I’m really excited today because I’m going to show you something that I don’t think I have EVER seen blogged about anywhere else. But, it’s a stat that is crucial in deciding whether your going to start betting selections live or not!

It isn’t talked about because most people don’t know about it. Of those that do, the majority think it is too complicated to bother with. Now, just to reassure you, it isn’t complicated at all but it will take a few minutes to work out.

Would you rather spend a few minutes working out whether selections are ready to be bet live, with real money. Or lose your money because you don’t want to spend the few minutes it takes to investigate this stat?

You’re probably itching to know what this stat is. Drum roll please…

It is called… Probability Of Negative Return.

Okay so I know that it doesn’t sound very sexy. In fact it sounds a little boring. I tried to think of ways to spice the name up and… well I couldn’t think of any good ones!

But trust me when I say, boring as it’s name is, what it doesn’t sure isn’t boring. Particularly to your bankroll.

What this four word stat does is to tell you the chances of going bankrupt over these selections.

There is always some randomness in betting. You could get 10 losers in the row, 1 win and then another 20 losers. But, at the same time you could get 5 winners, 2 losers, 3 winners, 1 loser and then another 5 winners.

This stat is going to take all our results and ask the question… what if?

- What if the winners and losers didn’t come in this order?

It looks at every possible combination and then tells us the probability of making a loss over the course of these selections.

Pretty cool isn’t it.

You’re going to want to have at least 100 selections to make this valuable but as always with statistics, the more the better.

In order to calculate this statistics we need to perform the following calculations and I’m going to assume you’re doing this in excel so you can use the built in statistical functions as some get quite complicated if trying to do completely by hand.

1) Calculate the percentage profit on each bet

This means if the bet had odds of 4.25 and won, then your profit is 325%, if the bet lost then your profit is -100%.

Next we need to…

2) Take the average of the percentage profits

We get all the percentage profits for each bet together and work out the average of all these figures. You can use =average(selection of percentage profits).

3) We need to work out the standard error of the result from step 2

Don’t worry, this is very simple in Excel. You enter:

=stdev(all percentage profits)/sqrt(number of selections)

Obviously you’re going to need to replace “all percentage profits” with the cell range containing the percentage profit figures from step 1 and “number of selections” with the number of selections.

Then the final step is…

4) We calculate the normal distribution of these figures

Okay so I know that sounds complicated, but Excel allows it to be very simple for us to do. You type in:

=normdist(0;average of percentage profits;standard error;1)

That’s it!

You’re going to get a number between 0 and 1, and the closer to 0 that this is the better for you. These are probabilities. If the result is 0.40 then this means there is a 40% chance of having made a loss over these selections.

If you know that would you start betting on those selections live? I wouldn’t.

I like to see a 0.10 or 10% as a starting point. This I considered within my risk levels to consider start betting on the selections but normally I will continue to watch them a bit longer. If they continue to drop then I will start betting around 0.08 or 0.07 and will be happy to let the bankroll grow and make profits from that point.

If the selections drop below 0.05 then I will consider adding a significant amount into the bankroll for the selections.

I know that this takes a bit longer to perform than you may want to spend, but if you’re one of the people that is prepared to spend a bit longer in your preparation then you will vastly benefit in your bank account.

It would be great to hear your thoughts on this article below.

1. Why do you assume that everyone will need or even want your help

2. Why do you assume that everyone has excel

3. What makes you think that you are winning more than I am

4. Need I go on

Don`t be so presumptious

Thanks for your comment John. This blog is here for those people who want my help, if you don’t then you don’t need to read it.

Hi Mike, I found that very interesting, what adjustments will I need to make to calculate my probability of negative return when Laying.

Regards

Paul

sounds v good m8 keep up the good work

Hi Michael,

Does your betting evolution software take this into consideration.

Thanks, Chris.

Hi Chris, yes it does.

Hi Michael,

Thanks for another interesting article. I am currently paper trading a Dutching system and I’m just wondering how I would go about calculating the percentage profit on a winning bet which had numerous selections – would I just use the selection that won. And also are all losing bets, again with numerous selections regarded as -100%.

Thanks for your advice.

Antony

Hi Antony, thank you for your comment. How you calculate the profit on a dutch can simply be done based on the profit from the winning bet because you’ve only bet a fraction of your stake on the horse.

Hi,

Very interesting article but could you please explain :

a. why I cannot see the 8 comments mentioned above, and

b. why this method is better than simply working out the basic profit and loss on the bets? Does it matter what happens if the bets are in a different order,(assuming level stakes, of course)?

Keith

Hi Keith, you need to keep scrolling down to see the comments. They are after the comment entry form.

This method is not designed to replace profit and loss, but to compliment it. Profit and loss is excellent for telling you just that. However before you start betting you want to know what kind of risk level is associated with the selections so you can determine if it is for you or not and to also help you in deciding bankroll size. It makes no difference what order the selections are in for this.

For example, if you had made a profit of 100 units you would be thinking this is great I want to start betting live immediately. However after calculating this stat you find out that there is a 60% chance that these selections could have made a loss, would you be so quick to part with your bankroll? Probably not. You may want to watch the selections and paper trade them longer until you have more data and this number has been reduced.

Re my last comment.

Sorry Michael but comments are not there to scroll to. I did get to them once by clicking on a link in the Most Popular Posts but that doesn’t work now either.

Oh great!!

I submit a comment and as the page reloads I now get all the comments!

Great, sounds like your browser may have cached the page without comments and needed a refresh.

The Stat You MUST Use But Nobody Knows!

Do you do the calcs over todays odds or over a period of odds for each dog

example LTO x 5 or 10 or 20

Could you show a sample of these calcs here on the site

(monkey see money do)

If you are simply going off todays odds then you are simply saying bet anything under 2.00

or am i wrong??

This stat has to be compiled over a period of time for your selections, the more selections (and therefore the more time) you do it over then the smaller the figure should be. I would suggest doing it over at least a month of results and ideally three months. This shows you your chances of a negative return over a decent timeframe. If you have some figures you would like us to check then put your calculations in a comment here and we shall check them for you π

just getting to grips with statistics, so this may be a daft question…

When using the normdist function in excel, is there an assumption that the distribution of your win and loss percentages, follows a normal (i.e. classic bell shaped) distribution? I wouldn’t have thought this would be the case with most betting. There would be a large number of -100% (losses) and then numerous wins, probably at lower odds, with a ‘tail’ of wins at higher odds.

Not a daft question and you are right, there are other distributions that can be used but I stayed with this as it’s the easiest to implement. However you should investigate other distributions.

Great. Are there any distributions you could suggest as a starting point?

Have a good new year

Thanks Dave, I hope you have a good New Year yourself. It depends on what you are modelling to be honest, the best thing is to start by looking at the distribution and then comparing the most common models to see which is likely to be the best fit.