I've been doing some thinking about proper wagering recently. I used to think I was a volume player and that anywhere I thought I had the slightest of edges I should play. In other words, I had a difficult time figuring out which plays would hit and which wouldn't (and yet, I had a 1-5 sliding scale for my plays... intelligent). Lately, if you haven't noticed, my cards have been about 60-70% of other contrarians out there (well, except Jonny, for obvious reasons).
Some reasons that a game might get trimmed: not anti-public across all sites, more than a point of difference between Pomeroy and the line (like UTEP tonight), or significant forward line movement. I may overlook one of these reasons if the other two are strong (like Fake Miami tonight), but overall, I think this is a better wagering strategy. Either that, or if I really have no skill at choosing the better sides, I'll be decreasing my variance. Obviously, I'll be playing less games, but I feel confident I'll still have more than enough action.
Hopefully, I'm not too far behind the curve and I'm not posting the obvious here.
For tonight:
7p West Virginia -17 -107 2x
7p Villanova +3 +102 2x
7p Miami (OH) -16 -103 3x
730p Nebraska +4 +103 3x
8p Loyola-Chicago +4.5 +104 2x
8p Wichita State +5.5 +107 2x
9p Kansas State +3 -105 2x
Passes are pretty much everything on the leans list below. UTEP, Provo, Charlotte, and SDSU were the hardest cuts.
Good luck tonight.
Subscribe to:
Post Comments (Atom)
13 comments:
Did you pass Provo just because of the Pomeroy score?
Yup.
FWIW, I'm not entirely convinced what I am thinking here is right, but I am feel more comfortable with the strategy.
FWIW, and I'm only being harsh here because I know you can take it, I think that's a pretty awful reason.
EVERYTHING else points to it being a play. On its face, from a square perspective, the line makes no sense. This is confirmed at every consensus site. Syracuse is overrated; #15 in the polls, #33 in Pomeroy. Syracuse has been worse than their record, 66th in Luck in Pomeroy. Providence has been better than their record, 238th in Luck in Pomeroy. I just don't see how a small discrepancy in the projected score, when we're not even into February, can possibly outweigh all of that.
I haven't fully fleshed out in my head how to handle the spread when it is one like this. Should I be counting it as one off or two off (since Pomeroy won't forecast a tie)? For consistency's sake, I left it off the card.
I agree with everything you wrote up there. I don't mind constructive criticism either, because you don't learn without it.
A couple of points here:
1. I've been wondering if I've been gambling scared, not unlike Squeeky in football in 2007. If I am, that's ok for now, I think. I had a terrible 2008 football and I am willing to sacrifice some EV for the decrease in variance.
2. In your last sentence you mention that "we're not even into February." At what point do you start valuing the predictive skill of those numbers? In my mind, ~20 games is a large enough sample for the predictions to carry value, but you've done more with this stuff than I have.
3. Sure, Cuse has been lucky and Provo hasn't. That won't have any effect on the prediciton. Unless you are talking about public perception, which I totally agree is on Cuse, but not the reason I passed the game.
Does all of this make sense? In this particular case, I'm fine being wrong (if I am), but in general, I'm ok with dropping plays that seem more marginal to me.
"Should I be counting it as one off or two off (since Pomeroy won't forecast a tie)?"
Well it could be anywhere between 0.01 and 1.49, so the average of that if you want to get technical I guess.
"At what point do you start valuing the predictive skill of those numbers? In my mind, ~20 games is a large enough sample for the predictions to carry value, but you've done more with this stuff than I have."
Of course they have value. That's not the question here. I just thing, considering all the other information available, they don't have nearly enough value to take the game from a definite play according to every other factor to a pass.
"Sure, Cuse has been lucky and Provo hasn't. That won't have any effect on the prediciton. Unless you are talking about public perception[...]"
I am talking about public perception. I think the Luck category is extremely useful, and I'm pretty sure I'm not the only one of us who looks at it for pretty much every game. I think there is a good argument that it's more useful than the prediction itself right now.
I think there is a good argument that it's more useful than the prediction itself right now.
At the risk of looking stupid, why?
I just thing, considering all the other information available, they don't have nearly enough value to take the game from a definite play according to every other factor to a pass.
At what point would it take you to move a game from a play to a pass. I admit that using a point is a pretty fucking arbitrary cutoff For example, if the prediction had Cuse winning by 3, would it change your opinion? 5? 10 (not that it would happen)?
"At the risk of looking stupid, why?"
Since when are we trying to handicap games? That's what you're doing when you put that much stock into the prediction. We are trying to gauge public perception and figure out which side has value. That is why I prefer Luck.
"At what point would it take you to move a game from a play to a pass."
Would've played at Cuse by 3. Probably at 5, too, if everything else was the same. I do not know about 10, since it would be such a weird situation. Probably not, I guess.
Since when are we trying to handicap games? That's what you're doing when you put that much stock into the prediction.
Really? I'd agree with you if I was basing this off Seth Davis' Friday column, but these are quantitative predictions based off sound mathematical theory. I can buy that I like shiny numbers more than the average person, but I'm not sure that is altogether a bad thing.
We are trying to gauge public perception and figure out which side has value. That is why I prefer Luck.
I agree 100% with this.
In figuring out which side has value, I would think that using a quantitative pythagorean predictor (along with market inefficiencies shown by the difference between the books' perception and the public) aids us in finding that value.
Again, I'm not saying to completely ignore the predictions. I'm just saying they shouldn't be of primary importance. I think it's silly to have hard and fast rules in the first place (precisely because of situations like this, when everything points one way and the predictions point the other), and even if you want to, one point is far too strict.
I think it's silly to have hard and fast rules in the first place
Sure. These aren't hard and fast rules, and I noted as much in the post.
and even if you want to, one point is far too strict.
This conversation has made me change my opinion. Obviously, this whole change in strategy is a work in progress.
fantastic discussion.
from using similar predictive numbers as Pomeroy but in the NBA, i think 1 point is way too small of a cutoff number if everything else looks good.
"Sure. These aren't hard and fast rules, and I noted as much in the post.
Fair point. I was just referring to your answer in my initial comment. "Hard and fast rules" was a poorly chosen phrase. What I mean was to not have a cutoff of one point or anything, but to consider the magnitude of the difference just as you would the magnitude of consensus #s, etc.
What I mean was to not have a cutoff of one point or anything, but to consider the magnitude of the difference just as you would the magnitude of consensus #s, etc.
That's precisely what I am trying to do. The bins I've been using when I put together the lines/differences were 0.5-1, 1.5-3, >3.5. This is clearly too strict.
This conversation was beneficial in helping me re-adjust my ideas. Thank you.
Post a Comment