Association Football Betting - My Personal Refinements

This post is motivated by a few considerations lately that I'd like to draw into one post. To an extent, I will be repeating some of what I posted earlier in the season about expected value and synthetic spreads. I would however like to bring together some aspects of the long shot strategy in the official picks with the edge on draws that I observed last year as well the new spread options available on Polymarket (and potentially Kalshi soon) and of course the moneyline no options. Hopefully these tweaks will help to extract a little extra value out of the football betting market.

The underlying methodology I am using in this post is expected value analysis, whereby to assess the edge on a bet, we estimate the win lose and draw probabilities convert their market odds to break even probabilities (or we do the inverse, we convert our probability estimates into fair value odds). In effect, we divide one by the other and multiply by our stake size to find the probabilistic average of all outcomes after we place the bet. We take away the original stake and the final figure is your expected profit or loss. We want to apportion our bets in order to maximise this profit.

Robert has noted on stream earlier this season that he witnessed the most profitable edge by taking long shots. Take Bayern Munich vs Club Brugge this evening as a hypothetical example. Suppose you thought Brugge are being underestimated by the betting markets (and I can see why you might think that). On Polymarket you can get Brugge to win at 6c. This represents odds of about +1567. Suppose you think Brugge has an 8% chance of winning. It's not much of a difference, but when you do your expected value calculation, you find that for a $50 bet, you'd expect to get ~$66 back - a 33% profit. Meanwhile, if you thought Bayern were undervalued, you could buy them at 87c. If you thought they had a 92% chance of winning, the expected value calculation gives you ~$53 back - a roughly 6% edge. Even though the percentage difference is bigger, your actual return is smaller. Now you can account for that by deploying different percentages of bankroll, but the point still remains that mispricings on the long shot side amplify larger than mispricings in the favourite's favour.

If you're a little bit patient, you can try and take them at 5c, which would be +1900 (and you'd probably get filled on them because Bayern are big betting favourites, so chances are someone will be willing to buy no on Brugge for 95c as your counterparty). That would send your edge up to 60%. From that point of view, it would make sense that the long shots are showing themselves to be more profitable - if the market probabilities are off, you can really hammer them if you know what you're doing. The main caveat is you need to withstand a fair bit of feast and famine, which is not easy if you're not confident in the underlying edge.

To address the above point, I do look at the other options available as a result of football having three outcomes rather than the two for most American sports (given draws are rare in American football). If I think a favourite is overvalued, I will generally be happy to buy no on them. I do this by setting a limit order at the highest ask to buy yes on that favourite and normally I will get filled on that sooner or later. One of the benefits of this is it reduces your variance since you can make smaller gains by covering the draw too. In the example above, I could buy no on Bayern for 13c if I set a limit order and wait for someone to buy yes on Bayern at 87c. Suppose the probability of Brugge winning is 8% and the probability of the two teams drawing is 13%. This implies Bayern only have a 79% chance of winning, and therefore that Brugge have a 21% chance of avoiding defeat. What's my edge here? Dividing 21% by 13% gives me 1.61, which taking away a unit stake means my edge is 61% - a little bit higher than my edge taking Brugge moneyline!

Ultimately, the probabilities I'm working with are estimates - I'm not even entirely sure if they're accurate - but for the sake of demonstrating the methodology: if you are confident in your estimates for the probabilities (and they might be as simple as converting Robert's good to lines to break even probabilities and making a comparable adjustment to the other two dependent moneylines), you can calculate which options will yield the best return for you. Combine that with interpreting order book flow (which is the principle I applied to determine I could probably get filled at no on Bayern for 13c since the betting public is generally happy to buy yes on Bayern at 87c) and building synthetic positions (such as the synthetic draw I've posted about before), you can get some really souped up entries all from identifying where your counterparty is buying an overpriced position - since all betting edge coming from your counterparty buying an overpriced position (equivalent to selling you an underpriced position).

My own general observations about which moneyline options provide the most value is that moneyline yes is best when the lines are priced tightly and one of them is somewhat mispriced whereas moneyline no provides the best value when a line is very clearly mispriced. The reasoning behind my thinking is that when a team is overvalued, that overvaluation propagates into the two other moneyline positions and so you'll generally only find an edge on one. The choice then is which of those two to take. This should be handled on a case by case basis. For the most part, this question is evaluated when making the official picks but it's a question you may want to ask yourself if you see a draw or underdog win advocated but with a very tight good to line. On the other hand, when a team is blatant overpriced, that normally means the other two moneyline options are underprices. Since buying no on a team is equivalent to a synthetic +0.5 on the other team, buying no will give you equivalent exposure. This reduces your variance somewhat and since overvalued favourites are easier to sell than buying underpriced draws, buying no is often the better play here. Once again, consider expected value and make your final call based on that.

We can look at this one step further and consider the new spread lines available on Polymarket. For Bayern vs Brugge, the line is set at 2.5 with Bayern as favourites. The new spread markets have been getting some play, better than MLB run lines but not as good as American football spreads (which sounds about right as spread betting gets good volume due to football being a global sport but it's not the most popular form of betting in the football heartlands). You can get Bayern -2.5 at 49c if you wait for someone to take Brugge +2.5 at 51c or you can take Brugge +2.5c at 48c if someone is willing to take Bayern -2.5 at 52c (based on order book flow, you should have no problem getting filled on the latter).

I won't go into expected value calculations for the spread lines because honestly, I haven't been looking at football spread since I finished my analysis of last year's official picks. I saw compellingly better value on moneyline and the +0.5s via the no option for moneyline so that's been my approach ever since. I did note that spreads performed comparably well to moneyline so long as they weren't low spreads, that is the -0.25, pk and +0.25. Bigger spreads still did OK but generally I saw that moneyline markets had a lower overround than spread markets on bookmakers such as BetOnline. With spreads available on Polymarket, that eliminates the overround on both, so the two are on more or less an even footing. Therefore, I would like to flag both spread markets and over under markets as a source for potential edge for the rest of the season and I cannot imagine that Kalshi will lag too far behind Polymarket in this respect. I will have to take a look myself and see if football totals observe the same behaviour as totals in American sports where they have a tendency to go up as the game approaches, especially games involving high market teams.

Finally, a note on draws. Robert mentioned on stream that he generally looks for away teams that are slightly better than the home team such that home advantage is roughly commensurate with the away team's edge at a neutral venue. I would generally agree with this but as a matter of liquidity, it is not always so easy to get filled at a good price on the draw. Draw markets are just not that popular with the general public and so it's hard enough finding someone who wants to buy the draw, it's even harder to find someone who wants to sell you a draw (at least at the price that you're looking for). It is something you have to be careful with in draw markets that you may find yourself jostling with sharps to find an edge. It's not to say that you can't, but more that in the scenario Robert is talking about, I see no on the away team as a legitimate alternative (and which one provides a higher edge I see sheerly as a matter of liquidity more than anything else). This is why I try to deploy my synthetic draw strategy when I see two mid to high market teams playing each other, even when those two teams would be evenly matched on a neutral venue (and thus the home team being the favourite). The bigger the team, the more opportunity you have to sell that team and hence leverage draw liquidity pools that you otherwise would not be able to access. Sometimes the best way to navigate sharky waters is to swim around the sharks rather than with them.

We live in exciting times for all sports betting markets. I have no doubt that the more markets open up, opportunity will follow. Good luck and good gambling!