Re: Are there any statisticians here who could help me with some self education?

On Jan 9, 10:13 am, es330td <es33...@xxxxxxxxx> wrote:
On Jan 9, 3:47 am, Inventor <Inventor...@xxxxxxxxxxx> wrote:



On Jan 9, 2:26 am, Matthew Speed <msp...@xxxxxxxxxx> wrote:

On Tue, 8 Jan 2008 20:52:23 -0800 (PST), Inventor

<Inventor...@xxxxxxxxxxx> wrote:
What would you like to do?

Let's start with an easy one.

Every NFL game has an O/U line. In 2007, for unique games the NFC
East went over the line 29 times and under 23 times. Since the magic
number to make money is ~52% win rate, had I bet the over in each of
these games I would have ended a net winner.

If, on the other hand, one teases each line up 6, the outcome are now
20-32, a whopping 61.5% for the under. The catch, of course, with
teasing is that it is a multiple event bet and both must win to pay
off. Is this number statistically significant, or is this fact
something that cannot be exploited through betting?

Presented with these facts, the next question that naturally comes to
mind is "How can one do better?" Now many, many variables come into
play.
Did one team perform better against the OU at home vs away?
Does the point spread in any way correlate with which way the OU bet
goes?
How far off was the line from the outcome?
If every line were one, two, etc points higher or lower, what impact
would that have on the outcome of the wagers.
Is there a pattern in over or under outcomes vs line?

These are all questions that can be answered yes or no using
statistical analysis but I haven't the first clue in knowing how to
set up trying to answer them.

If you want to bet the over/under, my only help is to offer my
freeware program called "Predict_Games", which is available on my
website,www.freedomodds.com. It is very bad at the over/under in all
sports except WNBA (Women's Professional Basketball), in which it
excelled last year. My online web-based program, "Predict" has
essentially the same over/under algorithm so it should work as well if
you don't want to hassle with the download and usage. Others here can
probably help you with over/under strategies that are more along the
mainstream of handicapping theory.

Can you explain what you mean by "tease it up"? I am unfamiliar with
the teaser wager type. I think we can work up some probabilities of a
multiple-bet paying off if we start with an assumption of a certain
success rate, then multiply it out n times, where n is the number of
bets in the multiple bet. For example, you have 55.8% success rate
described. A two-game multiple bet would have 0.558^2 probability of
winning. multiply that by the payoff and if the result is greater
than one I think that means its a good bet. But then, I'm a newbie at
this so take whatever I say with a grain of salt :)

Inventor- Hide quoted text -

- Show quoted text -

It wasn't clear if you understood teasers or not so I am adding this.
It is like a parley except that you get to move your line 6, 6.5 or 7
points in your favor. The wager must be a multi event bet and the
payoffs understandably fall considerably vs a parley bet.

For anybody reading this, I am not looking to bet O/U plays in
particular. I threw out that example because it was easily defined.
I am trying to learn the process of statistical analysis, so if I were
to come up with a betting scheme, e.g. in any Tampa Bay game when the
day number of the year is evenly divisible by the numbers 5 and 13,
what O/U line will give me a reasonable probability of a successful
wager?

(As I type this, concepts like the area under the standard normal
distribution curve are starting to come back to me. I may just need
to find a used intro stats book and read a few chapters.)

I found a nice online textbook for you. it is at:

http://davidmlane.com/hyperstat/

And it says:

If A and B are Independent
A and B are two events. If A and B are independent, then the
probability that events A and B both occur is:

p(A and B) = p(A) x p(B).

Which is what I attempted to describe to you. Please keep us posted
on your learning process, it would be cool if you did all the work of
relearning this stuff and then reported it to the newsgroup, haha.
Just think of how many people will be enlightened by your efforts!
Also, sometimes it takes a creative novice to come up with the next
revolutionary concept - will this be you? Anyway, good luck and have
fun.

Inventor
.

Relevant Pages

  • Re: Are there any statisticians here who could help me with some self education?
    ... bets in the multiple bet. ... you have 55.8% success rate ... The wager must be a multi event bet and the ... what O/U line will give me a reasonable probability of a successful ...
    (rec.gambling.sports)
  • Re: Are there any statisticians here who could help me with some self education?
    ... bets in the multiple bet. ... you have 55.8% success rate ... The payoff that this ... probability must be multiplied by to make the bet worthwhile would ...
    (rec.gambling.sports)
  • Re: backup archive format saved to disk
    ... having multiple storage locations with multiple copies and a rigorous ... For example, make multiple identical backups. ... There is no way, using any number of physical storage media, to ... if the probability of error in a data bit is less ...
    (Debian-User)
  • Re: Costing a fortune
    ... probability is close to zero ... Single severe injury and/or multiple minor injuries ... Single fatality and/or multiple severe injuries ...
    (uk.rec.walking)
  • Re: A Fun Recursive Prime Count Estimator
    ... Here's a recursive estimator for number of primes less than N. I like ... divisible by a multiple of 2 is zero. ... the probability that N is *not* divisible by any number up ... root, but we're saying that we don't know what the primes actually are, ...
    (sci.math)