Quick NCAAB question

[Justin7]

In men's Div I NCAA baskets, how often does the favorite win outright?

[Timmah2483]

I heard it awhile back, I think its 89-91% of the time. but I know early in the season that number is more like 70-75%

[Emily_Haines]

I show 74.8% over the last three seasons.

[Nicky Santoro]

how will this stat help in any way?? if it's 75%, it won't mean anything because a lot of those are -19 pt favs, and -15.5 pt favs and -9 pts favs.. they will win so often.. it's the -1.5 and -4 favs that will lose more frequently.

to make this stat useful, you should find out what % -4 pt favs win.. what % -2 pt favs win, etc. etc...

[durito]

2005-2010

Favorites (-1 to -42) are 9603-2894 Su 76.8%

[Pancho sanza]

74.33 %

1996 - present

[Timmah2483]

then what am I thinking of 89-91%.....

I bet its home court favorites.

[Justin7]

Thanks for the help.

I guess picking the su winner 70% on all games isn't very good. Back to work.

[Timmah2483]

Anyone have a stat for SU favored winners at home, say -300 or better. Im sure its around 90%

[Data]

Thanks for the help.

I guess picking the su winner 70% on all games isn't very good. Back to work.

Back to work? You do not have 101 type of data for the given league and yet you call whatever you have done thus far a "work"? This was the most ridiculous thing you posted since I joined SBR. Come on, you are not stupid, this "work" is not that easy.

[Pancho sanza]

Back to work? You do not have 101 type of data for the given league and yet you call whatever you have done thus far a "work"? This was the most ridiculous thing you posted since I joined SBR. Come on, you are not stupid, this "work" is not that easy.

rough crowd.

[Data]

rough crowd.

I am not a crowd. I wished I was...

[Justin7]

Back to work? You do not have 101 type of data for the given league and yet you call whatever you have done thus far a "work"? This was the most ridiculous thing you posted since I joined SBR. Come on, you are not stupid, this "work" is not that easy.

Data,

You are very proficient at critiquing others, and do exhibit a mastery of "101". This in itself is not sufficient to make a winning model. You have not posted anything suggesting that you are capable of doing so either. Until you do any successful modeling of your own, I don't think you're qualified to discuss what modeling work is.

[sideloaded]

Data,

You are very proficient at critiquing others, and do exhibit a mastery of "101". This in itself is not sufficient to make a winning model. You have not posted anything suggesting that you are capable of doing so either. Until you do any successful modeling of your own, I don't think you're qualified to discuss what modeling work is.

I'm glad someone finally said that.

[RudyRuetigger]

http://www.allbusiness.com/finance/3502623-1.html

[Data]

You have not posted anything suggesting that you are capable of doing so either.

I beg to differ. You have to account for that one can only hear whatever one can understand.

[Justin7]

I beg to differ. You have to account for that one can only hear whatever one can understand.

Data,

I'd love to see your input on this. At what point in your modeling process do you look at historical point spreads? And for what will you use them?

[Data]

At what point in your modeling process do you look at historical point spreads?

From the very beginning.

And for what will you use them?

Along with the MOVs, as a measuring stick.

[subs]

it's sad that this feud is going on.

if you feel like an argument why not just do what every 1 else does.

a/ go to the politics threads and talk sh1t.
b/ kick the dog/cat.

[Justin7]

There are plenty of metrics to gage how well a model is operating. Once you start using spreads, you are "poisoning" that data. You can only check it once against spreads. If you make adjustments and re-grade with the same spreads, you risk data mining problems.

In general, I would never use spreads "from the very beginning".

[Emily_Haines]

Here’s some data from my database for last three seasons plus first few weeks this year

favorites of…….

1-3 pts 1431-1096 (56.6%)
3.5-6 pts 1776-910 (66.1%)
6.5–9 pts 1621-460 (77.9%)
9.5-12 pts 1169-195 (85.7%)
12.5-15 pts 904-79 (92.0%)
15.5-18 pts 565-24 (95.9%)
18.5-21 pts 349-9 (97.5%)
21.5-24 pts 191-3 (98.5%)
24.5-27 pts 119-1 (99.2%)
27.5-30 pts 45-0 (100%)
30+ pts 55-0 (100%)

[jolmscheid]

Why not just parlay all 12.5+ point favs??

[jolmscheid]

I mean, if they win 90%+ of the time, wouldn't parlaying them whenever they come up be profitable?

[Emily_Haines]

I would think that the percentages would go up later in the season (Jan, Feb, Mar) as their is still a little uncertainty of how good a team is early on.

[LT Profits]

I mean, if they win 90%+ of the time, wouldn't parlaying them whenever they come up be profitable?

No. At 90%, the break even no-vig point is -900. Favorites of -12.5 or more usually have higher MLs than that, and many of them don't even have an ML.

[pavyracer]

31+ plus almost automatic win
26-30 points once in a while lose
21-25 points once a month lose
16-20 points 2-3 times a month lose
10-15 points 5-6 times a month lose
6-9 points 5-6 times a week lose
1-5 points half the times they play lose

[goblinburner]

Here’s some data from my database for last three seasons plus first few weeks this year

favorites of…….

1-3 pts 1431-1096 (56.6%)
3.5-6 pts 1776-910 (66.1%)
6.5 –9 pts 1621-460 (77.9%)
9.5-12 pts 1169-195 (85.7%)
12.5-15 pts 904-79 (92.0%)
15-5-18 pts 565-24 (95.9%)
18.5-21 pts 349-9 (97.5%)
21.5-24 pts 191-3 (98.5%)
24.5-27 pts 119-1 (99.2%)
27.5-30 pts 45-0 (100%)
30+ pts 55-0 (100%)

thanx 4 the info

[Data]

There are plenty of metrics to gage how well a model is operating. Once you start using spreads, you are "poisoning" that data. You can only check it once against spreads. If you make adjustments and re-grade with the same spreads, you risk data mining problems.

In general, I would never use spreads "from the very beginning".

Justin7, I am well aware that you have plenty of pseudo-scientific methods in your arsenal as you have presented them here in the TT over the years. So, when you say 'would never use spreads "from the very beginning"' that is not really surprising. Here are two simple reasons why a modeler (that you are clearly not) needs to use spreads while starting from scratch.

First, the modeler needs to get a grasp on scores and MOVs distributions for the given league. And if he has any clue he would want to use a method similar to what we could call Wong's radiuses and that will help him finding the relationships between expected and real MOVs. Failing to utilize spreads at this point clearly leads to producing distributions that do not reflect the reality.

Second, the modeler will want to create some kind of power ratings utilizing his methods of evaluating offensive and defensive efficiencies. While doing so he will immediately face the need to assess strength of schedule and at this point he has nothing better to use than the spreads for the past games.

[benjy]

http://www.allbusiness.com/finance/3502623-1.html

Cool linkage. Thanks Rudy.

[Justin7]

Justin7, I am well aware that you have plenty of pseudo-scientific methods in your arsenal as you have presented them here in the TT over the years. So, when you say 'would never use spreads "from the very beginning"' that is not really surprising. Here are two simple reasons why a modeler (that you are clearly not) needs to use spreads while starting from scratch.

First, the modeler needs to get a grasp on scores and MOVs distributions for the given league. And if he has any clue he would want to use a method similar to what we could call Wong's radiuses and that will help him finding the relationships between expected and real MOVs. Failing to utilize spreads at this point clearly leads to producing distributions that do not reflect the reality.

Second, the modeler will want to create some kind of power ratings utilizing his methods of evaluating offensive and defensive efficiencies. While doing so he will immediately face the need to assess strength of schedule and at this point he has nothing better to use than the spreads for the past games.

1. You don't need to look at game spreads to look at scores or MOVs.
2. You don't need to look at game spreads to determine power ratings or strength of schedule.

You're slipping, Data. Your modeling theory is approaching nonsensical.

I ask you again, have you ever made a winning model, or are you just talking out of your ass?

[Data]

I ask you again, have you ever made a winning model

I must have missed that question. Yes, unlike, you, I did make a winning model for a major league and have made a very good progress while I am working on a model for another major league.

[Justin7]

I must have missed that question. Yes, unlike, you, I did make a winning model for a major league and have made a very good progress while I am working on a model for another major league.

Do give details.

[C.S.]

Let me break out the ruler guys.

[Data]

Do give details.

That is what I do, have not you noticed?

Another thing that I share here is my perception of you as a square. I am an agnostic enough to admit that I might be absolutely wrong. I do think that based on everything I learned here I believe that I am right.

[Justin7]

That is what I do, have not you noticed?

Another thing that I share here is my perception of you as a square. I am an agnostic enough to admit that I might be absolutely wrong. I do think that based on everything I learned here I believe that I am right.

I haven't seen any details of any model you have made. I haven't seen any plays you have generated.

You claim you have a winning model in a major league. Based on the limited statements you have made, you seem to have no clue on modeling methodology. Surprise me though. Prove you have a winning model.