The cover of last month’s magazine brought a tingle of pride to the guys at Bluff Towers. After having put James Akenhead on the cover before the WSOP last year, we’ve only gone and done it again. This time, we put Michael “The Grinder” Mizrachi on the cover, and then he went on to make the final nine.
The issue of making predictions and covering success in poker has become a bit of a crusade for me. It’s important because in my column I’ve always wanted to inject a bit of scientific analysis into everyday poker commentary. In a world where 99 to 1 for Phil Ivey to win the Main Event (when there were 2400 players left) seems like a good bet to some people, there’s certainly some call for more scientific analysis.
So let’s piece apart how prediction really works in the poker press and work out how tough it really is.
Lesson #1 – Even a stopped clock tells the right time twice a day
I don’t know how many of you thought that Paul the Octopus was really psychic (if you do, you may as well stop reading now), but for those of you who don’t, that’s lesson number one. Randomness can be peculiarly correct sometimes. In the case of Paul’s prediction of Germany’s results in their World Cup matches, even dumb luck can get it right eight out of eight times.
Eight from eight is only 1/256 (if guessing correctly was 50/50), and that’s assuming there weren’t any systematic biases in the way Paul was given the food. That kind of event is not a big deal. I’ve said it before and I’ll say it again: really rare events do happen, they just happen really rarely. However, our minds have evolved (cleverly, usefully) to drown out the noise of mundane events and concentrate on the outliers.
When it turns out that two of your friends have a birthday on the same day, your mind doesn’t instantly pipe up and say “well, there have actually been (on average) 364 instances of one of your friends having a birthday and other friends not sharing that birthday since the last time this happened. Your mind isn’t built like that. But it’s true, it has been that long, and you just need to get your head around it. Getting your head round stuff that is in someway beyond your “natural” thinking is exactly what science is for.
Lesson #2 - Hagiographies
Philosopher Daniel Dennett has a nice anecdote for how reportage of a famous name can change people’s perceptions of how they are. While touring the White House, a guide showed him a portrait of George Washington. “This probably isn’t how George Washington looked back then,” said the guide, “but it is how he looks now.”
The phenomenon is most common with the recounting of the lives of Saints. These accounts, known as hagiographies, are not just complimentary, they spin a yarn which ultimately exults the person to heavenly status. And why wouldn’t they be of heavenly status? These guys are Saints! The process is circular.
So it is with poker players. No narrative ever written in the poker press would say of a player: “he binked a big one, got a sponsorship which bought him in for free to a number of major tournaments, and over the years he picked up enough cashes to make his resumé look like he’s half decent.”
Lesson #3 – Selection bias
Selection bias is when a set of subjects is chosen (either accidentally or deliberately) for their likelihood to show a positive result for one’s hypothesis.
Of course, in the poker press, selection bias makes sense. We concentrate on those most likely to succeed (it makes us look cool if we predict their success). Only in the social sciences would you want a really random sample to see what the average Joe does. Personally, I’ve always been fascinated by how the average Joe fares in poker. Admittedly, we do have Sharkscope, PTR and the Bluff db for that.
What’s staggering in poker – and I’m not sure anyone’s really picked up on this yet – is just how many hagiographies there need to be in our discipline before it looks like a meritocracy. At the Series or the EPTs, poker journalists hang around discussing who’s gone deep in a certain event. Everyone will be there reeling off names knowingly, and I must confess half the time these names draw a blank. “Oh, yeah, he won the EPT such-and-such,” they’ll say; thus his legitimate status is reified.
Why do we bother? Let me spell this out: we all really want poker to be a meritocracy. Everyone does. The pros want it because it means they’re not wasting their lives dedicating themselves to an empty pursuit. The websites want it because it means that poker’s a sport and not gambling. The journalists want it because otherwise their career would just be about producing reams of information about something which amounts to little more than caprice.
The good news is that poker is a meritocracy. Phil Ivey really is better than me, and I really am better than Mr Fishy. The bad news is it’s not that much of a meritocracy. If Phil Ivey, myself and Mr Fishy played a three person tournament with a standard blind structure, I’d reckon about a 55/30/15 split of wins for us respectively. We can argue the toss for this incessantly, but before we go off on that tangent, consider the following. Imagine Tiger Woods, a decent club player and myself played a round of golf together. Fancy a punt on the split of wins? I’d go for 98/2/0, or maybe better in favour of Woods. I’m simply never winning. Ever. Now that’s a meritocracy.
To repeat my earlier point, in order to make poker look more like a meritocracy, we have to have many hundreds of household names in the game (go on: try to name as many as you can; if you get lost, take a look at the list of Full Tilt red pros). That way, when a famous name wins a tournament, we can satisfy ourselves that our game is a meritocracy so that in our sport, just like in golf, most of the time the cream make it to the top.
Lesson #4 – Survivorship bias
The final problem is in my opinion the biggest one of all. Survivorship bias is most famously practised in the financial sector. Banks will start up scores of funds at the beginning of each year and plenty of them will do very badly. The badly faring ones will either be dissolved, merged into others, or the fund manager will be fired and the fund will be renamed. This way, only the results of the surviving or well-performing funds get included in the results.
To say that this happens in poker would be somewhat of an understatement. Allegedly, around one or two percent of all poker players are money making. Imagine a poker magazine filled with the hagiographies of those who didn’t make the grade, who didn’t bink the big one. It would probably fill an entire shelf in a library.
Thousands of people leave poker each year slightly in deficit. Don’t get me wrong, these aren’t tragic cases who have lost everything – those are very rare in my opinion. But, they’ve lost enough relative to their own means to quit. Most of them are probably soured enough by their experience that they will not come back. You know enough about variance now. As a mini thought experiment, try to imagine the volume of them that ended up on the poker scrap heap, not by virtue of bad play, but through running bad.
Granted, it isn’t as bad as I’ve painted it. Poker has such a pull these days, and can remunerate its glitterati so well, that financially it may well be one of the best things a hot young maths graduate can do with their time – especially when jobs are thin on the ground. The game has for a number of years attracted some fine minds who have worked their arses off to be good at their discipline.
My point is that if poker were as much of a meritocracy as we present it, there would be fewer of these “big names” and they would win major events more often. We wouldn’t be expending so much energy making these names big in the first place, and we wouldn’t be sweeping so many after-the-fact failures under the carpet.
Next month in part two, I shall analyse a set of players who, in the pre-WSOP issue, Bluff Europe made predictions about. We suggested ten “WSOP rookies” to follow who were big in the internet game, but were thus far too young to play at the Series. These hot tickets are the ones touted to be the big names of the next generation. Some, like Annette Obrestad and Daniel “djk123” Kelly, had already made it.
To study how these ten got on after the fact is “good science”. When you make a prediction, you have to follow it up with a report on how your prediction turned out. All too often, predictions are bandied about in the poker press and swept under the carpet when they aren’t realised – a most egregious case of survivorship bias. I know I can’t change the poker press, but if I can make one little corner of it a bit more scientific, I’ll be happy.
This article first appeared in Bluff Europe magazine.