Portfolio of the Week - 2009

[peanuts]

oil and NG looking like a good trade here. I might be a little early, but I like buying quality after a selloff

XTXI long please and thank you

[Karel]

Go EXM!

I didn't have much to shout about the last few weeks, but perhaps it is time to become a bit of a loudmouth again

Regards,

Karel

[Rob]

. . . perhaps it is time to become a bit of a loudmouth again

Indeed.

Congrats, Colossal One.

[peanuts]

looking good, Karel.

Ody, please sell XTXI and I'd like to short SWHC please

Thanks for running the contest. It's much fun and you do a great job

[Rob]

No Complaints

Please scratch [JASO] and put me into GS long. Thanks.

Should have stuck with JASO; it would have brought me 11.44%. After four bad weeks, though, I guess I can't complain about 6.9%.

[Karel]

Long EXM

No complaints here either! And something rather strange: EXM now stands at 7.76. When the competition opened this year, on January 5th, EXM opend at 8.23. So EXM has lost YTD. But my POTW balance shows a 90% gain. Indeed, no complaints!

Ody, I am still long EXM. What else?

Regards,

Karel

[Odysseus]

Well put Rob, a colossal win by Karel this week ! EXM was up past 30% at one point an end of +27% is nothing to be sad about That is pretty interesting that you played EXM up and down and made quite a bit of cash, you're 2nd place for the year at the moment, only behind IIC. Congrats !

Still a good win Rob with 6.9% ! You would have came in 2nd both ways A well earned 2nd placing of course, in fact your 7th. You've tied Peanuts for the most silvers, congrats !

JohnHenry came in 3rd this week with LOPE ! I had to look that company up just for the odd name factor. I have to say, JohnHenry just showed up on the scene and has only made 4 picks, but has placed 75% of the time. Keep it up sir !




Just wanted to bring something to the attention of everyone because I don't know what I'm doing wrong When we first started I wrote up the equations for calculating the Math Expectancy and Kelly Value for each person, but the numbers always seemed odd to me. I double checked from here and though my equation looks correct, the numbers don't look right to me.

Karel brought it to my attention a few weeks ago but I just haven't had a chance to delve much into it until now. If anyone has any suggestion on my error I'd be greatly appreciative. Just look at your sheet from the POTW link below, it should tell you the numbers and equations that the cell is representing. Here is the Math Exp. and Kelly Value equation I am using in each of those cells in case you can't see them.

Peter Hansen's profile:
Math Exp: ( J28 * I28 ) - ( 1 - J28 ) = Probability * WL - ( 1 - Probability)
Kelly Val: J28 - ( ( 1 - J28 ) / I28 ) = Probability - ( 1 - Probability ) / WL

J28 = % Winners = ( B28 / ( B28 + D28 ) )
I28 = W/L Rate = B28 / D28
B28 = Winners
D28 = Losers

Thanks and good luck all !

[Karel]

[...]
Just wanted to bring something to the attention of everyone because I don't know what I'm doing wrong When we first started I wrote up the equations for calculating the Math Expectancy and Kelly Value for each person, but the numbers always seemed odd to me. I double checked from here and though my equation looks correct, the numbers don't look right to me.

Karel brought it to my attention a few weeks ago but I just haven't had a chance to delve much into it until now. If anyone has any suggestion on my error I'd be greatly appreciative. Just look at your sheet from the POTW link below, it should tell you the numbers and equations that the cell is representing. Here is the Math Exp. and Kelly Value equation I am using in each of those cells in case you can't see them.

Peter Hansen's profile:
Math Exp: ( J28 * I28 ) - ( 1 - J28 ) = Probability * WL - ( 1 - Probability)
Kelly Val: J28 - ( ( 1 - J28 ) / I28 ) = Probability - ( 1 - Probability ) / WL

J28 = % Winners = ( B28 / ( B28 + D28 ) )
I28 = W/L Rate = B28 / D28
B28 = Winners
D28 = Losers

Thanks and good luck all !

I think the error is in the I28 formula, Win/Loss ratio. That should calculate the average % gain of the winners/avg % loss of the losers. You could do it with formulas, if there is something like an AVERAGE.IF (not in Excel, where I use SUM.IF/COUNT.IF) or just add a winner (loser) to the correct AVERAGE(X2+X5..) formula.

Regards,

Karel

[Peter Hansen]

Aixg long

Ody you are doing an excellent job ....wow you really do get into the math analysis......GREAT WORK. AIXG LONG FOR THIS WEEK !

[Odysseus]

I think the error is in the I28 formula, Win/Loss ratio. That should calculate the average % gain of the winners/avg % loss of the losers. You could do it with formulas, if there is something like an AVERAGE.IF (not in Excel, where I use SUM.IF/COUNT.IF) or just add a winner (loser) to the correct AVERAGE(X2+X5..) formula.

Regards,

Karel

Karel, you are a genius ! I think you are very correct. I only applied it to Peter's portfolio, can you tell me if those numbers look better? To me they do based on what I know of Math Expectancy and the Kelly Value (which from what I understood the latter should always be positive as it is the amount of your portfolio you can risk)
Kelly %

The Kelly % statistic was developed by John Kelly at AT&T's Bell Laboratories in 1956. John Kelly's original work dealt with long distance telephone transmission signal noise. You can read his original AT&T paper on signal noise by downloading the PDF at the bottom of this section.

Kelly's work is recognised today as being able to determine the optimal betting size which will maximise the growth of your portfolio over time.

The formula for the Kelly % is:

Kelly % = W - (1 - W) / R

Kelly % = Percentage of capital to risk per trade for maximum gains.
W = Historical Winning percentage of the System (trades).
R = Historical Average Win/Loss Ratio (AVG Dollar Win/Loss).

The Kelly % statistic can be applied to a system as follows:

Peter has a trading system with a winning percentage of 50% with average profits twice the size of his average losses.

From this brief description we know that Peter has a profitable System.
Putting these figures into the formula for Kelly % gives us W = 0.5 and R = 2. Substituting this into the Kelly % formula gives the Kelly % as 0.25. The Kelly % tells us the amount of equity Peter should risk in order to obtain optimal profits is around 25%.

It should be noted that this amount of risk can be spread over the number of trades on average the System will have open at any one time.

It is common to use 80% of the Kelly % as the optimal risk that is tolerable to a trader. In the example above, Peter has a system that on average has 10 positions open at any one time. He can spread his (25% x .8 = 20%) 20% risk over his ten trades. For Peter to obtain optimal results from his system, he should risk 2% for each trade.

btw, I'm going for WMT long this week

[Odysseus]

whoa scratch that, I meant WMT SHORT, sorry about that

[JohnHenry]

Mnro

Congrats to everyone in the green....MNRO lone please...thanks

[Karel]

Karel, you are a genius ! I think you are very correct. I only applied it to Peter's portfolio, can you tell me if those numbers look better? To me they do based on what I know of Math Expectancy and the Kelly Value (which from what I understood the latter should always be positive as it is the amount of your portfolio you can risk)

Hi Ody, you have the W/L ratio negative but that should be a positive value, as it compares sizes only. As a result you get a high Kelly Value, which is unlikely for a string of trades with a slight overall loss. I get 0.0389 for the KV when I drop the sign. The rest looks OK.

Regards,

Karel

[Rob]

Rob's Pick

Rob will go with PCP long this week. Thanks Spo-Dee-O-Dee!

... no wise cracks from the stoner section, please.

... and don't forget to .... Stay ....... De- ........ MENTED!!!! (obscure reference to an old radio show you may or may not know about ... Ody knows.)

[Odysseus]

Thanks again Karel, now it is

Math: 0.06
Kelly: 0.04

I threw an ABS() to negate all signs. Now just have to set up the formula. Unfortunately I tried accomplishing with google app's functions, but none seem to do the trick very well. Though I'm going to see if I can manipulate SUMIF to function the way I want, otherwise I may have to do it manually, or seperate all losses and gains into 2 different columns