[sdenton4]

For things which have some combination of luck and skill, there tends to be a baseline skill for, average results given no/low skill. So in a thousand person event, you'll have some number of people performing at baseline and a smaller collection that have actually shown up with some skill. Depending on how high the skill floor is and how much variance there is, this often means people performing at baseline don't have any real chance of winning.

But! The variance still matters a lot for the skillful players: their chance of winning is 1/10 instead of 1/1000, and for the baseline folks the chance of winning is basically zero.

[jonahx]

I know you were probably putting those numbers in as placeholders to illustrate the concept, but worth pointing out the advantage that (even a lot of) skill gives is nowhere close to 100x over weaker players.

[csa]

> but worth pointing out the advantage that (even a lot of) skill gives is nowhere close to 100x over weaker players.

Not op, but I think you’re right… it’s probably 1000x or infinity depending on how you look at it.

The ev of the median player in a typical tournament is negative, while the pro is positive. As a measure of skill, that metric can’t really be expressed as a multiple.

As for skill level, I think a prob being 100x a rec is probably about right — most people have no idea how much better pros are than they are.

That said, poker sustains interest from recs precisely because the format leans more towards the luck side of the luck-skill continuum than complete information games like chess or go.

So a rec can play head up against a top pro like Phil Ivey and can still win, but that victory would be a function of luck rather than skill. Iterate that spot 100x or 1000x, and the rec doesn’t win very often.

[jonahx]

> So a rec can play head up against a top pro like Phil Ivey and can still win, but that victory would be a function of luck rather than skill. Iterate that spot 100x or 1000x, and the rec doesn’t win very often.

The variance in poker is extremely high, allowing long-term losing players to often have winning sessions in cash games, and to (often enough) cash and even win tournaments.

After the entry ticket is purchased (ie, not subtracting it's cost in your EV calculation), the long-term losing rec still has positive expectation in a tournament. So does the pro. EV_pro <<< 100 * EV_rec. This will be true even if the pro is Phil Ivey. Any professional poker player would much, much rather take, say, 50% of the combined winnings of 100 random recs in the WSOP main event than 50% of Ivey, assuming both were offered at the same price.

[csa]

> After the entry ticket is purchased (ie, not subtracting its cost in your EV calculation), the long-term losing rec still has positive expectation in a tournament. So does the pro. EV_pro <<< 100 * EV_rec.

ROI is a standard metric for measuring MTT success. Pros are typically positive, recs are typically negative.

Flipping the metric to EV after buy in strikes me as a straw man, but I will roll with it…

I am fairly certain that I would take action on the results of a basket of 100 mtt pros being 100x greater than the results of a basket of 100 mtt recs in the wsop ME. Like I said, I’m pretty sure that would be closer to 1000x, since the vast majority of the recs will not cash, while the pros are much more likely to cash and run deep. I think this is a decent proxy for roi. If you’re interested, I will be happy to negotiate terms and have a mutually agreed upon escrow-holder for a bet next year.

I agree with your comment that there is a lot of variance in poker, but that variance can be mitigated quite a bit via skill except in “short stack” MTTs that devolve into shovefests relatively early.

[jonahx]

> ROI is a standard metric for measuring MTT success. Pros are typically positive, recs are typically negative. > Flipping the metric to EV after buy in strikes me as a straw man, but I will roll with it…

That, as you said, makes it trivially true that the pros are infinitely better than the recs. By that metric it doesn't even make sense to have the conversation. The original comment I responded to was: "their chance of winning is 1/10 instead of 1/1000". My thought experiment, far from being a straw man, simply extends that idea from "winning" to "amount made in cashes."

> I am fairly certain that I would take action on the results of a basket of 100 mtt pros being 100x greater than the results of a basket of 100 mtt recs in the wsop ME.

Very unlikely. To give a sense, based on payout structure this year if even a single rec makes the final table (at least one did) that alone puts the number to beat at a minimum of 100M (9th place prize = 1M)... and first is 10M.

> but that variance can be mitigated quite a bit via skill except in “short stack” MTTs that devolve into shovefests relatively early.

No, even deep stack tourneys are still really, really short compared with anything like a "long run" sample. Jonathan Little, the pro who own https://pokercoaching.com/ and has millions in tournament earnings, once went 50 tournaments without even cashing, let alone winning. Daniel Negreanu, arguably one of the best tournament players of all time, recently had a 2 million dollar losing year: https://www.youtube.com/watch?v=Nu-QqzikpWU.

> I will be happy to negotiate terms and have a mutually agreed upon escrow-holder for a bet next year.

I will absolutely take that bet, but you'd need to define some way to select what counts as the pool of pros and the pool of recs. Alternatively -- this wouldn't be a bet but a way to test the theory -- look at the list of everyone who cashed this year and cross reference it with Hendon Mob db or something similar... and define a pro as someone who's had lifetime winners of 300K+ or something. You'd need to control for the number of pros among the entrants as well, but if you're right you'd see the pros represented at 100x their entry proportion in their cashes proportion (that's very rough and doesn't account for size of cash, but it should be an easy enough experiment to do if you want to see that I am correct).