Please change the FAQ on bankroll requirements.

[El_Hombre_Grande]

Many of the posters here are responding as though anecdotal rumor trumps math. It doesn't. Undoubtedly Pzhon is correct. The real answer is a) bankroll requirements assume that you are a winning player (a dubious assumption for most) and b) the amount of a sufficient bankroll depends on a number of interrelated factors: your proven winrate, standard deviation, and tolerance for risk.

I would agree that the FAQ could be more much more "correct" by explaining this analysis. But I think it would be a shame also if a newbie couldn't leave the FAQ with a reasonably solid and useful answer as opposed to a "it depends." If all its going to do is theorize, its not really a useful FAQ. 20 is in some sense "wrong," as Pzhon points out, but its not nearly as "wrong" as many of the ideas that can float around in a newbie's head.

[pzhon]

[ QUOTE ]

Well I suppose if you keep all your winnings as part of your bankroll, you don't really have a 10BI bankroll at all, do you?


[/ QUOTE ]
You might when you start, and when you are considering starting to play at a level (or leaving it) is when you are most concerned with the adequacy of your bankroll.

There are alternative methods for determining the size of bankroll that you need. Instead of aiming for a low risk of ruin assuming that you never move up or down or withdraw (these assumptions do not have to be true), you could assume that you withdraw all winnings above your starting amount, and aim for a low risk of ruin within a fixed number of hands, say a million. The math is harder, involving differential equations instead of algebra, but it can be done. I have been thinking about writing an article for the 2+2 magazine on this approach.

Regardless, any sensible method will say that you need a larger bankroll to be safe when your win rate is lower, as it is in tougher games like NL $400, rather than in ridiculously soft games, like NL $2. The FAQ says you need a fixed number of buy-ins, so it should be changed. If you like a number for NL $25, then you shouldn't like that number for NL $2 or NL $400.

[pzhon]

[ QUOTE ]
the guideline was meant to stop the endless luckyjimm style of moving up with just a few buyins then busting the whole roll and starting over again.


[/ QUOTE ]
Really? I know the origins of the 300 BB guideline, but I didn't know the 20 buy-in guess (which is far more conservative than 300 BB in typical online low stakes games) was based on anything. Shouldn't the FAQ give correct information instead of trying to stop people from doing something?

[ QUOTE ]

anybody that wants to be more aggressive with their moving up is going to do it regardless of what the FAQ says.


[/ QUOTE ]
The FAQ should not mislead people about what is conservative and what isn't. It should give correct information, and then people can make their own choices.

At NL $0.01-$0.02, if you win 20 BB/100 (much more is possible) with a standard deviation of 60 BB/100, then the FAQ's suggestion of 5000 BB (20x250 BB buy-ins) will give you a risk of ruin under 10^-24. That's like the chance of winning 3 lottery drawings in a row. Isn't that a little too conservative? It's ridiculously more conservative than playing NL $400 with 20 buy-ins.

[ QUOTE ]
if we're trying to protect new players from going bust, id say being more conservative with the FAQ is the way to go.

[/ QUOTE ]
Give correct information, and let people rationally risk busting out if they want to.

Bankroll management assumes busting out is a disaster. However, for most microstakes players, it isn't a disaster to lose their balances. It's an inconvenience. For a midstakes player, it is much more serious problem, since it is less common for people to have an extra $10k lying around than $100. So, the mathematical guidelines are already quite conservative for microstakes players.

When you overstate how frequently large downswings occur to winning players, you are telling people that a large loss is a weaker indicator of poor play than it really is. You also discourage people from taking shots at higher levels when they are ready, and when they would have a low risk of ruin.

Rational bankroll management is not about minimizing risk. To minimize your risk, quit poker. Bankroll management theory should help you to decide whether the added risk of playing in a higher stakes game is worth the added profit you expect to make. To make the correct tradeoffs, don't overstate the risks, or pretend that NL $400 is just as soft as NL $2.

[pzhon]

[ QUOTE ]
pzhon is correct. If he wants to offer up something short and easy for noobs to undersatnd as a new bankroll FAQ, I would seriously consider adding it to the FAQ and sticky as long as AJ, isura and orange had no objections.

[/ QUOTE ]
Thanks, I'll work on one.

[traz]

If this is going into a sticky, I think it needs big red disclaimers everywhere, otherwise alot of people are going to go busto (even if it's their fault). What they're going to take away from it is "oh, well if winning players only need 10 buy-ins, then thats all I need too"

[djg]

http://forumserver.twoplustwo.com/showfl...rue#Post8465243

http://forumserver.twoplustwo.com/showfl...e=0#Post8449205

[blah-blah-blah]

obviously if you can do the math correctly the bankroll you need to play different limits is going to change according to your winrate etc. having a 10 bi role is fine if you move down whenever you get down to 5-6 bi's. I think if a change were to be made to the FAQ, it should include some of the math and the fact that bankroll requirements will change as you move up, but also keep the part that says "as a convention people usually say to keep 20bi's to the limit you want to play."

[yad]

Everything pzhon has said in this thread is obviously correct.

I do think that whatever goes in a revised FAQ should make clear what the goals of optimal BR management are, and what they aren't.

[pzhon]

[ QUOTE ]
If this is going into a sticky, I think it needs big red disclaimers everywhere, otherwise alot of people are going to go busto (even if it's their fault). What they're going to take away from it is "oh, well if winning players only need 10 buy-ins, then thats all I need too"

[/ QUOTE ]
You are suggesting that correcting the FAQ would be dangerous. How would it be more dangerous than the current situation? If you think a corrected FAQ should require disclaimers, then you should be even less satisfied with the current entry.

A losing player who thinks a bankroll gives him skills will bust out with 20 buy-ins. Do you think a lot of players have been busting out by following the current FAQ entry? The FAQ should say that bankroll management is for winning players. It doesn't say that now.

[pzhon]

[ QUOTE ]
http://forumserver.twoplustwo.com/showfl...rue#Post8465243

http://forumserver.twoplustwo.com/showfl...e=0#Post8449205

[/ QUOTE ]
What points do you think these threads make? They don't speak for themselves, and it's easy to jump to an incorrect conclusion from a quick reading.

For one thing, while most people will look at the graphs ZBTHorton gave and assume that he has established that he is a solid winning player before hitting that 40+ buy-in downswing, that's not true. Just as I mentioned happens in stories of downswings, he changed from the relatively soft Party games to the shark-infested games on PokerStars.

ZBTHorton ran at 5 PTBB/100 for 40,000 hands at NL $200, after which a rough 95% confidence interval for his win rate ranged from about 0.5 PTBB/100 to 10 PTBB/100. I'm sure he'd like to believe that he was neither lucky nor unlucky in that period, so the 5 PTBB/100 would represent his true win rate, but it's likely that he was lucky in the first 40,000 hands, and unlucky in the next 40,000 hands to lose it back. It's much more common to see a result that is +2 standard deviations followed by -2 standard deviations (or +1.5 followed by -2.5) than to see par followed by -4 standard deviations. If his true win rate is 2 PTBB/100 in those game conditions, then in some sense, a 40 buy-in downswing is about as surprising as it is for a player who wins 8 PTBB/100 with the same standard deviation to have a 10 buy-in downswing.

As for the other thread, there is a lot of unsubstantiated speculation, but don't miss this comment by Pokey: "As a result, the simulated paths should trend upward more sharply and have far less volatility than the Poker Tracker numbers would have us believe." I'm not sure I agree overall, but that particular factor Pokey identified means that the mathematical models may still be overly cautious, despite recommending 10 buy-ins or less in some cases.