How do you you calculate variance?

[Redwing41]

When you start betting the most important thing is first finding +EV bets, second, finding the right sized bankroll.

But how do you find the proper size bank roll for different win% 50,55,60

At 50% odds, +100 pay out. What is the largest deficit and the largest winning and when does it taper off to 50%. Do I use Z score or standard deviation?

[u21c3f6]

Google Kelly criterion.

Can I assume that your mention of +EV bets that you no longer believe you can Martingale -EV bets into profits?

Joe.

[Redwing41]

Google Kelly criterion.

Can I assume that your mention of +EV bets that you no longer believe you can Martingale -EV bets into profits?

Joe.

Incorrect. I gave the results to a math student to look over. As of now I believe that martingale at those adjusted price is +EV. I believe it is called the bernoulli or binomial probability. Anyone that can understand the numbers will know that covering 3 losses in row, 2 losses in a row and 1 loss in a row = roughly 87.5% of all losses. Therefore, making it +EV money (when you read the charts) since the remaining 12.5 percent losses is made up of 4 losses, 5 losses and 6+ losses. It doesnt equal as much as it seems (martingale series losses) seeing as losing 5 in a row (5,6,7,8 straight losses) will eat up a lot of the losses quickly.

The stats in general sense looks like this

total losses = 51.2%
First 3 losses = 44.8% lose no money
4-6 losses in a row = 5.6% lose 7 X flat bet + 1 flat bet
7+ losses = .8% lose full progression amount + 1 flat bet.




Kelly criterion is what I was looking at and it doesnt really work out to determine this, since Kelly criterion is flexible. When you lose it adjusts the bet amount, so it does not accuratly give you the total unit bankroll for flat bet. Unless I take the kelly and keep assuming every bet is a loss and eventually it will be close to 0. I can then bet the amount of bets till it reaches 1, even that im not sure is correct.

I am trying to find in 50 bets what is the standard deviation. In 1000 bets I would assume that the standard deviation lowers, near 50% because the bets pile up and each loss change the percent by very little

I will have to use Z score. I remember being able to get data on it about this but need to research more in order to interpret it better.

[statnerds]

i'm not even sure how to phrase this properly as to not ruffle feathers

best i can do is you cannot paint a -EV to be +EV by using any type of (chase) system.

if you become skilled enough at finding +EV wagers you would have zero need for said system.

[Redwing41]

i'm not even sure how to phrase this properly as to not ruffle feathers

best i can do is you cannot paint a -EV to be +EV by using any type of (chase) system.

if you become skilled enough at finding +EV wagers you would have zero need for said system.

Thats fine. I dont get easily offended. I am getting +EV in sportsbetting. The adjusted martingale is something to do on the side.

The math plays out to +103.25 ($1 starting) per bet or 3.25 cents every 1 dollar bet. If even those that are called "statnerds" can not find this out, it is better for me as I can use it secretly.

[Waterstpub87]

When you start betting the most important thing is first finding +EV bets, second, finding the right sized bankroll.

But how do you find the proper size bank roll for different win% 50,55,60

At 50% odds, +100 pay out. What is the largest deficit and the largest winning and when does it taper off to 50%. Do I use Z score or standard deviation?

So you really need to specify a number of wagers to find largest deficit and largest winning. You will always have standard deviation, it will never just "taper off to 50%", it tapers off to that mathematically when you get to infinity, but you will never get there in actuality.

A z score represents the number of standard deviations away from the mean, so you could use a z score if you would like.

with a 50/50 bet

Your calculation of standard deviations with be .50 * Sqrt(n)

So for 1000 games, you would expect your results to fall at 500 wins +- 15.81

So for example, say that you went 450 for 1000. So 550 loss - 500 expected = 50/15.81 = 3.16 z score

This would put you at less then the bottom .0001%


The largest deficit would be losing all of your money, the large winning would be great then 1,000,000,000.

You can calculate likely values, like where is the top 90/ Bottom 90% after 10,000 or something.