Will Wilde is a London-based maths graduate who is the head trader at Football Form Labs. They provide soccer betting software that allows you to analyse a treasure trove of data across 50 different leagues. He’s on the podcast this week to share his betting insights and experience.
- How to use past data to identify profitable trends
- Why he bets into some of the biggest markets in the world
- Exactly how he stakes his bets and manages his money
- His expected strike-rate and profitability
Today’s Guest: Will Wilde
Dave Duffield: Good day Will. I haven’t had you on the show before so maybe you can just give the listeners a brief rundown on your background.
Will Wilde: Sure, I’m the head trader for Football Form Labs which is betting software to help people bet on soccer. We’re based in the UK, but we cover leagues all around the world. We run our own strategies, we let users create their own strategies in the software and also give them advice on how to be a better punter through editorial output which can be simple as a recommendation. Or it can be a lot further in depth of how markets are formed and how people bet.
Dave Duffield: And has some of that come about from your maths background. I believe you have a maths degree?
Will Wilde: Yeah, I’ve studied maths. It’s heavily data driven, but we also like to present it in a format where we can give users the tools to make their own decisions on data. It’s not going to be as simple as saying, this has happened in the past so this will happen in the future. We’ll drill down further into statistics and try and work out which ones are relevant to an upcoming fixture or other sporting events.
Currently we’re covering a lot of the Rugby World Cup and have a huge game weekend, but with the same tactics that we use to analyse football matches we delve into other sports predominantly rugby, tennis, cricket, golf, NFL, baseball. We cover a large range of sports where we can get a comprehensive database of past results and use that to try and predict what’s going to happen in the future and make some money.
Dave Duffield: So we’ve run a podcast series here the last few weeks with some fairly standard questions amongst Australian racing and harness punters . This one will be a little bit different in that I’ll just only ask the ones that are moreso relevant to the work that you do. The first one is probably an obvious answer in that we’ve been chatting to people about whether they are more data driven or gut instinct and experience. It sounds like you’re heavily focused on the data.
Will Wilde: We are massively focused on the data, and I think a key thing is having experience with markets as well. I think most good gut instinct comes from a lot of practice in the markets. I think it’s probably a bit harsh to call it gut instinct. It takes years of practice to get that good to know when the price is wrong. I think even gut instinct is driven by data even though maybe people aren’t aware of it at the time that data that they’ve accumulated even if they’re not actually tracking, even if they’re not keeping records all of the time. We use a data driven approach, we try and look at trends in the past that will be reproduced in the future. Maybe if they won’t be reproduced in the future if there’s a reason a team may change their behavior.
A lot of the time in football just the introduction of a new coach can massively change the way a team plays. Over here we’ve got a team called West Brom in the Premier Leagues, they’re now managed by Tony Pulis. He’s done some remarkable things at other clubs keeping them up in the league, but he doesn’t play a brand of exciting attacking football because a lot of his games tend to be pretty dull affairs. He actually was involved in a 3-2 thriller on Monday, but apart from that most of his games are pretty dull, whereas before he took over at West Brom they were a very pleasing team on the eye with more goals in their games but maybe less effective. Whereas under him, they should stay up comfortably , which is the goal isn’t it?
Dave Duffield: How do you find trends in the data that it likely to continue, because there are some untrained people that would look at very recent performances, and probably small sample sizes, and the backfitting and all that type of stuff. How do you avoid any of those pitfalls?
Will Wilde: I think the key really is finding value. How is the market going to be priced up? Is the mark going to be priced based on how well the team’s are known, for instance in the Champions League tonight we’ve got Man City playing away. That game looks like it’s being priced up as a standard non-Bayern Munich German team playing at home against the English champions-elect.
Now, anyone watching Man City know that they’re not playing anything like that at that level, certainly at the last couple of weeks. Mainly because they’ve got some big injuries in their sport. Their key man David Silva is injured again tonight, and without him they lack creativity. What we’ll do, we’ll go back and look at the game to where David Silva hasn’t played for Man City and how they’ve done without him. That’s how we would approach that game, and try and model that one against the prices available tonight.
Dave Duffield: You talk about the Champions League there, do you tend to concentrate on big games and big markets like that where I’m sure you can get on for massive volume if you so choose, or you more looking to cherry pick nice smaller leagues and less widely bet games?
Will Wilde: I think that’s a very good question. I think a good answer to that would be the evolution of a standard bettor. If I was recommending someone getting into betting now, to really make money from it you’re going to have to beat the most liquid markets in the world. Now, the most liquid markets in the world are priced up by some of the best brains in the world, but that’s to say that value won’t exist at some point between the price opening and the price closing or even in play.
If you could understand why markets are modeled as they are you can take advantage of the times where they might be wrong, because the price will move up and down in any liquid market, and at some point it’s a back for someone and at some point it’s a lay for someone. I think if you’re disciplined enough to say, okay I’m going to back Man City tonight at 2.2, should they ever get there, but I’m also going to lay them at 1.8 to have a decent range, at some point there’s a good … That bet might get hit. I think our strategy is certainly to beat the most liquid markets in the world, the toughest market but also the most popular markets. We do that for a few reasons really, some are commercial.
A lot of people will be more interested in the Champions League tonight than the League One game last night that maybe 10,000 people are watching compared to a couple of million maybe that watched Chelsea last night. I think commercially, it makes sense to try and analyse the toughest games. From a professional punting perspective, there’s little point in analysing the smaller games because you can’t make that much money off of them. At some point liquidity dries up in the lower leagues, and if you’re going to turn this into a job in itself it’s got to pay for itself. The only way you can do that is by being with the biggest leagues.
Dave Duffield: Do you tend to bet early when markets aren’t fully formed or maybe there’s some information in the market that isn’t widely known, or do you bet once all the objective factors are known and you can be pretty confident that there’s no late injuries, weather, anything like that?
Will Wilde: Yeah. I think that certainly the biggest change is the starting lineup. One problem that gamblers have got is that the market will move once the starting line up lineup is known. And someone will know the starting line up before you do. Our approach typically is we use a predictive team lineup method where we have experts in every league that we cover. They will between 2 and 4 hours before the game give us a predicted lineup based on their expert knowledge and their information from all the teams in that leagues. While that’s never going to be 100% accurate, what it does allow us to do, it lets us feed our model, which is a player ratings model, with predicted team lineups.
We can get on before the price moves. We can’t get on much before that because there isn’t too much liquidity more than 4 hours before the start of the game. That’s where we operate at the moment, before 4 and 2 hours before the game. Certainly, prices are more wrong the earlier they are available. The reason for that is that bookmakers cover every game and every market within that game. They haven’t got the resources yet to say okay, we’ve got to be on the ball, searching Twitter every second, Preston North End say, give a press update that one of their players might be injured.
They put those prices up there but they’re limits are going to be a lot lower because there’s a lot more uncertainty in the market. As soon as the teams are announced, and certainly when the game goes in play everything changes completely. Liquidity goes up a lot more, markets will follow a similar trend in every match, which would be based on the activity in that game. I think that that’s a great time for punters if they had seen something in-game or they know the teams have maybe a different tendency to the norm in-play, that they can get a great opportunity, but for us we tend to bet between 2 and 4 hours before the match.
Dave Duffield: Have you found soccer to have more or less betting opportunities because of the low scores. Obviously the there’s some variance there. What’s the average score in a game? Maybe 2 and a half goals or similar to that.
Will Wilde: 2 and a half goals, I think it means there’s a lot of luck involved. Most models now have a certain high luck dependency. They expect variance in terms of the number of chances that you convert really. I think a lot of sports are best modeled because there are more points. For instance, tennis, you got a reference point every minute whereas in football you’re only going to get 2 or 3 a game. Surprised that football is so popular as an in-play betting sport because of that. I think before the game it’s an opinion.
You get an opinion, but the nature of a game changes so much in football when a goal is scored. Whereas is in other sports, like in NFL, it’s until you get very late into the game the pattern of the game is very similar. Variance is a huge thing in football but equally it’s a very liquid market so if you do something the more you can do something because there are so many games covered, so many leagues around the world that you can bet on. You can have enough bets in a year to negate that variance. We place 3000 bets a year. On one game you might get very unlucky but over the course of the year you hope it would even out.
Dave Duffield: What about in-play? You mentioned that it’s a very popular option. What percentage of your overall betting volume would be in-play?
Will Wilde: I think a very small percentage. One of the reasons for that is that the football models are very efficient now and the value that I can get pre-match won’t necessarily improve in-play unless there’s something I’m expecting to happen in play, which if I could predict that there was going to be a goal in the 30th minute for the Away team and then I wanted back the home team, why would I take a pre-match position anyway unless it was backing the home team and expecting them to come in in-play. The reason we don’t be a lot of time in-play is that if we make something value pre-match it’s very rarely going to be value on the opposing side in-play.
It would be a bad bet to hedge any of our positions and because we’re having so many bets a year we would lose a large chunk of our edge by just greening up and just having a guaranteed profit. I think trading generally as a way to circulate your funds and reuse them and work for a better position it should in a sport where you might have a lot of positions in one match, say a 20/20 cricket match. It makes perfect sense. You’re going to have a lot of entry and exit points. In football because, as you mentioned, with the lack of goals it’s very rare that your opinion is going to change if you’re betting on a model.
The model is always going to say that the home team is value whether they’re 3 nil up, whether they’re 3 nil down. Whatever the marker is it’s going to be wrong by a similar amount that it was at the start unless there’s been something in-play to massively change your opinion. I think because models are so efficient in football and liquidity is so large you can always use an opinion-based approach rather than a model-based approach in-play because the market isn’t that wrong in the first place. You’re betting against an efficient market so if you think you’ve got a small edge why not take it in-play?
Dave Duffield: In terms of your staking and the money management, how do you determine what sends me back to bet and also is that influence by the size of the overlay? Is there any aspect of the Kelly Criterion price herein involved?
Will Wilde: Certainly is. I think the more value you’ve got the more you should stake for sure but Kelly is based on knowing the chance of every outcome available for which in sport is impossible to get, so I think a certainly limited Kelly is the best approach. We bet actually solely on the Asian Handicap, for any listeners unaware, is a way of turning a proper market from a 3-way market which includes a win, draw or loss into a two-way market where one team is given a number of goals advantage at the start of the game in increments of a quarter of a goal.
One of the reasons we use the Asian Handicap is because it’s very liquid, but another is that the odds are normalised to, on average, even each side. You’re probably betting $1.98 / $1.98 the best prices would make. You’re never going to get a situation where you make a 3/ 1 shot 6/4, so you’re going to have a lot more on the 3/1 shot. It’s what we make our handicap, the supremacy that we make it compared to the actual handicap. While we have a Kelly part of our stake it’s a very standard staking system because the returns are … You’ve already got a standard return-based staking system. You haven’t even tried to make it by that just because it’s the Asian Handicap that you always get a binary payoff.
We’re very cautious in our staking but we average 0.7% of our bank. I think in terms of how anyone should work out their stake going forward I think when you’re learning to bet and when you’re new to it I think it makes sense to have a higher staking system, even higher percent in your bank because you don’t necessarily have this physical betting bank. If only when you start taking it very seriously you’ve floated a bit of money to your bank, and not much higher, but I can imagine 2% or 3% is going to be a good way to test the system. Also, have it matter enough that you’re going to give it enough attention when you’re learning.
When you take it to a professional level, you don’t want any result to have a huge bearing on your annual results. We have about 0.7% of our bank on every bet, but the reason we have such a low amount is that we place a lot of bets every year so I think you still want to turn over a large enough percentage of your bank to make a good return over a year. If you don’t do that then the profit you’re going to make is going to be very small, and maybe you don’t give it enough attention because you don’t have enough turn over.
Dave Duffield: At $1.98 for the Asian Handicap what would you expect to win as a winning strike or a percentage of your bets?
Will Wilde: The market is very good. If we won 55% of our bets in a year we would make 1000% return on our bank. It’s not going to happen. We’d be looking at 52% strike rate against the market that’s going to be paying you out, let’s say, a 49% market. It’s like 2% over around and we try and make 52% or 53% but a very small amount over that many bets. It’s a good market, and it’s a tough market to beat.
Dave Duffield: Then how do you manage your bank? Whether you’re winning or losing do you adjust the size of your bet?
Will Wilde: Yeah, I think it’s a good question of when you should adjust the size of your bank. I think it’s good practice. If you’ve got a profitable system then compounding the bank is going to make the returns better. In a perfect mathematical system you would rebalance your bank every bet. That’s not practical at all whenever you’ve got a lot of bets going on at the same time over a short period. It would also mean that the most recent bet has a lot more influence on your bank than previous bets if you’re on a good run. We re-base our bank every quarter.
On average, that would be every 750 bets and we turn over our bank 5 times. I think you want to do it at least every time as maximum for every time you turn over your bank. Say you got a bank of $10,000 when you place $10,000 in bets have a look at what you’ve got in your bank and then re-base that to your standard stake, but I wouldn’t do anything less than one full turn over of your bank. We do it around five.
Dave Duffield: All right, we’ll leave it there for today. Will I appreciate you coming on the show, definitely a different perspective to what we’ve heard before. All the best, also I believe your speaking at the Matchbook Trader’s Conference?
Will Wilde: Yeah, yeah, it should be a great event. If you or any of your listeners are over in the UK in London on the 14th, be great to come and join us and we’re hoping it’s going to be live-streamed. It’s a really good list of speakers so it should be great day, and hoping Scott can make his dream work.
Dave Duffield: No worries Will, all the best, appreciate that.
Will Wilde: Thanks a lot.
Get More Betting 360
Make sure you don’t miss our punting tips to come! Subscribe on:
or you can directly download this episode by right-clicking, Save As Here.