|
PureBytes Links
Trading Reference Links
|
I never saw anything posted here, but one such formula
is
fraction to wager = 2P - 1
where P is the probability that the value of the stock will
increase in the next time interval.
For example, if you looked at 1000 closing prices of
IBM, and saw that on average IBM increases 51%
of the time (51% *is* about average for most U.S.
equities) then the optimum amount to wager
would be 2% of your capital (2*0.51 -1)
Doesn't sound like a lot, but unless you have
other information, it is the optimum amount to wager.
Turns out you only want to "bet" on stocks with
P > 0.5 (makes sense, unless you have some other
information).
The questions are, how many samples do you have to
look at to get a good estimate of this probability?
Is 1000 samples enough?
You can maximize your gain by betting 100% of your capital,
but that also maximizes risk. You minimize risk byt betting
nothing, but that minimizes gain. The 2P-1 formula
maximizes gain while simultanesouly minimizing risk.
So, the question is, if 2P-1 is the optimum fraction
to wager, is that the fraction of everything I own,
or the fraction of some other pool of wealth?
The "fraction to wager" proof is derived in
Reza, "An Introduction to Information Theory"
Dover Publications, p. 450, 1994. It is
also mentioned in Manfred Schroeder "Fractals,
Chaos, Power Laws".
A slighly more elaborate wagering strategy (not
for the mathematically faint-of-hear) is shown
on the web at
http://www.physics.ubc.ca/~blok/p510gambler/p510gambler.html
>
>Some time ago I saw a formula that spit out the amount of capital (as
>a percentage) that you should devote to any given trade based on the
>number of wins, number of losses, average dollars made per win, and
>average dollars lost per loss.
>
>The theory being that if you devoted LESS than this per trade, you
>were not making as much as you could, and if you devoted MORE than
>this per trade, your risk/reward ratio went up (increased risk of
>ruin).
>
>Anyone have this formula handy? I'm trying to comparea couple
>different systems to each other and while this formula may or may not
>be what I use to determine how much to put into a trade, it seems
>useful for comparison purposes.
>
>Thanks in advance,
>
>
>
>Mike
>
|