The Truth About Sports Tipsters: Betadvsior - A Case Study
Posted 22nd December 2013
Note: A postcript has been added at the bottom of this article with new and relevant informaiton.
In December 2012 I published a book How to Find a Black Cat in a Coal Cellar: The Truth About Sports Tipsters. One of the central themes in it was to show that, on average, sports betting tipsters of which many are selling their advice for money, actually offer little more than guesswork and luck. Now that Sports-Tipsters has finished verifying the advisory service network Betadvisor.com, possibly the best known sports tipping service on the Internet today, I thought it would be worth telling the story of their performance as I have seen it. Readers, as usual, should understand that this story reflects my opinion only, and not those of Betadvisor, who of course will have a completely different view of events.
Betadvisor.com is a multi-tipster platform allowing both sports and racing tipsters to sell their betting advice to punters wishing to buy it. By offering an outlet for tipsters who might otherwise not have the means or skills to manage and advertise their own website, Betadvisor takes a percentage of any sales from their subscriptions. From the 17th February 2010 until the 15th May 2011 Sports-Tipsters agreed to verify all sports tips (Sports-Tipsters was never responsible for verification of any of the racing picks) from all tipsters operating at Betadvisor, and analyse and publish their results as one complete aggregated record. After this time, Betadvisor had submitted 10,134 tips for verification, returning a yield of +0.76%. The time series of these tips is shown below. According to statistical analysis there was a 22% probability that this performance could have happened by chance alone. This is a long way removed from what I (and most statisticians) consider to be evidence that chance would probably not the cause (1%). The records of the participating individual tipsters were not analysed separately during this time.
It was then agreed that since such a record was essentially artificial (since no single subscription package for all tipsters existed), public verification of Betadvisor's tipsters as a single record would cease. In its place, Laurent Marty (rugby tipster) and Geert van Elsen (football tipster specialising in Belgian football), in Sports-Tipsters opinion Betadvisor's two best tipsters at that time, continued to be publically verified individually on Sports-Tipsters. In addition to this, Sports-Tipsters agreed to continue to receive and archive all tips e-mails privately for all BetAdvisor's sports tipsters, and did so until verification was terminated on the 19th December 2013. Both Laurent Marty and Geert van Elsen continued to demonstrate their obvious abilities as tipsters. As of the 19th December 2013 Laurent had a yield of +12.27% from a total of 1311 tips dating back to the 19th February 2010. Geert van Elsen had a yield of +12.48% from 401 tips dating back to the 7th April 2010. Their profit charts are show below.
On the 22nd October I came across a forum thread on VerifiedTipsters discussing Betadvsior's verification, and decided it was necessary to explain the situation. In my post I confirmed that Sports-Tipsters continued to be in receipt of all sports tips, but observed that Betadvisor's summary performance page of all tipsters presented a misleading picture of the genuine success of their tipsters during the time that Sports-Tipsters received their e-mails. At that time, there were 41 tipsters listed on that performance page and all of them were in profit. I was confident that based on what I had verified between February 2010 and May 2011 this was highly unlikely to properly reflect the actual performance of all sports tipsters that had passed through Betadvisor since that time, and decided to set about analysing the performance.
In doing so, it became apparent that individual reporting for each tipster's performance on Betadvisor was including both the tips that had been made available for subscription as well as those that were recorded in house during any trial period a tipster had undergone through Betadvsior's Tipster Academy. These Academy tips were never independently verified by Sports-Tipsters, and since they were neither ever publically available for sale, I took the view that they should not be included within the official statistical reporting. At the very least, if they were to be, then a clear distinction between the two periods should be made, with a date provided for when any tipster moved from the Tipster Academy to the main Betadvisor service. Consequently, in the interests of transparency I asked Betadvisor to make such a distinction clear.
On the 12th December Sports-Tipsters wrote to the managers of Betadvisor to explain that it was no longer willing to support it. Although some progress had been made to demonstrate the distinction between trial Tipster Academy tips and main Betadvisor tips on its certification page, I felt that it was still not enough. Importantly, it was still not possible to identify from individual tipster's records, which of their sports tips had been available for sale publically and independently verified (by Sports-Tipsters) and which had been trial tips (through Betadvisor's Tipster Academy), not been independently verified and never available for purchase. Seemingly, Betadvisor felt that implementing such a distinction on every tipster page presented a technical challenge that would take some time to organise. Given that Betadvisor utilises a highly sophisticated database to manage and report on its tipsters, I always felt that this was an unlikely excuse. Certainly, after 7 weeks of waiting it had little credibility.
The overall profit from the 17th February 2010 to 31st October 2013 from 37,813 picks (including voids) was just +0.44%. The probability that this had happened just by chance is 20%, again far, far removed from any serious measure of statistical significance that might demonstrate that something other than chance was involved. [Read an earlier article about how to separate betting luck and skill]. The time series is shown below.
Between the 15th May 2011 and 31st October 2013, only 2 of the 118 sports tipsters that had been operating during that period achieved what Sports-Tipsters considered to be a statistically significant level of profitability (at the 99% confidence limit) that might be regarded as some evidence of skill over chance: Laurent Marty and Andre Nitu. (Extending this period back to February 2010, Geert van Elsen was also seen to be demonstrating a statistically significant level of profitability, but because his rate of tipping was so low, his profitability did not actually reach the 99% confidence limit for the 15th May 2011 to 31st October 2013 period). 70 of the 118 tipsters (or 59%) between the 15th May 2011 and the 31st October 2013 actually lost money.
Betadvisor has stressed that losing tipsters have sold far fewer subscriptions, and so it is unfair to group all the tipsters together into one analysis. Indeed, this was precisely why aggregate-single-record reporting was stopped in May 2011. But this of course completely misses the point. Betadvisor claim to be offering the "best sports betting tips". It says as much quite clearly in its logo. Furthermore, on its FAQs page it confidently claims that all its analysts are "profitable in the long run" (see Do you recommend certain analysts over others?). I suppose if "long run" means infinity, then just possibly this will give enough time for Betadvisor's losing tipsters to turn things around. What is abundantly clear, however, is that Betadvisor itself does not grant this luxury to its underperforming tipsters. How else could a summary page show 41 profitable tipsters and 0 unprofitbale ones? Consequently, then, this assertion is manifestly untrue. What's more, quite how anybody is meant to know what will ultimately prove to be a losing tipster before he has become one and at the time a potential subscription is purchased is anybody's guess. Undeniably, based on these statistics, there will be plenty of Betadvisor subscribers who will have lost money. If that was not the case, it would rather beg the question why losing tipsters were ever allowed onto Betadvisor in the first place, if they weren't even selling anything much anyway.
Betadvisor has rightly made use of its Tipster Academy to trial and select sports tipsters for its main service. Unfortunately, it would appear that not nearly enough rigour is applied to the tipster selection process. Too many unskilled and "lucky" sports tipsters are wrongly being selected for forecasting talent that few of them actually possess. Between the 15th May 2011 and the 31st October 2013 70 out of 118 seemingly slipped through the net. In my book, that is not nearly good enough for a site that claims to be selling the best sports tips.
Without more effort made by Betadvisor to present this reality to potential customers, it is easy to see how the service can be viewed (as many have) as little more than slick marketing. Sports-Tipsters was always supportive of what Betadvisor was attempting to achieve, but given this perceived lack of recognition of its weaknesses, I was no longer prepared to put my name to the Betadvisor project. The sports tipping industry should be about selling forecasting skill to those punters who haven't got it themselves. If all we have, on average, is luck, we haven't actually got anything worthwhile to sell at all. One might challenge this view on the grounds that at least some of the tipsters proved to be skilful. 3 out of well over 100? Is that really good enough? Presumably if we recruit enough people we'll always find a small percentage of them that genuinely have an ability that's worth selling. But that's a message we can never really expect services like Betadvisor to promote; it simply wouldn't be good business, would it? Better to say that you offer the best sports tips and hide behind the law of large numbers. As a business, it's their right to advertise their product as they see fit, provided that advertising remains within the confines of the law. It's my right to challenge that marketing approach provided I am not intending to defame the brand by using information that is untrue. There is no intention on my part to defame, and all of the information I have presented here can be verified and confirmed by Betadvisor themselves if they so chose to make the data available.
Betadvisor, however, are not unusual. Indeed, it is the case that all multi-tipster services, and all multi-tipster samples of data I have analysed, fail to deliver anything much more than luck alone. Of course, my book reported on my 11 year verification of nearly 250 tipping histories that showed a yield of +1.13% from 161,564 tips. Between 2004 and 2012 the Punters Lounge Tipping competition saw 579 individual tip 160,094 betting picks for a total yield of -0.95%. The customers whom I have referred to Pinnacle Sports since 2003 (arguably more sophisticated than the typical punter betting at UK or European bookmakers) have shown a total profit over turnover of -0.25%, from 56,876 bets. Arguably only 3 or 4 of them (out of a total of 102 for which data exists) have shown a statistically significant level of profitability that might be evidence of skill.
Perhaps most incredibly of all, Oddsportal's community of tipsters has tipped over 3.3 million betting picks since it was first established for an overall yield of just +0.85%. Of course it's true to say that this is a profit. But did it happen, on average, by skill or by chance? Yes, as for all the other samples of tipsters I have talked about there is always a small few who genuinely appear to have a real forecasting ability. But remember, here we are talking about on the average. Oddsportal make it very easy for its users to find 100%+ payout lines. Whilst we have no data on theoretical payout figures for these 3.3 million tips, it might be reasonable to assume that the average figure would be close to 100%. Tipsters, after all, don't usually tip less than best prices, and most best prices are at or above 100% payout on Oddsportal. With a yield of just +0.85%, can we really say that the average Oddsportal tipster is doing anything more than finding a best price in a falling market? After all, using just that very strategy (to find 100% payouts in shortening markets) with just one bookmaker (Sportingbet) I was able to show a yield of +0.91% from 5,751 individual betting selections.
VerifedTipsters has reported a yield of +5.1% from 60,000 verified picks. This result appear to buck the trend that punters who are prepared to advise others what to bet on are not capable, on average, of performing anything better than could be expected from chance alone. For the most part, I've not been party to the tips that have been verified by VerifedTipsters, so am unable to comment on this sample of data. Perhaps VerifedTipsters have a means of attracting better than average quality betting forecasters. Or perhaps this sample of 60,000 is too small, in comparison to the 160,000 tips verified by Sports-Tipsters, the 160,000 tips posted for the Punters Lounge tipping competition or the 3.3 million tips advised by Oddsportal community users. This is not for me to judge. However, until I see some significant sample of data that contradicts everything I have reported on here, I am prepared to stand by my assertion that almost all so-called tipsters are anything but, and that most of what is being sold represents nothing more than slick marketing in an attempt to attract customers to products that in fact are not what they claim to be. For the most part, that is not because the sellers are trying to defraud the buyers, but simply because the sellers haven't yet realised that luck is not the same thing as skill.
Postcript Posted 26th September 2014: In September 2014 Betadvisor changed ownership and updated their FAQs page, removing reference to the suggestion that all its advisors were profitable in the long run. The new owners should be applauded for that.