Credit Spreads with the Kelly Criterion
- Thread starter dcwriter2
- Start date
Robert Morse
Sponsor
I’m not familiar with this Kelly criterion but you do know that your assumptions above are near impossible.
I think its important you figure this one out on your own. It's quite a simple problem. You just plug in the numbers. If you cant figure it out, your research ability may not be up to par for trading.
Maybe edit your post to show what you have tried. Writting out an attempt seems to help me
Maybe edit your post to show what you have tried. Writting out an attempt seems to help me
So you'll lose $4,620 every 20 trades on average? No money management will help with that. From what I know about Kelly, it makes more sense for positive reward:risk trades, not picking up dimes in front of a high speed train (credit spreads).
Credit spreads do not need to be picking up pennies in front of a freight train. OP's numbers are not realistic at all.
OP is missing their win/loss ratio, which makes the calculation impossible. 95% is W, but win/loss is average gains/average losses. The answer explodes because the win/loss ratio is so small the fraction is blowing up (win/loss is in the denominator).
The result, the kelly percentage, should be less than 1, indicating a fractional percent of the account to bet. A kelly percentage with a denominator < 1 can be interpreted as "betting the farm on each play" because your odds are so astronomically low (your losses would greatly exceed your wins) that if you are going to play it's a literal dice roll - in other words DO NOT PLAY.
From his post:
95% winners (so 19 out of 20 win)
Average winner = $20
Average loser = $5,000
Horrible expectancy. Unless he made a typo, I think it can be dismissed right there, regardless if he uses Kelly or any other money management system? Martingale might be fun with those $5,000 losses, though...
The kelly criterion is negative for OP statement - meaning he has no edge and should not bet a dime on the trade.
Even with the largest edge, the kelly will never go above 1. If it ever was larger then one you would eventually go bankrupt. Unless it was a riskless bet and your success win rate was exactly 100%
Assuming discrete outcomes. The kelly would be:
b = 1/((5000/20)) = 1/250
p = .95
kelly = (1/250*.95 -.05)/(1/250)
Kelly = a negative #
Also kelly has a limit where the limit = 1.
However in most real trading examples, the payoffs are continuous rather than discrete. This makes calculating the Kelly criterion much more complex.
Even with the largest edge, the kelly will never go above 1. If it ever was larger then one you would eventually go bankrupt. Unless it was a riskless bet and your success win rate was exactly 100%
Assuming discrete outcomes. The kelly would be:
b = 1/((5000/20)) = 1/250
p = .95
kelly = (1/250*.95 -.05)/(1/250)
Kelly = a negative #
Also kelly has a limit where the limit = 1.
However in most real trading examples, the payoffs are continuous rather than discrete. This makes calculating the Kelly criterion much more complex.
system is not profitable
You need to make N trades (where N is a large number) to compute your expectancy, average R:R, probability of win/loss, then you can plug into the Kelly Criterion and compute your Kelly which gives you the bet size to maximize your gain, assuming you have unlimited capitals. For most of us, since we don't have unlimited capital, the risk of ruin is very real, trading fractional Kelly improves survival, which is what folks call risk management.
You need positive expectancy otherwise your Kelly will be negative => you will lose everything if you keep betting, i.e., Kelly tells you not to bet.
