[ QUOTE ]
The concensus is that bankroll management (money management) is overdone on 2+2. Reviewers tend to say this, in OP comments about most bankroll-related articles.'
[/ QUOTE ]
I tend to agree with Ghazban that the articles reach the choir, and not the people who are saying, "Tell me 10 buy-ins is plenty so I can skip levels," or, "Tell me 1000 BB is needed so I can blame my 500 BB downswing on luck." However, bankroll managment is simpler than the way it has been presented in articles here, and much has been said that is irrelevant or incorrect. There is still room for improvement.
[ QUOTE ]
The article focuses on bankroll growth strategies and does not mention much about the downside Risk of Ruin aspect. RoR in fact is not mentioned at all.
[/ QUOTE ]
Risk of ruin depends on what you will do when you hti a down swing. When will you move down, and what would that do to your win rate?
The artificial ROR models do have some use. They tell you about the probability of downswings of various sizes. But they don't actually say much about your risk of ruin.
[ QUOTE ]
The last item is Optimal F. Optimal F is a fixed-fraction betting method that optimizes bet size for net profit. As such it is an important BR approach to understand. It’s a much more aggressive approach than Kelly. Optimal f optimizes theoretical net profit but results in huge BR swings.
[/ QUOTE ]
My understanding is that optimal f is not much different from a fractional Kelly system, and that agrees with the article you cited.
Optimal f does not optimize expected profit. No bankroll management strategy does.
The article should have been shorter and simpler:
<ul type="square">Bankroll = c * SD^2 / WR, where c is the reciprocal of the Kelly fraction you use. I call c the comfort level. Most people are happy with a value of c between 2 (half Kelly) adn 4 (1/4 Kelly). The standard deviation for full single table tournaments is known to be about 1.7 buy-ins, so plug in various values to Bankroll = c * 1.7^2 / ROI. E.g., if you are aggressive and have a ROI of 30%, Bankroll = 2 * 2.89 / 0.3 ~ 19 buy-ins.[/list]This is simple, it is consistent between advantage gambles, and it take into account the variance conditional upon cashing, which the article did not. It does not introduce the irrelevant ITM percentage. The article's calculations also assume c=1, which is quite unpopular.
Why do people recommend 50 or 100 buy-ins for SNGs? First, some advice is used out of context. Expert win rates are lower in tougher games. A 6% ROI in $215 SNGs is fine, and someone who is working on the $5.50 SNGs should pay close attention to the advice from a player who is beating high stakes games. However, the number of buy-ins needed does not translate. A ROI of 30% means you need only 1/5 as many buy-ins to be as safe.
Second, some people want something different from a bankroll than a low risk of ruin. They want to be able to cover the worst downswing they have every heard of. If someone who does not have a track record of being a solid winner at a level reports losing 72 buy-ins, these players will say you now need 73, regardless of your skills or the level you are playing.
It may be interesting to some people to see that some of the same ideas show up in other fields such as finance. Football involves risk and reward, too, but football fans/players/coaches are among the least receptive to logic, and they have plenty of entrenched bad ideas such as punting on 4th and 2. The growth vs. value example was poor. It's a continuous tradeoff, not a dichotomy. I'd like to see better selection when reporting the common wisdom of other fields as something to emulate.