I'm interested in the history behind risk of ruin calculation in BJ, and why people use this over Kelly criterion. Both have significant flaws, but I'm curious why people consider RoR the better statistic.

The flaws of Kelly criterion, as I understand it, are that it can produce unrealistic figures. The theoretical underpinning idea that you can't lose 100% is broken by table minimums. Naive Kelly criterion doesn't account for sometimes needing to play at a disadvantage, to get to hands where you have an advantage.

The flaws of RoR are that it assumes that you will continue betting the same amounts down to the final dollar. If you or your investors pull the plug, or even change your bet unit, you lock in a partial loss. This could lead to a very serious misunderstanding of how much you may lose.

What's less obvious, and a flaw in both, is that tapping/using the profits invalidates the calculation. Even if you only return to where you started (e.g. take out the profits), the underlying stats incorporated the chance that you would be in that profit position,.and assumed that position would be more secure than you're making it by returning to square 1.

An ideal model would incorporate a full decision process of when you will resize bets, when you will draw down profits, etc. But in the absence of such a model, why is RoR preferred?