• Welcome to OGBoards 10.0, keep in mind that we will be making LOTS of changes to smooth out the experience here and make it as close as possible functionally to the old software, but feel free to drop suggestions or requests in the Tech Support subforum!

Which Team Will Finish Last in the ACC? CORRECT ANSWER: BOSTON COLLEGE

In this year's ACCT, the 12 seed will be...

  • Total voters
    142
  • Poll closed .
Haros said:
Answer the following question phrased slightly differently than before: Would you be more impressed with a model that gives teams a 60-65% chance of winning won a) 63% of the time of b) 80% of the time? Simply A or B will suffice.
Click to expand...

RJKarl said:
Make no mistake if he hit 70+% of his 50-55% games, I'd be very impressed.
Click to expand...


I think that answers your question, Haros.
 
Haros said:
Except the purpose of his site isn't to predict individual games. He only started posting full team schedules with projected outcomes a couple years ago as an added feature. Nor does he say you can make money in Vegas using his rankings (nor is that the purpose).

Answer the following question phrased slightly differently than before: Would you be more impressed with a model that gives teams a 60-65% chance of winning won a) 63% of the time of b) 80% of the time? Simply A or B will suffice.
Click to expand...

If "a" purpose of his site isn't to predict overall outcomes (this thread is not about a single game between two teams. it's about a range of games between between numerous teams), why does he play the parlor game of saying Team X will finish with this record?

If the "probability" his formulae has determined is Team X being 4-0; Team Y being 3-1 and Team Z being 0-4 and the results are X is 1-3, Y is 0-4 and Z is 2-2, then his methodology for saying this is probable is wrong.

In fact by having the data of about 90% of the season for each of the teams and each of their opponents, your "probabilities" should be within a reasonable margin of error or you shouldn't be presenting them.

I realize the theoretical answer is A but the reality is you don't need Kempom to pick a game that is 60-90%.
 
The goal is not to pick the game, the goal is to project the finish of the team and where the team stands as the season ebbs and flows. Teams get better, teams get worse, and the goal is for this to be captured by KenPom. He is not trying to pick a winner or a loser, he is trying to pick the probability of a team winning or losing. This is why people continue to ask you RJ which option you would rather have between A and B as Haros threw out.

KenPom wants to try to get the probability that Team A beats Team B, so even with Duke's 11% chance to beat us, he would want Wake to win 11 times to ensure that the formula is close to correct. The 11 times that Wake does win does not mean that his high probability of having Duke win was wrong, it means that his formula is closer to being correct.

The miscommunication that I think is occurring is because most people will say "oh there's a 90% chance we'll win that" to mean that "we should win and if we don't there should be some sort of explanation for why". KenPom is just attempting to quantify games in a system where 90% will quite literally mean 90% and not just as a colloquial meaning for "we should win this game beyond a doubt".
 
ACC Tournament KenPom Projections (as of 2/27):

Thursday

Noon: (8) Maryland v. (9) Virginia Tech (winner plays 1 seed Duke)
2:30: (5) Miami v. (12) Boston College (winner plays 4 seed Virginia)
7:00: (7) N.C. State v. (10) Georgia Tech (winner plays 2 seed UNC)
9:30 (6) Clemson v. (11) Wake Forest (winner plays 3 seed FSU)

This has Wake losing to Georgia Tech and then losing the tiebreaker (tied head-to-head, GT with the best conference win), State and Clemson tying for 6th (with Clemson winning the tiebreaker in head-to-head), Duke beating UNC again (and winning the tiebreaker for 1st head-to-head), and UVA tying with Miami for 4th and winning the tiebreaker (head-to-head).

Georgia Tech plays the worst two teams in the conference in their final two games though so they could in theory make it up to a tie for 9th with VPI at 5-11, but would still lose the tiebreaker and finish 10th. I think more realistically Wake beats GT, GT beats BC, VPI wins one of two against Clemson/State and Wake ends up playing Maryland at noon, VPI plays State in a rematch of their final game of the year in the 7 game, and then GT gets Clemson in the 9:30 game. BC will finish last.
 
Last edited:
Wake can't finish better than 9th even if we win out, and it's extremely unlikely that we finish worse than 11th. Given the tiebreakers I think it's even difficult for us to finish 10th. I'd give odds: 55% 9th, 45% 11th
 
simpsons.gif
 
Wakeforest22890 said:
The goal is not to pick the game, the goal is to project the finish of the team and where the team stands as the season ebbs and flows. Teams get better, teams get worse, and the goal is for this to be captured by KenPom. He is not trying to pick a winner or a loser, he is trying to pick the probability of a team winning or losing. This is why people continue to ask you RJ which option you would rather have between A and B as Haros threw out.

KenPom wants to try to get the probability that Team A beats Team B, so even with Duke's 11% chance to beat us, he would want Wake to win 11 times to ensure that the formula is close to correct. The 11 times that Wake does win does not mean that his high probability of having Duke win was wrong, it means that his formula is closer to being correct.

The miscommunication that I think is occurring is because most people will say "oh there's a 90% chance we'll win that" to mean that "we should win and if we don't there should be some sort of explanation for why". KenPom is just attempting to quantify games in a system where 90% will quite literally mean 90% and not just as a colloquial meaning for "we should win this game beyond a doubt".
Click to expand...

But each game is only played once. If we win tomorrow, then something he said would happen once out of ten times (approx) will have happened on the FIRST try. That's very unlikely, and makes us think that he probably had his percentages wrong. I mean, would he use the same argument if he had said Duke had a 99.999999% chance of winning? He could still say that our win made his calculation "correct," i.e., "See? Wake won the one game out of 100 million that I said it was likely to win. That's why I didn't say 100%, silly!" But it's so incredibly unlikely that a 1/100,000,000 chance happens on the first try that the more logical conclusion is that he had his percentages wrong.

Or maybe I'm misunderstanding. Are you saying that, in the aggregate, he wants a 10% team to win 10% of the time? Like, if there are ten games tomorrow that he put 90% odds on, he wants the underdog to win exactly one of them? If there are 1000 games this season that he put 90% odds on, he wants the underdog to win in exactly 100 of them?
 
Wakeforest22890 said:
Wake can't finish better than 9th even if we win out, and it's extremely unlikely that we finish worse than 11th. Given the tiebreakers I think it's even difficult for us to finish 10th. I'd give odds: 55% 9th, 45% 11th
Click to expand...

So you're saying that you think we'll definitely finish 9th, right?

/rjtrollface.jpg
 
ProfessorDeac said:
But each game is only played once. If we win tomorrow, then something he said would happen once out of ten times (approx) will have happened on the FIRST try. That's very unlikely, and makes us think that he probably had his percentages wrong. I mean, would he use the same argument if he had said Duke had a 99.999999% chance of winning? He could still say that our win made his calculation "correct," i.e., "See? Wake won the one game out of 100 million that I said it was likely to win. That's why I didn't say 100%, silly!" But it's so incredibly unlikely that a 1/100,000,000 chance happens on the first try that the more logical conclusion is that he had his percentages wrong.

Or maybe I'm misunderstanding. Are you saying that, in the aggregate, he wants a 10% team to win 10% of the time? Like, if there are ten games tomorrow that he put 90% odds on, he wants the underdog to win exactly one of them? If there are 1000 games this season that he put 90% odds on, he wants the underdog to win in exactly 100 of them?
Click to expand...

I'll ignore the first paragraph. Your last paragraph is correct. If there are 100 games with a team having a 90% chance of winning, the underdog will win 10 of them (10%). At 70%, the underdog wins 30 of 100. Individual game outcomes matter only so much as they form the aggregate population. He doesn't care whether Wake specifically beats Duke, (or even whether his model is correct on individual games), but he back tests it to see if/where it is off so he can improve it in the aggregate.
 
Wakeforest22890 said:
Wake can't finish better than 9th even if we win out, and it's extremely unlikely that we finish worse than 11th. Given the tiebreakers I think it's even difficult for us to finish 10th. I'd give odds: 55% 9th, 45% 11th
Click to expand...

It's actually fairly possible we finish 10th. If BC beats GT, and then GT beats us, there would be a 3-way tie at 4-12. We would win the tiebreaker (3-1 aggregate record) and get the 10 seed.
 
Thanks, Haros. That makes sense. I was just being too literal with what Numbers was saying ("would want Wake to win 11 times") instead of what he meant ("would want 11% teams in general to win 11% of the time"). I should have been able to figure this out without asking. Tired today.
 
Ellen Kirkman better not read this thread. Her head would explode.
 
ProfessorDeac said:
Thanks, Haros. That makes sense. I was just being too literal with what Numbers was saying ("would want Wake to win 11 times") instead of what he meant ("would want 11% teams in general to win 11% of the time"). I should have been able to figure this out without asking. Tired today.
Click to expand...

Yeah sorry I meant he would want the "Wake" teams to win 11% of the time. He aggregates all the data.
 
You must log in or register to reply here.
Back
Top