- Joined
- 6/21/16
- Messages
- 2
- Points
- 11
Hi all,
I'm currently developing my undergrad thesis on the Kelly. (So I will put a 'thank you' note on the thesis for those who help me. Haha)
My doubt is on Kelly in normal distribution case.
On Edward Thorp's paper, The Kelly Criterion in Blackjack, Sports Betting and The Stock Market, he derive the following equations:
f* = The Kelly fraction g(f) = the instantaneous expected return or drift of the portfolio
g(f*) = the case with f=f*
m = Asset expected return r= risk free rate s = standard deviation
The point is on the g(f) equation. Why the Stan. Deviation would metter for this? Why is the g(f) not only g(f) = (f*m)+((1+f)*r), i.e. the weighted mean of the returns? It implicate that with a larger deviation I will have a lower return?
It is even more weired if you set f=1 (invest all in the risk asset), cause the equation gets g(1) = m - (s^2)/2. So the portfolio made up only with 100% in the asset will perform lower than the asset itself?
Thank you, guys
I'm currently developing my undergrad thesis on the Kelly. (So I will put a 'thank you' note on the thesis for those who help me. Haha)
My doubt is on Kelly in normal distribution case.
On Edward Thorp's paper, The Kelly Criterion in Blackjack, Sports Betting and The Stock Market, he derive the following equations:
f* = The Kelly fraction g(f) = the instantaneous expected return or drift of the portfolio
g(f*) = the case with f=f*
m = Asset expected return r= risk free rate s = standard deviation
The point is on the g(f) equation. Why the Stan. Deviation would metter for this? Why is the g(f) not only g(f) = (f*m)+((1+f)*r), i.e. the weighted mean of the returns? It implicate that with a larger deviation I will have a lower return?
It is even more weired if you set f=1 (invest all in the risk asset), cause the equation gets g(1) = m - (s^2)/2. So the portfolio made up only with 100% in the asset will perform lower than the asset itself?
Thank you, guys
