Hooligans Plays Baccarat

Kelly Calculator

  • Start date •
  • Replies
    19 Replies •
  • Views 3,201 Views
Follow
Housemoney

Housemoney

Well-Known Member
Since
Jan 30, 2010
Messages
85
Score
0
Tokens
0
Do the pros on here strictly adhere to the Kelly bet sizes? I went back and read a lot of Ganche's post last night and it makes me think I need to change my betting habits.

I like the over in the Canada/Russia game, but Kelly is saying it is a no play. I guess because of the increased juice vs. the win probabilty.
 
-125, equates to a win probability of 55.55%, as such the calculator is correct.

My staking plan is relatively conservative (equivalent to about one third Kelly) and whilst it works on the same principles as Kelly, it levels out more marginal figures.
By way of example Kelly would advise a very aggressive bet at -3300 on a 98% outcome, whereas I would be much more cautious
 
The biggest lesson to be learned with Kelly is to not overbet your edge. Many people (including myself) use a fractional variation of Kelly to reduce risk. If you underbet, you don't maximize growth. However, if you overbet, you drastically increase your chance of ruin.
 
Dunder
-125, equates to a win probability of 55.55%, as such the calculator is correct.

My staking plan is relatively conservative (equivalent to about one third Kelly) and whilst it works on the same principles as Kelly, it levels out more marginal figures.
By way of example Kelly would advise a very aggressive bet at -3300 on a 98% outcome, whereas I would be much more cautious
Click to expand...

I was reading some old Kelly threads today and I noticed that most people seem to take a conservative approach to their apparent Kelly edge.

is one third a good starting point or is that too conservative?
 
Last edited:
illmatick
I was reading some old Kelly threads today and I noticed that most people seem to take a conservative approach to their apparent Kelly edge.

is one third a good starting point or is that too conservative?
Click to expand...

That question is so dependent on a lot of thing. A lot of people go very conservative on their Kelly % because they're not really predicting their edge, they're using the historic edge for whatever method they have. At best, you would expect this edge to go down slightly, due to the fact that markets become more efficient over time. At worst, there are issues with not testing out-of-sample, overfiitting, etc, and the actual edge is much lower or non-existent.

Assuming that a person has done a great job predicting their edge, then it's really a matter of goals. Full Kelly can be great, especially for someone with supplemental income who has the goal of rapid growth and the stomach for big swings. Somewhere around full Kelly can be the only reasonable option for a small starting bankroll if you need to make enough to pay expenses and still grow. For most people with a more comfortable bankroll, fractional Kelly < 1/2 seems to be where they find their comfort level.

I played at full Kelly for about a year early on. These days I'm between 1/4 and 1/3 Kelly when possible.
 
If you have a pretty good handle on your edge and the number of bets you make, one of the best things you can do is just simulate a years worth of bets over and over at different Kelly fractions. That way you can get a feel for the swings at the higher %s, and the sacrificed earnings at the lower %s.

I think that's the easiest way to predict ahead of time where your comfort level lies.
 
I say I bet .25% kelly, but really it's a lot less, mostly due to the fact that a lot of the stuff I bet has small limits. It winds up being that the only things I can bet 1/4 kelly on are my tiniest edges, which causes some issues in growing my bankroll.
 
You must log in or register to reply here.