The season is over, and some of the model assessments among the soccer analytics community are starting to trickle in. Martin Eastwood is first out of the gate, whose eponymous index was included in Counter Attack’s Friday Football Predictions model comparison. It turns out the Index did rather well:
Looking at Figure 1 you can see that the Eastwood Index has consistently outperformed the bookmakers all season – and this isn’t just one bookmaker that the Eastwood Index has beaten but the combined knowledge of the industry as I’ve aggregated multiple bookmakers’ odds together and stripped out the overround to make the comparison as tough as possible.
Interestingly, the difference in accuracy seems to be greatest as both ends of the season. I expected the start of the season to be difficult to forecast as new teams have been promoted, players have been bought and sold, and managers may have changed clubs but the Eastwood Index seems to have coped with these variables better than the bookmakers’ odds have.
While Eastwood published his promising results today, the post follows nicely on a Twitter discussion I had earlier on the weekend. One prominent anaytics blogger was calling out a popular stats site for posting game recommendations from professional “tipsters,” those guys who have a lock on this, that, or the next thing. All he needed to do was post up their success percentage. A bitchy thing to do perhaps, but necessary.
I don’t think it needs too much stressing, but if you’re hoping to make money on the subjective recommendations of a guy on the Internet relying on nothing but his heuristic hunch and his self-proclaimed expert status, you’re going to have a bad time. But this got me thinking about ways to address sceptical attitudes about the predictive efficacy of analytics, and about the use and application of football analytics in general.
Let’s imagine for a moment a weird and implausible dystopian future in which governments forced citizens to bet all their earnings on football matches. Chances are, a lot more people would have a vested interest in football, and Bill Shankly’s claim that football is more than life and death would take on a whole new meaning.
Within this world, experts would come out of the woodwork with advice on how to make your weekly, government enforced football wagers. Some opportunist characters might claim you could make a quick buck fast by betting against the odds on one or two matches based on their advice. Others might go on the basis of pundits and opinion-shapers in the big papers. Still others would argue for a safe investment strategy that doesn’t stray far from the official betting lines. And then you’d get people like Eastwood publishing predictions based on an internal empirically-based betting model.
Remember: these bets are enforced. You don’t have the choice here not to play. Your entire livelihood depends on your ability to make safe wagers, and if you’re lucky, you can earn a little when the games break against the odds. Which approach do you choose?
Well, the smart investor would likely choose a proven, ever-improving predictive model provided by one of the analysts, right? Particularly as the tipsters’ success rates are all noise and no signal. While this should be obvious, even in our normal, everyday world, many investors make decisions based on gut feelings, some of which turn out well, and some don’t. This isn’t to completely dismiss subjective impressions in making investment choices, but to show that they cannot compete with a a solid statistical model built for long-term success.
Perhaps over time, those model forecasters would show the most consistently stable return on investment. If you were a football betting advisor, you know which portfolio you’d choose. You probably wouldn’t look too much into the model’s methods, and you’d stop watching football, confident it will all balance out in the long run.
This should be self-evident, and yet time and time again many in football choose to trust the subjective impression of journalists or TV pundits over the boring, hard evidence provided in by good statistical analysis. Remember: in betting analytics, statistics are not used to proscribe how football should be played. Rather, for the most part it’s more interested in football as it is already played. No one is currently or at least credibly calling for wholesale changes to how the game is played based on this kind of analysis, or at least not yet.
As long as football is won or lost on tactical decisions and good players with luck thrown in to make things interesting, there will always be a place for storytelling provided by chumps like yours truly. The problem is when chumps like me start sniffing their own farts and trusting their subjective impressions over the record of evidence. If your livelihood depended on it though, the choice should be obvious.