I Should Be So Lucky

Hi. It’s Clive from Racecracker here again this time talking about luck.

Firstly, does it even exist? Well, pretty obviously yes. There is a group of people who believe that everything is pre-ordained, and that luck does not exist, but I mention them only out of courtesy. I think that all of us would agree that it exists but what do we actually mean by luck? I mean the effect of random variance on past events.

Obviously, if you are the beneficiary of a million pound lottery win I’d call you lucky, wouldn’t you? In fact, I’d go further. If I was a more usual losing lottery player, I’d call you a lucky b******* implying that I never get that luck or of course, more correctly that I haven’t yet had it!

I’ve never yet met a lottery winner although one guy I used to work with resigned part way through his third week saying he’d won £5,000 per week for life but the fact that he turned up stacking shelves in Tesco’s a few weeks later made me somewhat doubtful.

The far more common occurrence is the statement “We hit the post four times. We were really unlucky” or something similar. This I don’t really count as luck. It’s far more likely that my team, Palace for the record, got their striker free with a packet of Rice Krispies and were beaten by a team of £50 million players owned by some oil rich nation. We lost because they were better than us, whatever my eyes saw.

I am reminded of the old Gary Player quote, “The more I practice, the luckier I get.”

So, how do we tell if something happens by skill or chance?

If you use tipsters, you really need to know. Not only are you spending your hard earned on their recommended bets, but you are further lining the tipsters’ pocket for the privilege. This is fair enough if he’s skilfully providing winning tips but what if he’s just been lucky. That luck will turn bad in the long run and that’s actually worse for you because it’s much harder to ditch a tipster that has been winning you money. If you keep full records, you will probably spend at least the amount he has won for you before ditching him.

Luckily, I have a luck simulator to hand. One given to me by a football tipster, Mike of Winabobatwo. A very good guy Mike, and also very good at Maths and Computers.

In this simulator, I can assume I am betting on coin tosses to keep it simple. Here I know the true odds of winning (2.0). If I assume my tipster has no edge i.e., he will win 50% of the time, then this luck calculator looks at how many bets I would need to place to be sure that the tipster has no edge and is effectively of no use to me. And you might be surprised at the results.

Let’s try 100 bets first.

The luck simulator effectively tosses a coin 100 times on 40 separate occasions and reports the results. Only in 20 tests did the tipster make between + or – 5%. That means on 20 occasions he either won or lost more than 5%. Yet we know in this instance there is nothing but pure chance (Or luck) leading to a figure outside the 5% range.

Obviously, we must try more tosses or in the real world more bets so let’s try 40 sets of 500 tosses. Sorry, it’s getting better but still not what we’re looking for.

This time we get 27 within the 5% range but still 13 outside the range and in one the groups of 500 tosses, one recorded a loss of £460 to £10 stakes but actually is no better or worse than any other group. The “best” made £340.

I know which hypothetical tipster I would dump and which tipster I would keep.

But we know that in reality there is nothing to choose between the 40 of them. In fact, you have to go through 5000 bets before all 40 samples fall in between our + or – 5% range. That might be fine for a test on my PC but it’s no use at all for a real world test of actual bets. You would have to stake £50000 just to get an accurate picture of whether the tipster has been lucky or unlucky.

Unfortunately, the picture worsens at larger odds. Let’s replace our hypothetical coin with a hypothetical dice. Now the true odds have changed to 6.0.

On the first go of throwing 100 times on 40 separate occasions we only got a pathetic 4 times that results fell between our 5% ranges and even at 1000 goes you still only get 22 times. My luck simulator only goes up to 10000 trials and even that only produces 37 of the 40 in the correct range we are looking for.

So, we haven’t made much progress in analysing our real world tipster, have we?

So, what else can we do? Well, that venerable site The Secret Betting Club spend their lives analysing tipsters. Surely they must be able to help. Well, I caught one of their highly recommended podcasts recently in which they let slip some highly useful information. They said that all of their top recommended tipsters had at least one year of loss amongst their last 8 years. All of them. Possibly the best online tipsters out there monitored for years by the best monitors out there and the luck still causes one every eight years to be a loss.

This really isn’t helping much, is it?

Next I turned to the renowned Archie Test.

In case anyone doesn’t know, Archie stands for Augmented Reality Computer-Human Interaction Evaluator. All clear now? In fact, for our purposes, we are using a simplified version, but I won’t bore you with the mathematical details of it. Suffice to say we are going to use readily available information (providing you keep records or have access to the tipster’s results and trust them!) together with our refined archie and let it do all the hard work

You will need only three figures

  1. The number of bets
  2. The actual number of winners from those bets
  3. The expected number of winners from those bets

The first two are easily countable within excel. The third requires a little simple work.

First, we need the average odds of all the bets in fractional format.

So, to give an example if the tipster selections had SPs of 8/1, 6/1, 4/1, 5/1, 5/1 6/1, 4/1 and 7/1 then that would be 8+6+4+5+5+6+4+7 = 45/8 bets = 5.625.

Then we need to turn this into a percentage to help us calculate the number of expected winners.

This is equally simple. Just divide 100 by the average we have just calculated.

100/5.625 = 17.78%. This is our expected win rate.

So, the expected number of winners from this tiny sample is 8 * 17.78/100 = 1.42 winners. A pathetically small sample that I wouldn’t recommend!

So now we have all the information we need to use our faithful friend Archie. Using another hypothetical example let’s say

  1. The number of bets was 800
  2. The actual number of winners from those bets was 160
  3. The expected number of winners from those bets turns out to be 140

To find Archie’s result we need to use one more sum

Calculators away now!

But what does this figure actually mean?

Well, the statisticians who use this sort of stuff all the time need to know the probability that their results are due to chance which is exactly what we want to know too.

They have a conversion table that converts good old Archie to probability.

So, our system with an Archie score of 3.46 has a probability of around 6% of being down to luck or put another way a probability of 94% being down to skill.

The controlling factor will always be the number of bets.

If you want more reliability, you need a bigger bet sample.

Before signing up for a system always check the past record and do an Archie test. As we have seen, the higher the Archie score the better.

If you reach a score of 6.5 you can be 99% certain that you’ve found a good one.

Still not 100% but the only things certain are death and taxes!

Clive JonesRacecracker