--- In [email protected], "brian_z111" <brian_z...@...> wrote:
>
> Did you consciously choose this 'portfolio model' or did you drift into it?
> Why do you do it that way as opposed to another?
> What advantages do you see in your method?
>
>
{hint: this is a replay of message #138150, from which I deleted my comments
about others software.
Since I'm using primarily Amibroker, I feel it was not fair for me to express
judgements in other testing engines. I just eliminate the most evident grammar
errors.}
Brian,
as you are always a very kind person, I don't want to give the impression I
don't like to answer your questions but I feel Hicks already exposed my exact
points in massage #137869: "There are two main ways to combine systems. One
option is each has its own equity pool and the other is they share an equity
pool.".
Ok, let's try another way: leave the multisystem field for a moment and think
about trading ONE system on ONE market. After some research, we can think to
need a starting capital of X.
Now let's think to expand to a portfolio of N different market with equal
volatility and contract amount (I'm simplifying!!!): is the capital required
N*X? No, because diversification does reduce risk
not always (ask LTCM) but
most of the time
.. (just as example, one can re-read William Gallacher at pag
188 of his book "winners take it all").
So - JUST AS A THEORICAL EXAMPLE let's say we need a starting capital of Z
which roughly equals = (the square root of N multiply by X).
Now, we are going to introduce multisystem : on a portfolio of N markets, we
trade M systems. What is the starting capital needed?
Well, it depends on the correlation about systems (by the way, speaking about
correlation without defining "a time window" makes no sense
.. but I already
told
I'm strongly simplifying!).
But theoretically, only in ONE case we need a starting capital of Z*M.
Yes, it is obvious: it is only when the M systems' results are 100% correlated
(that's also means you have no reason to trade M systems
. but this is just an
extreme case).
In any other case, I would need a trading capital that is LESS than Z*M, which
LESS means a more efficient use of it.
So, in the end, people saying "I will allocate X% of funds at system A and Y%
of funds at system B
. and so on
.. they are probably in a total
unconscious way supposing that all their systems are perfectly correlated.
If this is crystal-clear why are still people thinking in terms of X%
allocating at strategy A and Y% at strategy B?
Because we have sometimes different objective. For example, in very big hedge
funds, allocating X% of capital to every trading scheme allows to more clearly
tracks the risk and reward of each approach and also valuate the jobs of the
different group of traders in charge for it (and decide accordingly the year
end bonuses too!!).
Ok, that's all for now because this thread is very interesting but my actively
following it is starting to impede my normal daily routine!!!
