Learning how to get better at betting on football

I used to be a professional sports bettor. I also created the best data set in the world for predictive gambling on football/soccer when I built StatsBomb. However, I hadn’t gambled in ages, so I wasn’t sure how I would do in coming back to the “hobby.”

My criteria at the start was to keep things casual, to only bet amounts I wouldn’t notice losing, and to keep my weekends to myself instead of sweating through every match and line move like before. I had worked 60+ hour weeks for two decades at Pinnacle, Brentford, and StatsBomb — I needed a break.

I also promised to produce my bets and analysis ahead of time via the Variance Betting subscription. This was mostly to hold myself accountable for all the work writing the column every week requires. I genuinely only expected 10-20 people to sign up and pay for the sub, because “old dude writes about betting on sports” is not obviously compelling content, but by the end of the season we were headed toward 500 subscribers.

The bankroll was approximately $10k, though I never really deployed enough at once to stretch that. For the first half of the year bets were about $200 each and then I bumped that up to $400 each later when I was pretty sure I had an edge.

How did we do?

Well, we’ll get to that in just a minute. But first…

A Cautionary Tale

Many data and analytics companies run out predictions as part of a marketing effort for their products. Seth Burn (one of the people who gave me my start in sports betting twenty years ago) has classically catalogued these on Twitter and in print for a long time and ridiculed the awful ones. This includes IBM’s Watson at the US Open. It includes the various data crimes committed by SAP in the name of hockey analytics. And it includes sundry other companies that have dipped their toes into the world’s competitive predictions markets.

Today, we’re focusing on a former competitor of the company I founded, Opta. But the analysis isn’t mine, it’s done by an account on Twittter/X named @craptor [SeeWraptor]. He’s a poker player and gambler, and decided to test Opta’s model predictions against Pinnacle’s lines and see how they did across a Premier League season.

It SOUNDS like a good idea, right? A data company — the official data provider for the Premier League for like 20 years — producing useful material for gamblers and fans to understand how strong different teams are just makes sense. And given their history, Opta should be hugely credible in this area.

Not so fast, my friends.

The Experiment:

Opta Analyst’s result for the season: $-4735.

Now some caveats: This is every match. There is definitely negative selection bias in having to bet every match with a model, and it doesn’t include outside information. But it’s every match with a Kelly model attached to it that bets far smaller amounts when it thinks the lines are correct.

So yeah, this is terrible performance done by the public-facing arm of the Premier League’s chosen data company.

It once again goes to show that even when sources of model numbers and math and gambling seem VERY CREDIBLE, you need to be careful in listening to them and applying their opinions.

Meanwhile, this is what I said about my own foray back into betting in September.

I promised to be honest about when I was losing — which happened a lot in the first couple of months. I discussed it in detail in October, and have made it free for everyone to read.

I also promised to tell you when I felt we were getting lucky.

In general, I wanted to use this as a test to find out if I could still beat the hardest markets in football/soccer after all these years. Because the gambling game has changed.

The lines we are wagering on (Asian Handicaps and Totals) are the toughest markets to beat in all the world. They have more info put into them and more money bet on them than literally any other betting lines that exist except MAYBE Sunday NFL football now that Americans are allowed to gamble freely.

It is a HARD task to consistently win in these markets, and I would not have been hugely surprised if I failed at it, especially given how much I had done in my career to provide better ways to analyse players and teams in the sport, and to make that information public.

Zvi and I did the same thing in American sports when we started working at Pinnacle in 2007, and it’s almost certain that we would have earned more money staying professional gamblers than we did from rebuilding Pinnacle. But rebuilding Pinny was probably more fun.

Ted’s Results

English Premier League
145 wagers, +4.5 Bets. $41,600 wagered, -$670. -1.6% Hold

English Championship
194 wagers, +19.5 Bets. $60,000 wagered,+3700. 6.2% hold

UEFA Champions League
76 wagers, +10.5 Bets. $23,000 wagered, +4800. 21% Hold.

Total +7830. 124,600 wagered. 6.2% hold

So 78% return on a $10,000 bankroll. That seems… good?

And it was a good season, but also saved by a filthy Champions League run. It happens. I do not expect to get that lucky again in future years.

We’ll take each one of these leagues in turn and see how we did.

The English Premier League ended worse than I thought it would, and this is despite getting on the NOT Man City rollercoaster early and often. I also made good money on Nottingham Forest’s run throughout most of the season, not because we were convinced that Forest were the seventh best team in the league, but because the gambling lines barely rated them from start to finish.

So yeah, a disappointing PL performance. We won more bets than we lost, but the vig or bookie margins put us into the negative.

The question then becomes why? Why was this season, though basically breakeven, worse than so many I have bet on before? Variance? Or was there something that cost us that was more systematic?

Home Field Advantage is the main culprit here. Last season was the second-lowest HFA on record behind the COVID games played without fans. Obviously any betting we do includes varying HFA factors, and when that differs significantly from historical norms, results are likely to suffer. This is especially true when you are betting with models behind the numbers, and it’s also very difficult to adjust these factors on the fly, even with human intervention.

If you go to SoccerStats and scroll down to the bottom, you can see the calculations for how many goals the home team scored per match vs the away teams. The difference in 22-23 was .41. 23-24 was .32. LAST season…? .09. To put it another way, in the 24-25 season in the English Premier League, home field advantage was almost non-existent. Home teams won 41%, while away teams won 35%.

Meanwhile, the Championship maintained a very busty HFA, which further illustrates the point of how hard it can be to adjust this stuff on the fly.

The $1M question is whether home field in the PL will revert to norms this coming season or whether it will stay low for currently unknown reasons.

The Championship ended up basically as I wrote previously. 6% hold is exactly in line with my historic expectations. There were a few annoying late bumps (especially around Plymouth Argyle), and Burnley defied all mathematical laws to finish where they did defensively, but it was another good season.

The UEFA Champions League… whooboy, that was a second half heater.

ALL of the profit from that league came in 2025, and it came via the normal handicap betting, a bit of totals, and some gifts on the To Advance + Outright lines that worked out in our favour. For the final match, I was able to hedge an Inter To Win the CL bet I made at +975 for 100 against a 400 profit on PSG to win and lock in a nice leveraged return.

The new format turned out to be very fun to follow both from a gambling perspective and a football perspective.

Season Outrights 24-25

One thing that is NOT included in any profits are the season outright bets I discussed on YouTube/Podcasts in August of last year.

Here’s a reminder of what was suggested back then on video.

I’m going to publish a new one of these early next week for Variance Betting subscribers — there will be no YouTube versions of these because the podcast seems to have a totally different audience on that channel.

Leeds to Win (We had 45% likely) (Win)
Leeds to be Promoted (Win)
Burnley to be Promoted (Win, small bet. Model liked it, I was less excited. I was wrong.)
Luton to be Promoted (LOSS)

NOT Arsenal to Win (Win)
Man City to Win (Loss)
Liverpool to Win (Longshot at 8-1 — I avoided this for Slot uncertainty. *sad trombone*)
Arsenal Top 4 (Win)
Liverpool Top 4 (Win)
Manchester United NOT Top 4 (Win)

Market-Implied was very excited about Southampton being relegated and I felt Ipswich was highly likely vs the market. Whether you found ways to get bets in on this was up to you.

I disagreed with the model that Palace might have a relegation worry.
I thought Wolves had value in potentially being relegated. This was RIGHT… until they changed coaches and became midtable. (Obviously a loss.)
I also suggested Everton to stay up.

What Now?

Given the results above, that is a pretty interesting question. I am going to continue Variance Betting in public as I did last season — we’ll cover my bets and analyse all the matches from EPL, Eng Champ, and the Champions League for the same subscription price as before.

I’m also very likely to change from betting $400 to $10,000. As I have said many times before, this is much more like the stock market to me than what people think of as gambling or sports betting, and I only place wagers on matches where I think I have an edge.

Those who want to follow along, welcome aboard. And for everyone else, you can support us by buying our Season Previews in a couple of weeks while continuing to enjoy all of our great content across the newsletter and podcasts for free.

All the best, and a special thanks to the hundreds of people who supported this silly experiment. I hope it paid off for you like it did me.

—Ted Knutson

If you enjoyed this newsletter, we’d appreciate it if you would forward it to a friend. If you’re that friend, welcome! You can subscribe to The Transfer Flow here. We also have a podcast where we go in depth on transfer news and rumours every week. We’re on YouTube here, and you can subscribe on Apple Podcasts or Spotify by searching for “The Transfer Flow Podcast.”