There’s an old saying which goes somewhere along the lines of “gambling is a hard way to make easy money.” Whether we like it or not, the bookies are typically the ones smiling at the end of the day. How often have you heard them whine about having a bad day at Aintree or Cheltenham? Seldom do you hear anything about the millions of pounds in profits they take in during most years.
Only a handful of people manage to make consistent long-term profits betting on any sport let alone horse racing. You may already know several reasons why most punters end up routinely tearing their betting slips in frustration:
- Lack of Betting Banks: This is a set amount of money you leave aside for betting only. It helps breed discipline but most punters bet with whatever they have in their pockets at the time.
- No Staking Plan: You could use Kelly’s Criterion or another staking plan to ensure you don’t blow all your money on one or two picks. Again, few punters follow such a rule and end up ‘chasing’ losses instead.
- Greed: I think we’ve all suffered from this affliction at one time or another. This ugly monster can cause us to lose our hard-earned winnings.
- Emotion: Things can quickly unravel when you bet with your heart and not your head.
- Impatience: Too many punters jettison a system far too early without ever really giving it a chance to succeed.
- Laziness: In simple terms, if you don’t make the effort to research horse racing statistics, you’ll ultimately end up losing.
The last point neatly brings me onto the issue of statistics. I know a lot of people who have no interest in statistics. In their world, they are only betting on the existing event so past outcomes are irrelevant.
In a way, I kind of understand their viewpoint. Statistics are often completely misused which ultimately makes them worse than useless. A prime example is an overreliance on a betting system that is ostensibly unproven. The only way to get around this is by understanding statistical significance.
What is Statistical Significance?
According to experts in the field, statistical significance relates to a result that isn’t down to chance or luck. In the wide world, ‘significant’ means important. In the world of stats, it actually means something that is ‘probably true’.
Statisticians tend to quibble about the exact definition of the term. For some, finding an outcome that is probably true 95% of the time is enough (which means there is a 5% chance that it is false). When it comes to betting on horse racing, you should push that figure closer to the 99% mark. In terms of following a system, this means there will be a 1% chance that it is a false idol.
The whole point of statistical significance is to help you understand that profits can only be attained over a long period of time. It is designed to assess data to see if a system is down to chance or if it is actually a viable money maker. Statistical significance is achieved only after a certain number of outcomes have been analysed; or bets in this case.
(By the way, ‘system’ could simply refer to a ‘trend’. Examples could be Richard Johnson’s rides at Bangor or Frankie Dettori’s rides in Class 4 events.)
This is why you need to be wary of systems that claim a huge ROI. The next time one of these emails comes your way, take a minute to consider if the data that will inevitably accompany the ‘system’ is statistically significant. Here’s an example:
Johnny X sends you a message telling you he has a new system that achieved a 62% win rate from 140 picks. First of all, if the email doesn’t give details of the average priced winner, ignore it completely. I mean, getting 62% of your picks right means nothing if these ‘tips’ are gems such as Barcelona to beat Getafe at the Nou Camp or horses with an SP of 1.25.
If Johnny X was to include more data on prices, then you can perform due diligence. Let’s say his tips led to a 12% ROI with average odds of 1.72.That would represent a decent profit and could be worth further investigation. After all, you would theoretically break even with a 62% win rate if the average odds were 1.61.
A Quick Note
The following information is based on the assumption that you’re using the best market prices. For horse racing in particular, this tends to occur on the Betfair Exchange. Then there is the small matter of the bookmaker’s advantage, also known as the over-round which leads to the profits (vig) enjoyed by the bookies.
On the exchanges, the over-round is lower than on traditional bookmaking sites. For example, it may be 106% on the exchange and 114% on Ladbrokes. This means you’re at a 6% disadvantage on the exchange while Ladbrokes have a 14% advantage right at the start.
Therefore, you would actually need to win well above 62% of bets at 1.61 odds to combat the over-round. Average odds of 1.72 should be enough to overcome the bookmaker’s vig in the long term however.
Back to Johnny X…
But wait! Is 140 picks enough of a sample size? I used a simple spreadsheet found at www.football_data.co.uk which included the following data set:
- Number of bets
- Average Odds
- Standard Deviation
- T statistic
- 1 in x Probability
The P-Value is most important here. It is basically a figure which outlines the percentage of chance involved. Remember, I suggested that you would want a 99% confidence limit to be satisfied that any set of data has statistical significance. This means we need the system to have a P-value of 0.01 or less.
Using the data from Johnny X’s tips above, I find that the P-value is 0.043. This means there is a 4.3% probability that the system he uses is down to luck. This equates to a confidence level of 95.7% which is well below the figure we need to achieve statistical significance.
How Many Bets is Enough?
This obviously depends entirely on the data set you’re presented with. Let’s return to Johnny X once again. In order for the system to meet statistical significance standards at the current level (12% ROI at average odds of 1.72), he would need to achieve it over 255 bets.
Obviously, if the ROI is higher, fewer bets would be required. For instance, the system would need to hit a 20% ROI at average odds of 1.72 for just 88 bets in order to be statistically significant.
Yet there is a caveat. If you manage a few long odds wins, it can skew the picture spectacularly. For instance, a system with an 80% ROI at average odds of 5.4 would meet a 99% confidence level after just 58 bets. At those odds you wouldn’t need too many winners from 58 picks (still very impressive mind you!).
It’s always best to look at several hundred bets if possible before committing to any system of data. Shorter odds and lower ROI are ‘less sexy’ but far more achievable. Surely a 10% ROI over a long period of time is better than ‘lose most of your money’ tactics?
For example, a system that achieves a 10% ROI at average odds of 2.0 over 538 bets results n a confidence level of 99%. Wouldn’t you feel confident in any system that manages consistent profits for over 500 bets and has just a 1% chance of giving a false impression?
Of course, statistical significance can also help you determine if a system is a complete fake. I’m sure you’ve come across a few ‘fixed bets’ sites which claim to have ‘insider information’ on football and horse racing. Then there are ‘systems’ with huge ROI rewards and you can quickly uncover the possibility of fraud.
For example, you could receive an email from a tipping site that claims a seemingly impossible ROI, strike rate and average odds combo. When you try to calculate this data and find the odds of it being down to luck are around 40 million to 1, it’s a fair chance that the site is using fake information!
One of the biggest enemies of the punter is impatience. This can happen in two ways. First, they could leap headfirst into an unproven system without waiting to see if it has statistical significance. If you decided to get on Johnny X’s system after 140 bets, you would be taking an unnecessary risk as you would need to wait another 115 bets before confirming that he was in fact onto something profitable.
The second method is to dump a system that has already proven to be statistically significant. For instance, you may follow a system and find that you’re down 42 units after 3 months. Then you could stop following it only to discover that it provided patient punters with a 200 unit profit over 12 months!
The big lesson here is to resist the urge to measure profits over days, weeks or even a couple of months. In previous Race Advisor articles, we have shown you multiple trends, sometimes over a 5 year period. While I don’t expect anyone to wait that long before netting consistent profit, it is an indication that you won’t learn how to bash the bookies overnight.