Get It Quietly

Football, bollocks and a bit of poker if you're lucky.

Location: Enfield, London, United Kingdom

Saturday, September 04, 2004

Even More Fun with Spreadsheets

Thanks to Dan who sent me the following comment on my recent bankroll post :

"The $500 bankroll for $30 (50/30/20) SnG tournaments seems too low. The session risk is the buy in ($30+3) implying session risk equal to huge 6.6 percent of bankroll ... would not Kelly imply a larger bankroll (around $1700) so session risk says closer to 2%? I realize bankroll is a function of edge under Kelly; even so 6.6% session risk seems very "lively" to me. "

I promised to go and look up the Kelly criterion as it applies to this case and I will. In a bit :-). I'm not entirely sure how the theory applies to Sit and Goes, particularly in small samples. There are only 4 basic outcomes (if three places are paid). What I did was to fill in probabilities of each finishing position and give it a good old Monte Carlo spin. The figures I plugged in are as follows :

P(1st) = 18%
P(2nd) = 15%
P(3rd) = 21%

which equate to an edge of just under $15 for a $30 Sit and Go (after the rake), and I'm figuring to play 2 of these per hour by playing two at once, usually overlapping, by which I mean starting the second game when the first is half way through. Anyway, when you give these a spin over 20,000 trials, the maximum downside does vary a bit, but it's almost always between $400-$600. It's not a rigorous bankroll definition but it's probably close enough.

Dan went on to add "I suspect a decent living can be made multi-tabling $200 SnG for 5-6 hours a day if approached carefully with respect to session risk." I would suspect a lot more than decent if you're good enough, although that's much too high for my taste. The real import of my figures below is how they compare to each other. There is no doubt in my mind that if tournaments are your bag, Sit and Goes are by far the best bet on-line. The simple fact that you can play so many more of them in the same time elevates them high above multi-tables in terms of hourly value, and they are subject to a much smaller variance. Cash games I wouldn't know, but I'm sure that the required bankroll for any cash game where you expect to make $30/hour is much higher than $500 !


Blogger butch said...

I'm not sure this definition of a bankroll is a particularly useful one. If you are playing with $500, go down to $100 and continue playing, you are then accepting a large chance of going broke. I'd prefer a bankroll that let me experience one of these swings your simulations show still leaving me with enough to continue playing.

In regards to your comments on cash games, I think you would be surprised as to what bankroll you can live with. IMO the lowest varience route to making $30ph by cash games would be multitabling the lowish limit NL games. Playing 3 0.5-1 NL games with a $50 buy could achieve this win rate, and I would expect a swing larger than $500 to be very rare. But again I would recommend at least $1000 bankroll so you can cope with the downswing if it happens.

2:52 PM  
Blogger Andy_Ward said...

This comment has been removed by a blog administrator.

9:32 PM  
Blogger Andy_Ward said...


I see what you're saying - but I think you could say the same of any theoretical bankroll requirement. Any time your roll is below the requirement, you are in jeopardy of going broke, independent of whether you started with enough before a losing streak.

But perhaps we're only quibbling over definitions. As I said, what's really interesting is the comparison of the figures for different games.

Regarding cash games you may well be right - if you play them regularly then you're in a much better position to comment than me.

Thanks for taking the time to post,


9:49 PM  
Blogger butch said...

I think you agreed with me more in your first post ;). Maybe I am being a bit pedantic and I appreciate that this wasn't the real point of the simulations. I guess it comes down to what we want our bankroll to represent, and within the poker world I don't think people are consistent here.

I tend to think of a bankroll as being an amount big enough that I have a close to zero chance of going broke. Other people think of it as being the biggest downswing they are likely to expect at the game they play. As I think we will agree, these two things are different ( the second version would result in me going broke, the first lets me continue to play the game if I experience a bad losing run.) I guess what I was trying to say in my first post is that I think quoting the second one as a 'bankroll' figure can be misleading, particularly to new players.

I'd also point out that it depends very much what you do with your winnings. If you cash out down to 500 at the end of the month and spend them then it will not be enough to avoid going broke in the long run, if you use them to build your bankroll it probably will be sufficent to keep playing the game indefinitley. And of course you always have the option of adding some more money from otehr jobs etc.

Finally from an EV point of view, it probably makes much more sense to play in games where you don't have enough cash to play indefinitley. After all, if you lose half your roll you can just drop down to playing half the stakes etc. But if we approach this from a 'bankroll required to make $30ph indefinitley' point of view, I think we should quote $1000.

In terms of cash game experience, I have around 1000 table hours logged myslef and have discussed this various people with a similar number. The hardest thing is not to get bored and move up too quickly.

By the way if you don't play cash games, who is the A ward on stars with a QPR avatar?


1:18 AM  
Blogger Andy_Ward said...

I may come back to your bankroll points when I have more time - then again, this isn't a forum where anyone has to have the last word or "win" the argument. I'm happy for people to read all the comments and make up their own minds.

As for stars, I was going to say "Small world" when I thought is it really that small or is someone pretending to be me ? I really hope not. Not out of a sense of outrage but a sense of unease - it would be like having a stalker ! Anyway it isn't me.


6:33 PM  

Post a Comment

<< Home