If you’ve heard the word Archie batted about various betting forums and communities. You’re probably wondering who or what this is. Well today we are going to put an end to any questions and explain exactly what Archie is.
It is in fact a method of scoring selections and is commonly known as an Archie Score. The purpose of using an Archie score is to determine with what probability selections are based on skill or chance. Obviously the more likely they are based on chance then the less sure we can be of them, while the more likely they are based on skill the more confidence we can have in placing our money on them. You can use this on either your own selections or someones that you are following.
Luckily this is not too difficult to work out. There is some maths involved but I am going to break it down for you so that it is as easy as possible.
In order to calculate the Archie score we need to know how many winners we would have expected from our selections, based on chance. This is actually very easy to do. You start by converting the odds of all your selections into probabilities. This is done by dividing 1 by the decimal odds. For example…
| Decimal Odds | Probability | |
| Selection 1 | 10 | 0.10 |
| Selection 2 | 5 | 0.20 |
| Selection 3 | 7.5 | 0.13 |
| Selection 4 | 3.75 | 0.27 |
| Selection 5 | 2 | 0.50 |
Of course you will have far more selections in your sample, but in this example we are just using five selections. You can see that we have converted the decimal odds from each of the selections into a probability. Now, to work out the number of expected winners we simply add together all the probabilities. In our example we would have 1.2 expected winners.
So what is the Archie calculation. It looks like…
If you don’t like maths then please don’t run away, I promise it isn’t as hard as it looks at first glance. We already know how to calculate expected winners. The only two other pieces of information we need are pretty simple…winners and runners!
No calculations required to work out either of those pieces of information. With these three pieces of information we can work out the Archie score. In our example we have 1.2 expected winners. Let’s assume that we have actually had 2 winners and we know that there have been 5 runners. We have all the information we need. Our Archie score sum now looks like this…
This gives us an Archie score of 0.70
That’s great, but what does this actually mean?
| Archie | Likelihood Of Chance |
| 0.30 | 58% |
| 0.50 | 48% |
| 1.00 | 32% |
| 1.50 | 22% |
| 2.00 | 16% |
| 2.50 | 11% |
| 3.00 | 8% |
| 3.50 | 6% |
| 4.00 | 5% |
| 4.50 | 3% |
| 5.00 | 3% |
| 5.50 | 2% |
| 6.00 | 1% |
| 6.50 | 1% |
| 7.00 | 1% |
| 7.50 | 1% |
| 8.00 | 1% |
| 8.50 | 0% |
| 9.00 | 0% |
| 9.50 | 0% |
| 10.00 | 0% |
| 10.50 | 0% |
| 11.00 | 0% |
| 11.50 | 0% |
| 12.00 | 0% |
In the table above you can see exactly what the Archie score means. We look down the left hand column to find our Archie score, which is between 0.50 and 1.00, and then we look into the right hand column and can see that this means the results are between 32% and 48% down to chance!
Would you want to be betting on these selections at the moment? Probably not, you would wait until you had some more results.
The bottom line is that the higher the Archie score the more the results are due to skill and less likely to be based on chance. This can be one of the most useful measurements in knowing when to go live with your own selections or which tipster to choose.
congratulations
Yes your right you sure did confused me if that was an easy one would hate to see a hard one. Kind Regards Glyn.
Sorry to confuse you Glyn, let me know where it got confusing and I shall try and clear it up for you.
Who Is Archie?: If you’ve heard the word Archie batted about various betting forums and communities. You’re… http://t.co/9tIRFfdsVP
Interesting article. I have a question. How many runners would be required before performing the calculation in order to achieve a statistical sound result? It seems intuitively likely to be more accurate the greater the number of runners. Does this follow?
You are right, the more selections the better. Personally I like to have 100 winners as a minimum, although if you have a low strike rate method this may not be possible.
I guess the question to ask then Is what Archie/Chance do you start backing the selections at? Anyone have an opinion on this? Good article by the way.
Thanks Luke. It is personal really, at what point are you happy to conclude your bets are down to skill and will continue. Personally I like to aim for a score of 5.00 which is a 98% probability that results are down to skill.
Thanks for the reply Michael.Guessed it was down to opinion but was just looking for some sort of figure 5 seems like a good starting point.My current system has a ROI of 25% over 69 bets.Expected winners 38 and actual of 48 so it seems like a ‘winning system’.I then calculated the Archie score of 1.16 which is telling me maybe I shouldnt have as much confidence ,as the stats show that maybe the majorty of the results are down to luck.Thats they way I see it.
Good way of calculating but,? is picking winners or losers down to calculations of odds?, I prefer stats gleened from experience, just lately when horses at odds of 1-12, and 1-6 and 30-100 lose how would you cope with these as more or less dead certs calculated by the odds?, I believe as many systems tell us that long term actual stats will give you part of the answers to calculating winners or losers, then you have to go into other realms of info and look at these stats and by combining these and other stats you put yourself in the profits.
Thank you for the comment Fred. Very interesting post 🙂
As you say, this isn’t going to help you pick winners. What it will do is to help you understand whether the selections you are picking are more likely to be chance or skill. Of course like anything to do with sports betting this is not a hard and fast rule, but a very useful method of deciding whether to start betting live on selections. It isn’t a method of finding profitable selections within itself though.
I have the following results for a system that I am looking at; 497 runners, 64 winners and 79 expected winners. Seemingly a disaster, as separate profit/loss calculations would confirm, but the Archie score, according to my calculations, is a respectable 3.39. Can this be so? Does the equation still apply if the actual number of winners is less than the expected number (as in this case)?
Hi Paul, thank you for your post. This can indeed be so, in order to understand it we need to re-visit what it is that Archie is measuring…
Archie measures the likelihood that selections are due to skill. The larger the score the more likely it is because they are based on skill, the less the score the more likely it is they are based on luck.
With that in mind, we can create Archie scores for both profitable and non-profitable selections. Non-profitable selections can also be based on “skill” in that the chance of you getting those results randomly are not high. In other words with a high archie and negative return, you would be very skilfully picking losers more often than you would be by randomly betting.
I hope that clarifies it.
Yes it does, thank you. It is useful to know that there are other methods of measuring skill/success other than by profit/loss.
Regards
Paul
Who Is Archie? http://t.co/7xhdjTIeez
Hi – great write up. How is the actual Archie figure converted into likelyhood of chance figure though? EG 0.3 Archie score equates to 58% likelyhood of chance – how do you get from the 0.3 to the 58%? Is there another calculation? Thanks
Hi Steve, I will get Patrick to give you a response the next time he’s online 🙂 Thank you for your patience.
Hi Steve, the Archie Score is based on a calculation called the Chi-Square test which is exceedingly bloody calculated!
hi Michael
I am trying to utilise the Archie score dynamically in a spreadsheet and wondered if I might ask a question.
The spreadsheet is for all runners in a given year and this is with it not filtered for any particular model, just every single runner
Here is the example
Runners = 122327
Expected Winners = 15112
Winners = 12846
The first part of the equation is 122327*(15112-12846) squared = 628119297212 (I imagine we are only squaring the winners less expected winners, yeah?)
The second part of the equation is 15112(122327-15112) = 1620233080
628119297212 / 1620233080 = 387.67 which a long way above 12
Just trying to work out where this answer fits in the above table, as it only goes as high as 12, or do we just assume that anything higher than 12 is also 0% chance of luck?
Now obviously this is simply a full year’s racing data, so is not indicative of any system or model, but the cells in the spreadsheet used for the calculation change dynamically as autofilters are used, so number of runners, winners and expected winners will all change dynamically as the sheet is filtered, as will the Archie score. I will then simply conditionally format the answer in the Archie score box to reflect the likelihood of chance.
Does that make sense at all?
It looks like you’re doing this right. You have a much higher score than the 12 due to the volume of selections. You would always have a percentage chance above 0%, but it will be tiny. The process you are using in the spreadsheet should work, I can’t guarantee it without seeing the file, but it all makes sense.
cheers Michael
I am not so sure it is working as it should, as tried reducing selections using filters, but even with just over 7000 selections, the Archie score was over 99.
If you don’t mind, may I send the spreadsheet to you? If has a full year’s worth of racing data, but it is more just to check the Archie score is being done right. You will find it down the bottom of the data alongside the calculation boxes which do POT etc
Thanks Michael
Of course, please send it in 🙂
Sent overnight mate
cheers
Hi Michael
Just checking if you had a chance to glance at that Excel file I sent to see if the Archie calc is indeed amiss
cheers mate
Shane
Hi Shane, nothing has arrived. Did you have a typo in the email address by mistake? If you want to use the Help button on the bottom right of the page, you’ll get a response you can reply to directly.
hi Michael
No, just as the last time, I sent it via wetransfer, as the file size makes it not able to be emailed
Can resend if you like
cheers
Hi Shane, yes can you please resend as we can’t see it.
Hi Michael
Just resent now. You will find the Archie calculations at the bottom of the page alongside the normal calculations boxes you have seen
cheers
Shane
Hi Shane, it doesn’t seem like the file is coming through. We’ve not had anything, are you sure you’re using our right email address? If you’d like to send us a quick message through the button on the bottom right of the page then we can confirm it’s coming to the right place.
Hi Shane, it doesn’t seem like the file is coming through. We’ve not had anything, are you sure you’re using our right email address? If you’d like to send us a quick message through the button on the bottom right of the page then we can confirm it’s coming to the right place.