Adaptive Models

[Maverick22]

Lets Say I Took All The MLB Games That Occurred Yesterday.

After All The Games And their Data are recorded. I create A Model That Can Accurately "Pick" Those Games At 100% For That Day Only. Where the model uses the stats that occurred BEFORE yesterdays Games... ( All stats Yesterday -1 ). This model would be created daily...Following all the games... I've got my TI-83 On Deck and a database full 'o stats.

Is it foolish, to attempt to make tomorrow's picks, based on the 100% model of yesterday?


And if that wont work...Does this daily Model have any value?

[adlai]

no...

[Justin7]

Two problems...

First, each game has a non-100% probability. It is hard to "predict" an upset - you can only say that it is more likely than the market price suggests. If you are predicting upsets everytime they happen, you are probably skewing your model horribly.

Second, it is very difficult to make good projections for players/teams using only the current season's data. I'm not saying it is impossible, but I'm not smart enough to do it.

[Maverick22]

no...

If you can, Please be as ambiguous as possible

[adlai]

If you can, Please be as ambiguous as possible

in 2 words, data mining. and not the good kind. i'm an applied econometrician by trade, with a graduate school education. this type of 'model' building isn't useful and it breaks all kinds of statistical rules. and the fact that you are using a ti83... i mean, come on buddy.

i might be a little biased though. i'm one of the few statisticians that believes hardcore statistics has no application in sports betting.

good luck mav... most sports statistics, in a betting sense, are just pure coincidence.

[Nuggz]

I would say most statistical trends are coincidence, but i think its foolish to say that working with statistics plays no role in making picks.

[Dark Horse]

Is it foolish, to attempt to make tomorrow's picks, based on the 100% model of yesterday?

Yesterday
All my trouble seemed so far away
Now I need a place to hide away
Oh I believe
In Yesterday

[username474]

i might be a little biased though. i'm one of the few statisticians that believes hardcore statistics has no application in sports betting.

wouldn't that make you an antistatistician ?

[sycoogtit]

If you can, Please be as ambiguous as possible

[adlai]

i might be a little biased though. i'm one of the few statisticians that believes hardcore statistics has no application in sports betting.

wouldn't that make you an antistatistician ?

well of course, because the only application of statistics is sports betting.

[Maverick22]

in 2 words, data mining. and not the good kind. i'm an applied econometrician by trade, with a graduate school education. this type of 'model' building isn't useful and it breaks all kinds of statistical rules. and the fact that you are using a ti83... i mean, come on buddy. i might be a little biased though. i'm one of the few statisticians that believes hardcore statistics has no application in sports betting. good luck mav... most sports statistics, in a betting sense, are just pure coincidence.

How are statistics coincidence?

A pitcher that gives up 1+ home runs every game is not COINCIDENCE. That's maybe circumstance to him being terrible at what he does...

There is nothing coincidental about how Michael Jordan has 7 rings. Nor that Phil Jackson have a career win percentage of .705

Its coincide to roll a roulette wheel and hit 17 5 times in a row... That "statistic" is useless, b/c roulette spins are of a random and unpredictable nature...

Statistics are history of what has happened, where is the "unexpectedness" there?

It's a funny (Save for the word foolish.) how all your posts are boasting of how much a statistician/econometrician you are, yet you make statements like 'sports statistics are useless". Save for the word foolish.

[JayTrotter]

A more appropriate use of statistics in sports handicapping, is to evaluate the market reaction to certain situations.. evaluation of whether the price is right... This is a market, and it's a battle of perceptions.. if you could evaluate certain opportunities where the price of a line is not congruent with the actual price, ie. a team that should be -140 is -220, you would be better off than trying to predict tomorrow based on yesterday..

[JayTrotter]

my suggestion... find someone who is terrible and bet against him...

[Maverick22]

Lines, odds and prices mean nothing when you lose.

I'll invest efforts in other areas, instead of doing lots of detective work to find on odds and right prices...

I'm focused firstly on picking winners, and second on anything else.

[adlai]

How are statistics coincidence?

A pitcher that gives up 1+ home runs every game is not COINCIDENCE. That's maybe circumstance to him being terrible at what he does...

There is nothing coincidental about how Michael Jordan has 7 rings. Nor that Phil Jackson have a career win percentage of .705

Its coincide to roll a roulette wheel and hit 17 5 times in a row... That "statistic" is useless, b/c roulette spins are of a random and unpredictable nature...

Statistics are history of what has happened, where is the "unexpectedness" there?

It's a funny (Save for the word foolish.) how all your posts are boasting of how much a statistician/econometrician you are, yet you make statements like 'sports statistics are useless". Save for the word foolish.

i will reference jaytrotter's post. this is how i use statistics in sports betting. in what situations do individuals under or over bet lines? betting lines function like stock prices, in the long run, learning the behaviors of the gambling public is a much more profitable angle to play. this is how i use statistics in my strategy. been a pretty darn successful strategy in my lifetime.

and i will now take all your crap for my post.

[roasthawg]

Lines, odds and prices mean nothing when you lose.

I'll invest efforts in other areas, instead of doing lots of detective work to find on odds and right prices...

I'm focused firstly on picking winners, and second on anything else.

I dunno... for the most part this has been my thought process for the majority of my gambling career to date. Concentrate more on picking winners and less on the price I'm getting. But price is a huge part of the equation... at -110 a 53% capper has an expected roi of about 1.18%. At -108 that same 53% cappers roi goes all the way up to 2.07%... it almost doubles! That's a pretty major difference right there for a slight change in juice.

[Justin7]

I get 100 coins from a jar of pennies. I flip each one 100 times.

3 of these coins landed hit "heads" 60+ times. Are those 3 coins likely to continue hitting heads at a higher than 50% rate?

Any player stat includes a combination of chance and skill. In a short period of time, chance dominates skill. The domination is so complete, it is almost impossible to handicap MLB looking at only this year's stats.