Dr. Bob recommends using a Kelly multiplier of less than 1 when the accuracy of one's winning probability is unknown, and one is risk averse. This limits the short-term win and loss streaks.
Another tactic is to adjust bankroll used for KC calcs. ie. if I start at 1k and get up to 3k, using raw KC you'd use 3k for the latest bets. But I could build in a delay in raising my bankroll for KC calcs. eg. I keep using 1k for KC calcs until I hit, say, 5k of "true" bankroll, that would also have a dampening effect on win/loss streaks as percentage of true bankroll.
Before I fire up any simulations, any thoughts on which is superior method, and what do other people do?
Another tactic is to adjust bankroll used for KC calcs. ie. if I start at 1k and get up to 3k, using raw KC you'd use 3k for the latest bets. But I could build in a delay in raising my bankroll for KC calcs. eg. I keep using 1k for KC calcs until I hit, say, 5k of "true" bankroll, that would also have a dampening effect on win/loss streaks as percentage of true bankroll.
Before I fire up any simulations, any thoughts on which is superior method, and what do other people do?