Secrets of the Amateurs

Final results are now in. A couple of events haven't quite finished, but none of the selected players are still in so we're basically done. Before we start I want to put a huge qualifier on this. I've been watching a lot of the World Cup (obv), and it's quite tilting how often people come out with statistics that are based on a completely insignificant sample. There should be someone sat in the corner saying "Sample size ... sample size ... sample size". Only on one occasion did I hear someone point this out. During the Argentina-Mexico game, the Radio 5 commentator said "Mexico have scored all three of their goals in the last half hour of games in this tournament". David Pleat, bless him, replied "That's just a quirk". Stunned, the commentator simply repeated himself by saying "Well, it's a fact that all three goals were in the last half hour", entirely missing the point which is basically "so fucking what".

At least goal timing is something that might possibly be correlated to something real, for example England have scored many more of their goals, proportionally, in the first half of games than the overall average, in recent tournaments, which you could definitely argue is correlated with their inability to pace themselves during games, ie they run around like madmen for an hour and are out on their feet for the last 20 minutes. Pundits have also been saying stuff like "The reigning European Champions have never won the World Cup" as if there could be any possible negative correlation between the two. Any at all. I certainly can't think of any possible reason why this should be true, and if you can't do that you should be suspicious of any such conclusions drawn from small data samples.

So bearing all that in mind, we can very cautiously address my WSOP results. Here they are, in a nutshell :

TV Pros : 465 tournaments, $2.87m buyins, $1.95m returned, down $920k

Euro Pros : 323 tournaments, $1.33m buyins, $2.10m returned, up $770k

Internet Kids : 416 tournaments, $1.59m buyins, $1.03m returned, down $560k

TOTAL : 1204 tournaments, $5.79m buyins, $5.07m returned, down $720k [1]

So Euro pros win the day. Raise the 12-starred flag ! Overall, 1200 tournaments isn't a ridiculously small sample ; it's moving towards something that's at least a little significant. And bear in mind that the biggest single win by any player was Praz's $515k. In order to have recorded a collective ROI of 100%, which seems the minimum of what many people believe WSOP ROI to be, generally, the group would have had to find eleven more $500K scores. Let me put it this way. If the 20 TV pros had trebled their return, they would have wound up at +100%. I can guarantee you one thing - if that had happened, if for every "TV Pro" result there had been two more of the same, the media would have creamed their pants about "The Year Of The Pro".

I'm not going to comment on individuals who had poor WSOPs (tempting as it is), it just seems mean. You can Sharkscope any of them if you're interested. It did seem to me as the series progressed that people who were playing every tournament they could were struggling. We can definitely correlate this I think ; there must be issues with tiredness, and people who were multi-tabling for whatever reason (IMO it's burning money unless you have huge bracelet bets to make it worthwhile) were bound to suffer for it. In addition, "Winner take all" bracelet bets are going to negatively affect EV as players crash and burn rather than picking up place prizes.

Of 24 players who paid more than $100k in buyins, only three made a profit. And two of those (Mortensen for $12K and Kravchenko for $7500) were small. The exception was, of course, the superhuman Phil Ivey who made $107K profit. This is slightly skewed because Men Nguyen had $88K in buyins and made $700k+ profit, being the biggest winner of all by a distance.

Overall it was an interesting thing to do and in conclusion I think I'll put it this way. The exercise has done nothing at all to alter my belief that players' $ expectation at the WSOP is much lower than most people believe. I believe that the good players aren't much better than the field and there's a huge randomising effect because of simple tournament variance. For the rest, consult Taleb's "Fooled By Randomness". See you next year !

[1] If those don't add up exactly it's just rounding errors

Update : One final note. I'm banging on about not reading too much into results from 1200 tournaments, whereas this is the understanding of sample size more prevalent among poker fans (from 2+2) :

"Can people stop saying either Dwan OR Hellmuth is bad at PLO? You can criticize specific plays they make-- anyone can make a bad play-- but they both made a big run in this tournament and Hellmuth also finished 7th at an earlier PLO event at this WSOP. I think that pretty much establishes that they are both skillful players."

Hellmuth finished 7th and 11th in two of the six PLO/PLO8 tournaments this year. Better still, Durr finished 17th in one. And that establishes that they are skilful Omaha players. Sigh. Of course, they still could be good PLO players (and Durr almost certainly is) ... then again I should probably be grateful, because if everyone understood things like sample size and variance then there really would be no money in poker, everyone's solid.