Kelly sizing...

[rvince99]

IntradayBill,

It;s a "Leverage Factor," not a "fraction." The two are only the same value under certain circumstances (which rarely occur in trading). Using the leverage factor (which is what the answer that satisfies the kelly criterion returns) as a fraction to risk, will have you over-levered, ie beyond the peak of the optimal fraction, the optimal f. It can be converted to it's optimal f equivalent, but most don;t know how to do this.

Kelly was mistaken (and Shannon i n his review of such ) were mistaken when he thought the answer returned a fraction. -Ralph Vince

 

[intradaybill]

Quote from rvince99:

IntradayBill,

It;s a "Leverage Factor," not a "fraction." The two are only the same value under certain circumstances (which rarely occur in trading). Using the leverage factor (which is what the answer that satisfies the kelly criterion returns) as a fraction to risk, will have you over-levered, ie beyond the peak of the optimal fraction, the optimal f. It can be converted to it's optimal f equivalent, but most don;t know how to do this.

Kelly was mistaken (and Shannon i n his review of such ) were mistaken when he thought the answer returned a fraction. -Ralph Vince

Ralph, thanks for the answer but I do not understand what you mean by leverage factor in the context of Kelly. Can you give an example?

Let us say someone has 100K in the account and Kelly is 0.80. What does it mean according to you, if anything at all?

 

[rvince99]

Bill,

I can but not here. If someone wants more on it, they need to obtain the paper from The Journal of The International Federation of Technical Analysys #11 available through admin at ifta dot org or attend the 2-day course on these matters -- informatoin at my website ralphvince.com

 

[u21c3f6]

Quote from rvince99:

...
Kelly was mistaken (and Shannon i n his review of such ) were mistaken when he thought the answer returned a fraction. -Ralph Vince

:eek:

There has to be a miscommunication somewhere.

Assuming perfect inputs, Kelly is the % of your capital to invest for optimal capital growth. Yes, I know that an individual's inputs are probably not going to be perfect which is why many suggest a fractional Kelly and in my case I use and recommend half-Kelly.

If you have an investment that risks $1 to make $1 and you "win" 55% of the time, you have a 10% edge. Kelly is edge divided by odds or 10%/1 to 1 or 10%. Using any other % than 10% in this example returns inferior results. Half-Kelly (5%) will produce about 75% of the growth rate while 1.5 Kelly (15%) will more than likely bankrupt you.

If you risk $2 to make $1 and you "win" 70% of the time, you have a 5% edge. Kelly would be 5%/1 to 2 or risk 10% of capital for optimum growth.

If you risk $1 to make $2 and you win 40% of the time, you have a 20% edge. Kelly would be 20%/2 to 1 or risk 10% of capital for optimum growth.

There are of course formulas to use if you are making more than one investment at a time or you can just have separate capital pools that you draw from for different investments. Not optimal, but close enough and easier to handle IMO.

Joe.

 

[rvince99]

Joe,

You say

"Assuming perfect inputs, Kelly is the % of your capital to invest for optimal capital growth."

But that is NOT correct. In certain circumstances it will EQUAL the % of your capital to invest for optimal capital growth, but in MOST instances, in trading, it will not. That's the point of my argument.

But there's no miscommunication. It is an incorrect assumption that Kelly will yield the % of your capital to invest for optimal capital growth. Only in certain, simplified instances will it EQUAL that, and those conditions rarely exist in trading. And my point is that people using the Kelly Criterion are ending up with numbers that they thin are this, but have them greatly over-exposed to accomplish this.

-Ralph Vince

 

[u21c3f6]

Quote from rvince99:

Joe,

You say

"Assuming perfect inputs, Kelly is the % of your capital to invest for optimal capital growth."

But that is NOT correct. In certain circumstances it will EQUAL the % of your capital to invest for optimal capital growth, but in MOST instances, in trading, it will not. That's the point of my argument.

But there's no miscommunication. It is an incorrect assumption that Kelly will yield the % of your capital to invest for optimal capital growth. Only in certain, simplified instances will it EQUAL that, and those conditions rarely exist in trading. And my point is that people using the Kelly Criterion are ending up with numbers that they thin are this, but have them greatly over-exposed to accomplish this.

-Ralph Vince

Ralph, With all due respect, there has to be a communication problem here because if you know your edge and the odds, Kelly is mathemtically proven to be optimal. All other %'s will underperform.

Can you give me an example of edge and odds where Kelly is not optimal?

Joe.

 

[rvince99]

Joe,

The Kelly Criterion is NOT proven to be the optimal fraction to allocate to a risk opportunity. As I have proven in the paper in the IFTA journal (and will cover in-depth in Tampa) it is proven to be EQUAL to the Optimal fraction in only certain circumstances which are infrequently present in trading. These "certain circumstances" are the very circumstances cited in support of (false) proof that the Kelly Criterion Solution is the growth optimal fraction. In most cases in trading it yields a number GREATER than the growth optimal fraction, and hence one using it assumes greater risk for less return.

I'm not going to circumvent the paper in The IFTA Journal. You can take my word for it, or you can obtain the Journal / come to Tampa, or you can continue with the widely-held delusion that the answer to the Kelly Criterion will yield the growth optimal fraction to allocate. But I cannot circumvent the Journal article here. -Ralph Vince

 

[u21c3f6]

Quote from rvince99:

Joe,

The Kelly Criterion is NOT proven to be the optimal fraction to allocate to a risk opportunity. ... -Ralph Vince

Let's agree to disagree.

Given the edge and the odds, anyone can easily prove to themselves using just MS Excel that Kelly is the optimal % of capital to risk for optimal capital growth.

Joe.

 

[kut2k2]

Quote from u21c3f6:

Let's agree to disagree.

Given the edge and the odds, anyone can easily prove to themselves using just MS Excel that Kelly is the optimal % of capital to risk for optimal capital growth.

Joe.

Joe, you are correct. Ralph Vince has been blowing smoke up people's arses for years now because he is a vendor with something to sell, and trying to undercut Kelly is his particular schtick. As you pointed out, the mathematics behind Kelly are exact. The danger comes from oversimplifying assumptions like reducing a multi-outcome situation to a two-outcome situation (avg win, avg loss). When you follow the math like gummy and I did, you get a correct and accurate Kelly formula that works.

When it comes to "optimal f", as the old saying goes: there is no "there" there.

 

[rvince99]

Guys,

Say what you want -- I'm trying to help you here. I'm not trying to "sell" anything to you -- not books, not an IFTA Journal -- not even the 2 day course. (Of course, you can go Cal or someplace nice and pricey like that, and learn the WRONG things, by people who DON'T trade).

You sound like the guys sitting at the bar, drinking water, gorging on the free popcorn who are mad when the girls ignore them.

I'm trying to help you guys. I don;t NEED to sell anything (My partner and I are at about 500 consecutive winning trades, and have not had a losing trade in going on 3 years here, in markets in the US and around the world. If you don't believe me, wait a few months and buy into the forthcoming ETF on this).

-Ralph Vince