Tips to avoid survivorship bias

I’ve seen a lot written about survivorship bias recently and it is definitely something that anyone considering following tipsters needs to be aware of. The problem has been articulately described by Joseph Buchdahl in an article for Pinnacle Sports about survivorship bias. He writes, “Survivorship bias is the logical error of concentrating on the people or things that ‘survived’ some process whilst inadvertently overlooking those that did not because of their lack of visibility. Survivorship bias can lead to overestimating the chances of success because failures are ignored.”

Admittedly, the overall records of our Tipster Table tipsters does currently look better than it would have had every tipster that had ever tipped on the site still been included. I’m sure there have even been some tipsters who generated huge profits through luck early on and now find themselves treading water, or even worse on the decline. Some followers will have fallen victim to survivorship bias in terms of looking at the overall records of the active tipsters and deciding who to follow.

It’s clear that backing tipsters solely on their profit totals after 100 tips, sometimes more, can be problematic and there’s often not enough evidence after that amount of tips that the tipster will continue making profits in the future.

The fact that poor tipsters give up after backing a handful of losers clearly shows that they didn’t have the mindset to be a good tipster. It’s a good thing that they are no longer providing tips but there will still be some who have fluked a good start, continued for longer than they should have and gained some followers.

Tipsters from our competition get asked to leave if they clearly don’t have what it takes to be a profitable tipster. This helps maintain the overall quality of the tips on the site but it doesn’t stop the lucky ones or prevent the overall presentation being distorted. The evidence from our tipster competition shows that even with the tipsters who left and are now omitted are included, we have recorded profits in 19 out of the last 24 months. Include only the active tipsters on the site or the ‘Survivors’ and it’s 22 profitable months out of 24. There’s one day left for February’s competition and the profit total is the second highest ever and so the quality has at least remained high amongst our tipsters.

It doesn’t change the fact that there is still the possibility that some tipsters have got lucky and are continuing tipping of course or the fact that the tipsters look more successful as a group. It is however, the methods of analysing individual tipsters that needs addressing, more so than the overall information that is being presented. Indeed a top quality tipster is still a top quality tipster if he exists alongside only lucky survivors and no losing quitters, or both.

Survivorship bias aside, it’s common for followers to wait until tipsters generate excellent returns before following, only to give up after a losing streak. As soon as they give up the tipster hits a big winner again.  This can be just as likely to happen whether the tipster had previously benefited from luck or they are shrewd and knowledgeable tipster. This is often because of the followers confidence in the tipster and his false expectations. He/she should keep the faith for longer if they’ve accurately analysed a tipster and deemed him worth following.

But how do we analyse tipsters and how can we be confident that the already profitable tipsters are good tipsters that are likely to generate us returns in the future and that they haven’t just benefited from luck and survivorship bias?

Basically, we need to judge tipsters by more than just their current profit totals and ROI. If they are doing some or all of the following then there is good reason to believe that they will generate profits in the future:

  1. They tip early (Greater chance of obtaining value odds before market is most efficient)
  2. They frequently beat the start price
  3. They provide reasoning which shows superior insight or data analysis
  4. They’ve made a lrage numner of tips, preferably over 500 tips (less likely to be down to luck)

Possibly a bit controversial but I believe it may also be worth taking a gamble on a tipster before they’ve generated mouth-watering returns and made a large number of tips. Betting is after all, all about speculation. If a tipster has demonstrated that they tip early, they are beating the start price frequently and they’ve provided excellent insight then these are strong signs that they will start to or continue to be profitable. It’s always going to be better to get on board with a good tipster before he’s had what could be his best wins (not least because the more subscribers he gets the harder it will be to get on) and the signs above are a good way of evaluating this. In addition, you can compare the selections of newer tipsters with those who have already shown their credentials. We often see 3 or more tipsters pick out the same value outsiders on our tipster board for example.

Yes, betting on a tipster before they’ve actually shown excellent returns is more of a risk but it’s a calculated one and with added risk comes greater rewards. Just how much of a risk is this anyway? If you are betting early, beating the start price a lot, taking the best available odds from a bookmaker that utilises Best Odds Guaranteed (BOG) and staking sensibly, you aren’t going to go far wrong at all, regardless of the actual selections.