How to start building a sports betting predictive model
bettingexpert blog editor. Always taking the alternative route to finding the value.
How do you get started building a sports betting predictive model? What do you need to consider before setting off on the long journey? Andrew tells as few things to keep in mind.
Building a sports betting model can be difficult work. I won't lie to you.
It can mean long hours of tediously entering data, sorting spreadsheets, setting up databases, testing, re-testing and re-re-testing.
All this, with no promise that you will eventually 'crack the code'.
Would you expect anything different? Bookmakers have the sharpest minds working for them day in day out, using everything at their disposal in order to compose the odds that millions of people are trying to beat. You think you're going to come up with a system to beat them in a couple of afternoons analysis? It's not going to happen. It takes times and dedication, a sharp mind and persistence. But then, half the thrill is in the chase.
More often than not, the end product is worth the time and effort, even if it's just for the many lessons you will learn along the way, both practical and theoretical.
I've built a number of models over the years, covering everything from the NFL, NBA, Australian Rules Football and European soccer. And I can tell you, while no model you build will be light work, the first model is always the hardest.
So where do you start?
In this article I will discuss the fundamental things to consider before getting started on a sports betting model. And in doing so, try to impart to you some of the lessons I learnt along the way in the hope that it saves you some time and frustration.
Understand what you're attempting to do
It's pretty elementary I know, but you would be surprised by the number of people who miss the point and don't quite grasp what a betting model is trying to achieve.
And what is that exactly?
Well, each of my betting models attempts to assess the current 'form' of a team or participant, which is then compared to its opposition in an attempt to gauge the likely outcome of the contest.
What you're essentially trying to do with a model, in very basic terms, is create an independant point of reference from which you can ascertain the probability of all possible outcomes in a given match or contest.
Ideally you want your model to be able to recognise value in a given betting market. In other words, you want it to give a truer expression of a teams potential or 'form' than what the bookmakers odds do.
Once you've developed your model, you'll be surprised how often it can identify value in the market. Will it always get it right? Of course not. But a fully developed betting model will show you opportunities that the general betting public simply wouldn't consider.
Learn your probability theory
Yeah I know, it sounds like homework. Bor-ing. But you're not doing yourself any favours unless you understand the fundamentals of probability theory.
And it's not so much about learning and grasping theory, although it's important. It's equally as much about inspiration.
The more you read and understand about probability theory, the more imaginative you'll become with your models. You'll come up with all sorts of interesting and creative things to do with the numbers, taking angles you hadn't even considered.
Sure you can probably get by making a model with basic arithmetic. Maybe. But it's not going to be the cunning bookie killing machine that you've always imagined having at your disposal.
Know how to manipulate a spreadsheet
The first model I made was with graph paper. Yeah, that's right: graph paper. It was clumsy. It was inefficient. But it's all I knew. Then I discovered spreadsheets, and from there databases and from there I got into some very basic Php programming.
No you don't have to be a programming wiz to build a sports betting model. I'm not. But the more you do know about spreadsheets and the like, the better off you will be and the more powerful your testing and analysis will be.
And perhaps most of all, the more efficiently you will make use of your time.
So at the very least, know how to throw a spreadsheet around and learn how to make the data dance. And from there, work your way into building databases and writing queries.
Trust me. You'll be glad you did.
Know the sport and its betting markets
If you're starting your first betting model or system, I would recommend you begin with not only a sport you know well, but a league you know well.
And by knowing well, I mean like a ruthless expert.
If you don't understand the fundamentals of the sport or league, it's very difficult to know where to begin in your analysis and very difficult to know how to assess the performance of the sports participants.
And by understanding the fundamentals I also mean have a clear and comfortable understanding of the betting markets for that sport. The markets that you are going to attack is at the very core of your models identity. Is that market head to head betting? Is it line betting or handicap?
In other words, the manner in which you decide to assess a teams performance is going to be determined by the betting market you want to find value in. So know the sports betting markets as well as you know the sport itself.
You must also keep in mind bookmaker limits and market liquidity.
The amount of money you can get down on a particular league or bet type is something to consider before spending hours building your betting model. Sure, you might make a killer model for Polish 2nd Division football. But are you going to be able to bet at a rate that makes the time spent on the model worthwhile?
Personally I would stay away from the more obscure leagues. For one thing, bookmakers are far more sensitive to successful betting in these sorts of leagues. They will move quickly to restrict your betting if they feel you've got an edge in a league that they would readily admit to not knowing as well as they should.
Plus bet limits in these leagues usually begin pretty low anyway. And even if you move your action to a betting exchange like Betfair, you're going to have trouble getting your money matched in the lower leagues.
So aim high. Shoot for the big time. This is where the money is.
And it's where the challenge is too.
Build a model that will give you options and one that will provide for you long-term.
Data Data Data
Your model is going to need data. At the very least that means final scores, but ideally it means meaty in-depth stats that you can breakdown and incorporate into an algorithm.
And most of all – historical odds for which to test your model on.
Where can you get the data you need in the format you desire? Is it readily available in spreadsheet form?
Well, that can be the tough part.
There are plenty of sources on the net for statistical data for a wide number of leagues. Some are free. Some will cost you a pretty penny. It's worth spending hours trawling the web for sources. You'll often find the best sources in places you'd never expect, tucked away in the far reaches of the internet.
But you won't always find exactly what you are after, especially if you're looking to make a model for more obscure sports or leagues.
So there is always the option of doing your own data entry, even if it's to augment a data source from another provider. Personally, I'd advise this only as a last resort.
Because to be perfectly honest – data entry sucks.
But if you are going to head down this path and begin your own data source from scratch, just remember to repeat my mantra: “You only have to do it once. You only have to do it once.” It helps.
Oh, and remember to click 'Save' often. (The horror).
You can follow Andrew on Twitter @AndrewBexpert
If you enjoyed this article, you might also like:
You must be logged in to post a comment! Sign up + or log in in the top right corner.
Very detail information to make sports betting system. It is always hard to make system where are various topics to analysis.
Really good article like always, keep it going!
@ Andrew your a star, so much wisdom when it come to the game, some useful information above to consider incorporating extras into my own systems. I think your totally right about the spreadsheet as the amount of info we have to go through daily is to much for the mind to absorb and utilise in full. I have just written a tip for the Bolivia game tonight and chanced my luck on a 0-0 half time draw as we have seen so many so far even though I think we'll see a goal I am sticking with the system instead of instinct.