since we've started out with 10 straight heads. Tails is no more "due" than it would be if we started out 5-5, or started out with 10 straight tails.
As Tommy pointed out above, most people just don't understand the concept of regression to the mean. Once they do, there's not a whole lot to discuss.
Like a +5 dog is up 5 at half, and the favorite is -4 at half (+1 in game now), do we play the fave cause we think it'll regress to the mean of -5?
Cause you get lots of CLV sometimes or maybe this is a dumb question cause every game is different.
Basically this is a messy way of asking, are the 1st and 2nd halves independent events?
I'll try to say this as nicely as possible...this line of thinking is way off. You're not getting "CLV". Beating the closing line is a lot different than playing into an adjusted line based upon the results of the 1st half. Regression has absolutely nothing to do with betting 1st halves vs 2nd halves.
Lot more factors into a 2nd half wager then simply getting back to the FG line . (common error)
Many people view it as a great value or maybe even CLV, but it is actually a new wager with new factors. One has nothing to do with the other.
You try typing this shit on a coopertino iPhone.
LOL. I was going to post something like this but then I realized my spelling is the worst.
Sorry about that, forget you guys don't live inside my head so I don't write out every little detail, which sometimes proves detrimental to the process. Point is instincts are already leading the non math guys down a path toward the concept. The debate part is how do we bridge the gap? How do we get the math guys to patiently explain the concept (short of a sticky) time after time and how do we get the non math guys to understand it? Isn't a terribly hard concept, and as I asserted multiple times, we have a language barrier to overcome.
TL-appreciate the input as always. Not a fan of the coin flip example in this regard as I was trying to apply it to sports betting. So when I simply write about a team under/over achieve, I trust readers to take understand the full meaning, which again, is my fault. Used correction there cuz I didn't want to use expected twice in a span of 2 sentences, damn creative writing courses from high school!! But these events I am referring to aren't 50/50. Not a 50/50 change Verlander wins when takes the mound. Not an50/50 chance Brady is going to lose or throw for 150 yards when he lines up under center.
I meant to say, but got lazy and typing on these devices super sucks, we collect this data, we compare a teams short term pefofrmance against what we believe the true performance should based on observation of historical performance to this point. When good teams suck its a slump. Shit teams win they're hot (not my terms, for illustrative purposes only).
In this context, non math guys say due. I remember sixth sense defending the use of regression on forecasting turnovers.
On the whole I feel BT could be even better than it already is, and it is awesome. The debate goes back to how do we get the two sides to communicate without pissing contests ensuing?
Unless I am missing the intent of BT. Sharp forum? Hell yes. Kind and nurturing to noobs? Sometimes yes, sometimes no.
The idea that we believe there is more to sports wagering then a formula is incorrect ?
Just because some of us believe in more then math, does not mean we don't understand math or probability on a wager.
Rito, check your email
I am being sincere when I say this: This attitude will carry you a long way. I wish you the best of luck in the future my man.