My Gambling Thesis

In "A Brief Intro to Gut Forecasting" I talked a bit about what information we draw on when making predictions. Let's be honest, if someone asks you how you think a game will go, you can usually answer within a second. Thought only occasionally actually comes into it. But what I now want to know is - what good is thinking? How does it really help? 

The best-selling book turned Brad Pitt flick "Moneyball" is about how using a numbers-dependent approach to player acquisitions allowed Billy Beane to take the Oakland A’s, with a relatively tiny budget, and make them competitive against the big guns (the equivalent of making Stoke competitive against Barcelona). The key to the "Moneyball" approach was recognising that the most important stat for any baseball player, the one you prize above all others, is something called On-Base Percentage. In essence, this stat was found to correlate most strongly with victories. Knowing that it was OBP, rather than bicep circumference or body shape, that really mattered, the A's were able to pick up a squad of players disregarded by other teams for their unusual technique, their oddly shaped monobrows, or their club feet. What this shows is that predicting outcomes becomes less of vague, if you can figure out what numbers matter, and how to use them. 

So, let’s apply this to the sport we love. Let's go with Shots on Target - did it matter to Monaco that they had none when they qualified ahead of Arsenal for the UCL last-eight? Clearly not. Did it matter the next evening for Leeds when they had 8 shots to Fulham's 27, and beat them 3-0? What about form? Did Villa's 4 goals away from home all season tell you they were about to stuff Sunderland 4-0 on their own turf last weekend? I'm just saying, the numbers are useless unless you know which ones matter, and what they're telling you. Numbers alone aren't worth bubkiss.

So, we arrive at the point. You can crunch the numbers, check the form guides, you might even come to believe that it's the weather that best predicts outcomes. But really, nothing reliably does. You're just guessing anyway, and crunching the stats doesn't really get you any closer than just throwing your hat on a stick (unless you know which numbers you really need to crunch). At the start of any game you have three outcomes - home win, draw, away win - and you don't know, and you can't know, which is going to occur.