Mayor, I'm baffled by how the numbers can be so high in this analogy regarding ROR. For example: Your numbers, ROR 1/2 = 31.6 and ROR 1/4 = 56.2 . Without knowing how to do the math for this, and based on the direction your numbers are headed, I will estamate the ROR for 1/8 to be about 70%. How is this possible if we are playing with 0% - 2% edge? Can variance be this bad? Maybe that's why I've been on the "sucking eggs" side lately.
the risk of losing half the bankroll for Kelly bettors is 33.3%, which is pretty close to your 31.8% for a ROR of 10%. Does this mean that the ROR for Kelly bettors is close to 10% ??? I am confused because I thought that Kelly bettors would have a ROR of 0%.
True Kelly betting has you resizing your wager after each bet based on your current bankroll size and current edge. In this case the risk of ruin is 0%.
The typical way ROR is determined is that one computes optimal Kelly betting once, based on a fixed bankroll, then one does not resize bets at any future point. Without resizing, there is a real possibility of losing the entire BR. It is the latter that is typically assumed when discussed ROR or trip-ROR.
But won't any betting system reduce your risk of ruin to zero if you can always reduce your bet based on your bankroll? For example, I could set up a system where my max bet equals 1/16 of my bankroll. That way, I can split 4 times, double on each hand, lose all bets and still only lose half of my bankroll. What makes Kelly any better than that?
Kelly is better if your goal is to maximize the expected growth rate of your bankroll. Lots of fractional betting schemes keep your bank above zero, if you had infinitely divisible capital. But Kelly gives you the fastest growth rate, meaning the fewest average hands to double your bankroll.
there are some excellent posts by some very good players at BJ21.com and bjmath.com, as well as the paper by Kelly on proportional betting on bjmath.com. I imagine there may be some fine posts here as well but I am not as familiar with this site.