One of the big issues facing punters in the modern era is access to an extraordinary number of tipsters. Each of them claims to be successful to the point where one wonders: How do the bookmakers ever earn a profit?
Quite easily as it turns out. Back in 2005, Harvard Medical school released details of a study it conducted which featured over 40,000 bettors who placed almost 8 million wagers between them to the tune of well over £50 million. Overall, the entire group lost 10%. Moreover, just 245 punters showed a profit of over £1,000 or more.
This isn’t a surprise to those in the industry. For most people, sports betting amounts to little more than guesswork. Lucky runs eventually grind to a halt as your results revert to the mean. An unskilled individual who places around 200 bets at Evens at a bookie with a 10% margin has a 10% chance of earning a profit. After 1,000 such bets, this chance falls to around 0.1%. In other words, the more you bet as an unskilled punter, the more likely you are to lose.
Deep down, most people know this, so why does there appear to be so many success stories?
Survivorship Bias – Why It Paints a False Picture
The term ‘survivorship bias’ comes from World War II. When bombers returned home from missions during WWII, engineers checked out the damage caused to the planes of survivors. They found that the planes were primarily shot along the wings, rear gunner, and body. As these planes were already very heavy, engineers needed to strategically place armour on the planes to keep them safer without compromising mobility.
The engineers reasoned that the extra armour must be placed in the areas where the planes had been shot. However, a mathematician called Abraham Wald went against the grain and suggested placing the armour on the cockpit and rudder where there were no bullet holes. It probably seemed daft at the time but made complete sense.
After all, the planes that flew home survived after being hit. In that case, it was clear that the wings, body, and rear gunner areas were already strong enough. Engineers had no samples of planes that did not survive to work with, but it seems logical to imagine that when a plane was shot in the cockpit or rudder, it didn’t make it home. By adding armour to these crucial areas, the engineers improved the chances of a plane returning.
Until Wald intervened, the engineers had fallen prey to survivorship bias. They focused solely on data from survivors and didn’t take the downed planes into account at all.
The equivalent in the world of betting is when you focus on ‘winning’ tipsters only. On social media and so on, it is not as if tipsters shout about their losses!
What a Load of Monkeys!
The fact is, only winning tipsters boast about their success while losers delete their erroneous predictions and begin again with a new username or website. As a result, punters only see winners and are under the illusion that these tipsters are ‘good’ rather than ‘lucky.’ The vast majority of tipsters hit lucky streaks now and then but are losers over time. Of course, you don’t know this because they vanish into the woodwork.
Let’s say there are 100 tipsters. After 10 bets, only 50 remain because the rest have disappeared after bad tips. After 30 bets, we are down to 20 tipsters. After 50 bets, there are only two left. Punters make the mistake of assuming that these two are ‘oracles’ and will continue to win. In reality, as soon as they hit the skids, they will vanish and reappear under a new name.
In theory, you could get 10,000 monkeys to press a button to back or lay a horse at Evens. After a year, it is probable that at least one monkey (and probably maybe more) will have a huge success rate. Statistically speaking, it is pure luck. It is the same with horse racing tipsters. There are so many that there is certain to be at least one with an exceptional record. Over time, unless this individual is a genuine horse racing whiz, the tipster will revert to the mean and start losing heavily.
In 2008, Derren Brown neatly showed up the fallacy of survivorship bias. In the System, he anonymously sent five tips to a punter, and all five won. The punter became convinced of this mystery tipster’s wizardry, waged £4,000 on the sixth tip, and promptly lost. There was no cheating on Brown’s behalf because he genuinely picked all five winners.
However, Brown had actually sent tips to 7,776 people who he separated into six groups. He gave each group a different horse and ensured he focused on six-horse races only. After race #5, 7,775 people had lost which left one survivor.
The moral of the story: That awesome tipster you follow on Twitter could be on a lucky streak or else they have several usernames and are mimicking Brown’s example. Either way, a loss in the long-term is certain if relying solely on luck.
Between 2001 and 2011, a Pinnacle Sports contributor named Joseph Buchdahl accepted a total of 120 betting advisory services. From almost 25,000 tips, the average profit was over 17%, an astonishing return. Here’s the thing: all of these bets were made before Joseph used the site; he checked the past performance of each tipster.
When he started tracking them after accepting their tips, the profit on the next 90,000 tips was just 1%. It was quite clear that he had weeded out a huge percentage of charlatans who were on a lucky run which was inevitably going to end at some point. Even exceptional tipsters with a long track record hit the skids now and then. The difference between them and the rest is that they eventually start winning again.
The Archie Score – How to Tell the Difference Between Luck and Skill
Michael has written about the Archie Score in more detail in a past article, but I will briefly go through it again to outline what it means. It is a very useful method of determining whether selections are based on luck or skill. You can use it to analyse a tipster’s recent record to see if they are likely to continue or revert to the mean.
You can calculate the Archie Score using a fairly simple formula to find out how many winners you could expect from the selections based on chance. Begin by converting the odds of selections into probabilities. I recommend checking the recent record of any tipster you follow and complete the next steps.
I have made up 7 selections for this article.
| ||Decimal Odds||Probability|
You get the probability by dividing the number 1 by decimal odds. For example, 1 / 2 = 0.5. It won’t be perfect because it is best to stick with two decimal places, but you will get a pretty accurate outcome. Now, you add the probabilities together which comes to a total of 1.95 expected winners.
The Archie Score calculation is:
Runners x (Winners – expected winners)2
Expected Winners x (Runners – Expected Winners)
The other pieces of information we need are the number of winners. Let’s imagine that there were three winners from seven. Here is what the Archie Score looks like:
7 x (3 – 1.95)2 = 7.72
1.95 x (7 – 1.95) = 9.85
7.72 / 9.85 = 0.78
The Archie Score is 0.78. You can view the full table in Michael’s article, but I can tell you that the likelihood of this tipster being lucky lies between 32% and 48%. For the record, the lower the Archie Score, the more likely it is that the tipster is in the midst of a lucky streak. An Archie Score of 4.00 suggests a 5% possibility of chance and a score of 8.50 means there is virtually 0% chance of the outcome being down to luck.
Our hypothetical tipster is not trustworthy in the least. Would you trust anyone long-term when there is an almost 50% chance that their wins are down to blind luck?
The next time you see a tipster boasting about their wins, analyse their recent results and see if they are good or just plain lucky. Otherwise, you will remain one of the majority who is falling prey to survivorship bias. When it comes to betting, many of us have a blind spot when it comes to tipsters. We would rather believe that they are experts in their field, and are capable of helping you find the gold at the end of the rainbow.
Sadly, most online tipsters are full of hot air and make the most of a lucky streak. The Archie Score is an excellent way to determine the difference between luck and skill, but of course, the larger the betting sample, the more reliable your results.