Back Testing A Sports Betting Strategy – DESIGN A System
Last Updated On 28th November 2017: Sports betting or sports investing is a market that is designed to profit from people who want to bet small and win big. 50% of all money earnt by sports betting books is actually made through parlays and accumulators. These are where people bet under $100 in most cases and try to win $400+ with a mix of events all happening. There’s a major issue with this. When you build a parlay you are adding the bookies spread on top of itself, meaning that even if you win, the sportsbook actually loses less than if you had made each of these bets individually. This is why you often see sportsbooks giving individuals improved odds for accumulators.
Instead we recommend making well informed theories and designing a strategy for individual plays. This means with most bookies you only need to win 52.5% of your plays to cover the bookies spread. For example on under/over lines you will generally see -110 as an odds with the line set at under or over X.5.
Doing the maths we can say that if you win 52.5% on 100 bets you win = 52.5 units. You will have lost 47.5 of your bets for a loss of (-110 x 47.5u) = 52.25 aka a break even. This means that on lines of -110 or BETTER you would only need to win 52.5% of the time to break even. But that slight edge is how the sportsbooks earn money and why in most cases it doesn’t matter the end total of an event.
The issue arises when we need to make a high volume of bets with a high degree of confidence. Sometimes value is obvious (to you) and hence when you take those bets you think you’ll have a 60% edge. As long as the odds are good enough always bet these. But how do you find these opportunities if you are looking to become a full time sports bettor?
The answer is to build a strategy and back-test where there have been profits in the past. You can then look at the upcoming events and see what events match your theories. If you have a theory that hits 57% of the time that has been back-tested over 3 seasons and 300+ plays, then you can be relatively confident that this play will be a 52.5%+ play. And if the odds are -110 or better you should be making these plays AS OFTEN AS POSSIBLE.
Of course this is assumed edge.
How To Back Test Sports Betting Theories
In this post I’ll talk about how to backtest a strategy so you can find these opportunities on a regular basis. But first I want to talk about an important element to sports betting.
Treat Sports Betting As A Business
My background is in business and although that’s completely different to gambling some of the principles are very transferable. Such as the principle of removing yourself from the day to day “work” – The principle on working on the business not in it for example. If you want to scale a big business you need to have a team in place and with sports betting although you don’t need to have a team per say in place, you should consider how you can automate your processes. Back-testing potential theories is the key to this.
Remember it may cost to back-test a theory. Whether you hire someone to do this for you or you invest your time into this. The back-testing process itself is what we will talk about next and the basics of how to do this. I’ve manually back-tested over 100 theories before moving onto the scaling and outsourcing element and I can definitely say I should have outsourced this alot earlier than I did.
The Back-Testing Process
The theory behind back testing is easy. Let’s take an extremely simple sports betting theory: When A team loses they will go on to win their next game. Now of course this theory is terrible but let’s use it just for this example.
The next step is to build a spreadsheet that outlines when these plays qualify. My spreadsheets have the date of the event, qualifying game (previous game result) and the game that would have been bet on – with the result of course. This is then tracked as a win or a loss with the odds played. Let’s just assume for maths sake that every game is +100. This would mean that a 50% win rate would mean a break even.
Next you need to go and find the results for the previous seasons when this team was playing. If we are looking at an NHL theory then I like to go back 3 seasons, for other sports it might be less or more due to the number of games a team plays per season. Let’s say you decide to go back 3 seasons, that’s 240+ games to analyse for 1 team! If it’s a theory that could be used for all teams in a league that’s approximately 30 for the NHL then you’re looking at analysing 30 teams over 82 games over 3 seasons = 7,380 games. That’s a lot of games! And that’s only for 1 theory. If you have 10 theories to test well you get the idea.
Key: Ensure your theory is specific & measurable – For example you need to be able to find the historic odds for the event. I just use flashscore but if your sports/league isn’t on there then find a new way before going through each individual theory.
In some cases I test theories that I cannot get data for specific odds for. For example Team Totals are really difficult to find historic odds for and as a result the theory has to be SUPER profitable and not just a little bit so. This means that if we look for 57%+ in a theory with -110 odds. If we do not know the odds (hence team totals) then we need about a 60%+ to even consider delving deeper. I have one team total theory that hits around 50%… Sounds bad right? But it pushes about 30% of the time. So only has a 20% loss rate… Sound better now?
Once you’ve calculated this backtesting you can check to see if this theory would lead to a profit or a loss. If you would have lost money in even 1 season then the theory doesn’t qualify. I also like to make at least 10+ units a season with each theory otherwise it’s just too close to break even and doesn’t really make it worth our while.
Once you have 5-10 theories you are confident with you can very easily have a confident way to make 30-50+ units per sport per season. Scale this and increase your unit sizing after each season. Remember we are treating this as a business not just a way to make money quick. Next we’re going to talk about the 3 season or 100+ play rule.
The 3 Season & 100+ Plays Rule
When I talk about sports betting theories and back testing to my financial investing friends they quite often state a very important point – “Past results do not equal future results.” There’s a famous saying (I cannot find who by) that states the following: “Past Performance Does Not Equal Future Success” – Although this is 100% true we can reduce our exposure to this by increasing the amount of data we assess. The more individual games in your theories, the higher the chance of success as the lower the variance of the theory.
For example if you test a theory with only 10 results, then there is a huge amount of variance in this. This also increases something called confirmation bias. Confirmation bias is when you see something happening a few times and automatically assume this is a given or a profitable theory. To avoid confirmation bias and exposure to variance we increase the number of plays we test in each theory. As a general rule I like to have AT LEAST 3 seasons worth of data and 100 plays. Ideally having 100+ plays a season is best but sometimes with complicated theory set-ups they just do not happen often enough. Generally speaking the best theories are the ones that are extremely specific with 5 or 6 elements in place.
Building theories is one of the most difficult elements but one I’m looking to write a full guide on very soon. Generally speaking though you can just pick up ideas from being very close to a sport. For example if you watch all NBA or NHL games you will likely have an idea of what tends to happen in certain situations. If players are injured or getting into specific high scoring or low scoring games. Also if you know a goalie in the NHL is over/under rated or a team plays more defensively when they are playing a rival, then you can build a theory around this.
KEY: Remember you only need to back-test once AND you can profit multiple times from this one theory. Although it might take hours to test you can profit for years.
Scale & Outsource
Once you have created a few profitable theories it’s time to scale and outsource.
Scale – By scale I mean its time to look into more sports, more leagues and more theories. I recommend starting by specialising in one sport or league. For example we specialise in the NHL. This is our bread and butter and what gives us a consistent profit each year. This funds the research into other sports and new theories. Scaling is very easy when you have a team in place which is what I will get onto now.
Outsource – Outsourcing the research is something I recommend as early as possible. Hiring a VA (virtual assistant) to do the research for you, this is relatively cheap to outsource and I’ll talk about how to do this and potentially write a guide or build a course on this too in the near future. Just know that if you want to reach the 500+ units profit per year level then you’ll either be working 20 hours a day or you will need to hire help.
Sports Betting Tips and Strategies
Invest in your sports betting education and research.