Question for the math guys

**Miz** · 01-31-13, 04:21 PM

n = number of bets (subtracting pushes)
x = number of wins
u = n/2

sigma = (n/4)^0.5

z = (x - u)/sigma

Go into excel and type this function .... function = 1 - normsdist(type in the cell for z)

format that cell as a percentage. There is the likelihood that luck created the results (assuming a 50% probability is assumed).

Good luck

**gamblingisfun** · 01-31-13, 04:59 PM

Thanks for the help. I'll look into it when I get on my computer. What if the odds were say like -200 and I only went 50%? We all know that's losing money, is there a way to figure out (based on the odds of the average bet odds) what the odds of doing that poorly are? Also does your function work when the odds of each bet are not 50/50?

**statnerds** · 01-31-13, 05:08 PM

check back after your sample size is 500.

GL

**gamblingisfun** · 01-31-13, 05:30 PM

My sample size is about 2300 in real life and I picked 54%, I was just trying to figure out if there's a way to calculate the odds of that actually happening and being repeated. Thank you very much.

**Sawyer** · 01-31-13, 09:03 PM

Originally posted by gamblingisfun

My sample size is about 2300 in real life and I picked 54%, I was just trying to figure out if there's a way to calculate the odds of that actually happening and being repeated. Thank you very much.

2300? How many of these were steam plays?

**gamblingisfun** · 01-31-13, 10:32 PM

I don't play steam or any of that. Just work my "system" though I don't like that word really, but for lack of a better term that's what I call it. I just run my numbers and place my bets at 10am before I go to work.

**James D** · 01-31-13, 10:52 PM

Originally posted by gamblingisfun

I don't play steam or any of that. Just work my "system" though I don't like that word really, but for lack of a better term that's what I call it. I just run my numbers and place my bets at 10am before I go to work.

Do you track at all how often you beat the closing number?

**gamblingisfun** · 01-31-13, 10:56 PM

Originally posted by Miz

n = number of bets (subtracting pushes)
x = number of wins
u = n/2

sigma = (n/4)^0.5

z = (x - u)/sigma

Go into excel and type this function .... function = 1 - normsdist(type in the cell for z)

format that cell as a percentage. There is the likelihood that luck created the results (assuming a 50% probability is assumed).

Good luck

I used this formula for my personal record this year and found the percentage to be .0009%. I'm not sure if this formula assumed equal probability in all events (bets) i placed. Some odds were -250 some were +250, all in all the average odds were -115. Is there a way to factor that in there?

**gamblingisfun** · 01-31-13, 10:59 PM

Originally posted by James D

Do you track at all how often you beat the closing number?

I should have, and I suppose I still could with historical closing odds. I do have records off all odds of bets I placed. I'm no expert on steam or line movements but I assume that early in the morning, not as many would have pounded the faves meaning I got better odds, and likewise if I bet the dog early, I got worse odds than if I had waited. More or less I might have wound up sometimes getting better odds, sometimes worse. Might it all have evened out?

**James D** · 01-31-13, 11:15 PM

Originally posted by gamblingisfun

I should have, and I suppose I still could with historical closing odds. I do have records off all odds of bets I placed. I'm no expert on steam or line movements but I assume that early in the morning, not as many would have pounded the faves meaning I got better odds, and likewise if I bet the dog early, I got worse odds than if I had waited. More or less I might have wound up sometimes getting better odds, sometimes worse. Might it all have evened out?

Steam does happen early in day but it also happens midday, right before tipoff, anytime really. It may have evened out but if you are 54% after 2300 games I would guess it will not even out. I would assume you have beaten the number more often then bet into a bad number. This data would be significant, your 54% over 2300 is also quite significant.

**gamblingisfun** · 01-31-13, 11:25 PM

Originally posted by James D

Steam does happen early in day but it also happens midday, right before tipoff, anytime really. It may have evened out but if you are 54% after 2300 games I would guess it will not even out. I would assume you have beaten the number more often then bet into a bad number. This data would be significant, your 54% over 2300 is also quite significant.

It was during the baseball season, from mid april to near end of august. Placing my bets at 930am (ish) and most games didn't start until 6 or 7 all central times. I'm sure waiting for betting on big dogs or fading popular teams would have been better to wait for better odds. I really know nothing of steam betting, all I know is i hit 55.41% on ML avg odds -122, and 53.47% on totals avg odds -106. I used all reduced juice totals on 5dimes, but that early in the am they had no reduced juice ML only reduced RL, and I don't use that. I'm sure on a crappier book the odds would be worse, which would eat into my units won by virtue of me having to pay more for each loss.

**gamblingisfun** · 01-31-13, 11:27 PM

I'll be tracking my picks on here for baseball season, just want to see if what I had last year was a giant fluke or got extremely lucky, or if I actually have something. I know for damn sure the sample size was big enough for at least one season, I just want to get more actual seasons under my belt

**Miz** · 02-01-13, 10:30 AM

Originally posted by gamblingisfun

I used this formula for my personal record this year and found the percentage to be .0009%. I'm not sure if this formula assumed equal probability in all events (bets) i placed. Some odds were -250 some were +250, all in all the average odds were -115. Is there a way to factor that in there?

This is a shortcut estimate that assumes a 50/50 expected probability of win/loss. If you are doing ML bets, then you will have to use a more eloquent way. See Ganchrow posts. He graciously derived once for us and posted it in plain english for all to use.

**gamblingisfun** · 02-01-13, 03:38 PM

I looked up and I found the Ganchrow posts for finding the probablily that I wanted. I think I did it right, but most of that kind of advanced math is over my head, but from what I understood: I found the z-score by taking my units won and dividing it by the square root of my units risked. And then did the 1-normsdist(z-score) and it came up with a percentage of 3.354%. I take that to mean that's the probability that the results I got happened due to chance? Please correct me if I'm wrong as I'm no expert in that type of math. I also edited my spreadsheet a bit to make all bets to win 1 unit, and used that unit risked amount, because I think he said something like it works best if it's all the same unit size to win. Some of my actual bets were more than 1 unit, but for finding probability of results that I got, I wasnt interested in the units won. Though I did also find out the p-value when including my real bets of various sizes (which I have no idea if that works with the formula or not) and it came out to 0.843%.

**indio** · 02-02-13, 06:03 PM

Originally posted by gamblingisfun

Thanks for the help. I'll look into it when I get on my computer. What if the odds were say like -200 and I only went 50%? We all know that's losing money, is there a way to figure out (based on the odds of the average bet odds) what the odds of doing that poorly are? Also does your function work when the odds of each bet are not 50/50?

Simple binomials can calculate odds for things like this. Of course, the formula is calculating the odds on what the % chance of event occurring is entered into the equation. With things such as dice, cards, and keno drawings for example, this is accurate since the probabilities are based on true raw math. With sports, no one really knows the true probabilities of a result, only the market price. However, it does provide an interesting function in telling you the odds of being lucky, therefore telling you the chances that beating markets are due to your skill.

But, lets assume all -200 is really -200 and always a 66.67% chance of occurring. The odds of picking 50% or less when your chance to win each event is 66.67% is as follows :

10 events = .2131
50 events = .0108
100 events =.0004

You also asked about the chance of hitting 54% of 50/50 chances over 2300 tries. Assuming the events had a truly 50/50 chance of occurring, your odds of getting 54% or more right (1242 or more) is .000067, which when translated to odds is 14,924-1. To put that in perspective, the chances of shuffling a random deck of cards, and then dealing 4 cards and all of them being the same rank is 20,824-1.

Hence, if someone picks 54% correct of 2300 games that were priced at 50/50 chances, there is a 14,924-1 chance that he got really lucky, and a .999933 chance that he was picking many teams that had a better than market chance of winning.