Hi all
I am pursuing a few strategies and lets hypothetically assume I have a 3% edge. Looking at staking I am researching a quarter kelly system. However I wondered how to compensate for the fact that by definition winning bets will be smaller and losing bets larger. If you win then your bankroll is higher so by definition your next losing bet is with a higher stake. I'm not sure how to quantify this mathematically or how big a bite out of the edge this will take?
Kelly Staking Question
How do you know your small bets will win and your big bets will lose?
Wouldn't it be simpler to just make the smaller bets the larger bets, or have I misunderstood you?
You treat your bankroll as a fixed number, you shouldn't increase it just because you have had one large winner but when you want to stake larger.Cards37 wrote: ↑Sun Feb 18, 2018 3:03 amHi all
I am pursuing a few strategies and lets hypothetically assume I have a 3% edge. Looking at staking I am researching a quarter kelly system. However I wondered how to compensate for the fact that by definition winning bets will be smaller and losing bets larger. If you win then your bankroll is higher so by definition your next losing bet is with a higher stake. I'm not sure how to quantify this mathematically or how big a bite out of the edge this will take?
- marksmeets302
- Posts: 527
- Joined: Thu Dec 10, 2009 4:37 pm
Kelly doesn't care about the distribution of you wins and losses. Whether you have many small losses and occasional big ones or vice versa, the optimal fraction of your bank to risk is given by the kelly criterion. Assuming you strive for optimal growth of you wealth.
It is wise to use a quarter kelly system. The overestimation of your edge is punished much harder than the opportunity loss resulting from an equal underestimation. On top of that the psychological aspects can be brutal. You found this out already: after increasing the stake you fear a large loss, throwing you back significantly. Still, applying full Kelly on a perfect estimation of your edge will give you the highest growth of your wealth in the long run.
Unless you are straight out backing or laying I think Kelly is impractical in the betfair markets. They quickly become saturated and often the strategies can't scale any further.
It is wise to use a quarter kelly system. The overestimation of your edge is punished much harder than the opportunity loss resulting from an equal underestimation. On top of that the psychological aspects can be brutal. You found this out already: after increasing the stake you fear a large loss, throwing you back significantly. Still, applying full Kelly on a perfect estimation of your edge will give you the highest growth of your wealth in the long run.
Unless you are straight out backing or laying I think Kelly is impractical in the betfair markets. They quickly become saturated and often the strategies can't scale any further.
Because when you win your next bet is larger because of bankroll growth. So when you have a loser it will be a bigger bet. On average therefore your losing bets will be with bigger stakes than winning bets.
That's assuming that each winner is followed by a loser. If you have a winner followed by a winner it will be a bigger winner, and if you have a loser followed by a loser it will be a smaller loser!Cards37 wrote: ↑Sun Feb 18, 2018 9:55 pmBecause when you win your next bet is larger because of bankroll growth. So when you have a loser it will be a bigger bet. On average therefore your losing bets will be with bigger stakes than winning bets.
Your bet following a winner will only be significantly larger if you backed a short-priced horse with a massive edge. Your hypothetical 3% edge, and the fact that you are (quite sensibly) considering using quarter stakes, should mean that your percentage stake in relation to your bank will be quite small. So your next bet would only be slightly larger.
The Kelly criterium takes care, mathematically, of your concern. You just need to remember that the Kelly criterium is only mathematically sound when you know for certain what the true probability of winning is. When betting, the skill really lies in either accurate judgement of the true probability of winning, or good judgement of how much error there may be in your judgement.
When I was betting I used the underlying principle of Kelly and adjusted my stakes according to level of confidence in the accuracy of the price I arrived at, but using quarter Kelly is another way to be cautious.
Good luck.
Edit: My 800th post.