Statistical Arbitrage vs Matched Betting

by | Dec 24

9 min read

Hey guys, Joonas here, the founder of Edge Alerter. I’m going to go through the key differences between a statistical arbitrage approach to betting and matched betting. I’ve gotten a lot of questions from members, especially new ones, asking about how we’re similar and how we’re different.

This will be really good in clearing up how we are different and secondly, to just mathematically prove how the long-term expectancy is significantly higher from a statistical arbitrage perspective as opposed to the matched betting approach.

For those not familiar, my background is in Quantitative Finance. I’ve been a professional Derivatives Trader for many years. I worked at Bet365 for five years as an in-play trader and I’ve been betting successfully for many years.

The purpose of this analysis, matched betting is extremely popular globally. If you Google it, the internet does light up that there are an estimated over well over 1,000,000 people globally doing matched betting. For those not familiar, it’ll be interesting for you to learn how that works but then also as I mentioned just a moment ago, I’m going to give proof of how the matched betting approach is actually mathematically suboptimal especially in comparison to statistical arbitrage approach.

First question, what is matched betting?

Essentially you could break it down and call it the pure arbitrage of promotions. Then you might go, well I’ve heard of that word ‘pure arbitrage’ but what does that mean exactly? It means when you’re locking in a guaranteed profit. The matched better will do that through the promotions. They’ll back, for example, a team or player or horse with the bookies offering a promotion of maybe it’s an early payout, maybe it’s a second and third promo insurance within the racing and then they’ll lay it on Betfair typically.

They are trying to lock in, guaranteed themselves, maybe 5% to 10% profit on turnover. Something like that, but essentially locking in profit is what it’s all about and as I sort of detailed there, there’s a hedging component to it typically through Betfair but you can actually do that through other bookies as well. It’s locking in profit, it’s hedging, there’s this hedging component.

Then, statistical arbitrage. What’s that?

That’s all about focusing on purely expected value. We don’t need to pay away hedging costs or insurance fees out the door and we don’t need to pay Betfair a commission. It’s all about long term expectancy. That’s what that’s all about and the risk management for that, you might naturally ask the question, well hang on a minute that sounds a bit risky because you not hedging on Betfair or through other bookies. Yes, there can be higher variants however you can totally manage your risk through correct stake sizing. Those are the two key definitions there.

Let’s try and bring that out through a couple of examples.

Example one: Let’s imagine Betfair is liquid, by liquid I mean there’s a lot. The spread is tight, there’s a good amount of that liquidity in the book in terms of how much size there is there to back and lay as well. I will use $10 stakes for the purpose of this example. Imagine a horse is $3 to back with the bookies and $3 to lay on Betfair, it’s relatively rare to get that because the bookies are pretty quick to move but let’s just assume for the purpose of this example that that’s the case.

The matched betting approach, so how a match better would approach this is that there would typically be a promotion on this race. They would back the horse with the bookie and having 10 on that, if the horse does win, you’re going to win $20 profit. Then you’re going to lay the horse on Betfair at the same price of $3, which was the assumption here. You’re paying Betfair some commission there if you do win. If the horse doesn’t win you’re going to win $9.50 instead of the full 10. What is your net position after that.

If the horse wins your PNL is zero. If the horse loses you’re just losing $0.50 and obviously the hope here is that it runs second or third with this approach. From an expectancy perspective and let’s just assume that there is no promotion in this case just to keep the math a bit simpler. The expectancy here is negative 33 cents and the way you work out expectancy is by multiplying all the outcomes that can happen with the PNL that is attributed to each one. Minus 33 cents, that’s if you want to work that out in terms of percentages which is always very useful to do for backing because at stake was $10 that’s a negative 3.3% expectancy. That’s the matched betting approach.

Then the statistical arbitrage approach is similar in that we still take the $3 value with the bookie but we’re not bothering with Betfair at all. Then we ask, what’s the net position? We either win 20 bucks or we lose 10, so there is more variance here as explained but then the key thing is and there are really two parts to professional betting and trading.

1) Quantifiable long term positive expectancy.

2) Diligence stake sizing.

Expectancy is really critical to betting. Ask yourself, what’s the expectancy here and then multiply them out and it’s exactly 0. Instead of losing 3.3% where the expectancy is zero and obviously they’ve both got the same amount of promotional value in the second and third run. It’s 3.3% profit on turnover inferior but this is a liquid bet for example.

Example two: this is an illiquid Betfair market and so as I said, backing with threes it’s pretty rare, more likely you’ll be backing at threes maybe laying at three 330/340 something like that. We’re going to use 350 in this example. Again, I won’t walk through each line item but basically if you’re backing at three and laying at 350 the expectancy is negative $2.00. Off the base of a $10 stake, that’s negative 20%, so you’re giving up a lot or you’re donating a lot because you want to hedge and manage your risk. The startup approach here again, it’s zero, so it’s 20% profit on turnover different. That’s extremely significant.

To summarize the math in those two examples, in example one, the Liquid Betfair market, the matched betting approach has a 3.33% lower expected profit on turnover and in the Illiquid Betfair example, it has a 20% lower expected profit on turnover. So it’s extremely significant. The only other thing I would add to that is there’s a critical difference between profit on turnover and return on investment and whilst the 30% versus the 10% might not sound that significant, it obviously makes sense to most people but because there’s this power of compounding, after 10 bets your return on the investment, if your POT is 30%, it’s going to be over 1000% whereas if you compound 10% over 10 bets you’re winning at 159% return on investment so there’s a huge difference there.

The final comments, I just walked through those mathematical examples and they should give you some ideas into obviously insight into the expectancy and how the hedging costs are significant and over the long run those examples were negative 3.33% and negative 20% in comparison to the statistical arbitrage approach. Maybe you can assume that over the average matched better is probably about 10 to 15% profit on turnover inferior, so there’s this cost to expected profit.

Second point is around variance. One of the most common justifications for using the matched betting approach is to manage variants and that’s fair enough, you know intuitively, you’re backing it over here you’re laying it over there. You are capturing a spread, if you like, but ultimately through diligent stake sizing you are able to manage your variance and so it’s essentially insurance costs that you shouldn’t be paying.


Interestingly a lot of these matched betting services are actually affiliated with Betfair. Further to that, Betfair actually has the same parent company as SportsBet. There’s a lot of these matched betting services where they’re kind of in bed with the bookies effectively. They’re trying to get punters in their backing and laying in Betfair and chopping themselves up by paying these exorbitant insurance costs which is essentially what they are.

Sustainability as well, be careful with top price finders, simply because the bookies use them too. If all you’re doing is using these top price finders through these matched betting services to try and turn over bonuses for example then the bookies can find you fairly quickly.

The final point there is around the funding requirements. Betfair requires you to have the full amount of the potential liability down, so it is more expensive from a cost of capital perspective as well.

Then, I guess the final question is, alright I get it matched betting versus statistical arbitrage. I can see the difference is there and if I’m interested in a startup perspective or approach then how do you find a good system to follow?

Well, obviously, at Edge Alerter we’ve built up a reputation and a real strong track record of having a system that works. There are a lot of systems out there. Most of them don’t work. Most of them unfortunately are quite scammy at best, but this is the hard part to find a good startup system to follow and that’s an open-ended question. I’m not going tell you to use us over others but feel free to do the analysis and have a look around.

That’s it for me. If you’ve got any questions you can message me on Telegram or you can email us or you can even ask them in the comments section below this video. Thanks guys.