Hi kris45mar,

Wednesday, March 15, 2006, 12:39:07 AM, you wrote:

k> I have 25% DD with 4 wins out of the last 30 discretionary trades.

I was a discretionary trader for many, many years, but no longer. I
don't sit in front of the screen all day now, and my physical
condition has improved remarkably (less stress - more exercise). It's
nice having time for other things.

That said ... one can be a good discretionary trader. I survived and
made a living doing it for longer than I care to admit. But with the
record you have over the past 30 trades, you have to stop
immediately, or scale back position size harshly, or you may be
headed for trouble. You need a complete re-evaluation of your
methods, IMNSHO. Otherwise, you risk the emotional precipice -- at
least, what I assume would be the emotional precipice, but maybe you
have another pile in the bank somewhere, and it's 25% of a
meaningless number. I don't know. But I do know that 4 and 26 would
make me want to fade my own trades. ^_^

k> Looking through my last two years' trades tells me that what worked in
2004 is not working in 2005/6. This brings me to the point in my 2005 trading
plan where I defined conditions to stop
trading. I now need a change of direction: the plan is to continue to explore
AB, AFL and the superb posts on this board towards developing a mechanical
system. It can't be that hard for me to
develope one that does better than my 2005 trading year. Whether I can then
actually trade it is a whole different ball game.

So often, what worked in one specific year, or a couple of years,
suddenly causes a lot of grief. I don't know how long you have been
at this, or if you backtested (doesn't jive with the word
'discretionary' that you wrote), or for how long. What I do know is
that one year of results (good or bad) doesn't serve as very much of
an indication as to what the future might hold. Again and again, most
everyone on this list and elsewhere who is in this business for a
living will tell you that you need to find something that has worked
for a substantial period of time, in up, down, and flat markets, then
do a walk forward on out of sample data, finding reasonably comparable
expectations in the OOS period.

k> All I am asking is this:

k> Markets change over time (that is why there is no Holy Grail) and so
should our systems, or the ones we choose to trade with, not respond to this?
Or we may choose to stand aside for a while.
Or just trade different markets with concurrently different systems to create a
smoother equity curve overall?

Personally, I want a system with an inconsequential drawdown even
when multiplied. I also want a reasonably smooth equity curve, and I
don't want to see a small number of huge winners against a much
larger number of breakevens or losses. I also want to see a very
large sample period, and no inconsequential OOS period either,
containing a *lot* of trades.

But again, you said 'discretionary', so I am getting mixed signals
here.

k> Could you comment on whether you trade with one system only or more than
one? And if more than one, what would be a trigger to change if the Equity
curve is not the signal to do so? Drawdowns?
Sleepless nights? Declining expectancy? This has to part of our business plan
after all. In 2004 I achieved my trading goals, 2005 was not a successful one.
Message: time to stop doing what I am
doing: it is not working. Do something else. The goal then is to replace what I
am doing with something that does work.

k> I realise the answers to these questions are personal, but it is
invaluable to get some insight to the philosophies of others, in an attempt to
know where to start.

I have more than one live system right now. But less than five. ^_^

I think what anyone will tell you about your question above is
something like this: Take a truly thorough look at your system.
Break apart overall performance into smaller bites. Know what amount
of 'badness' is possible, imagine you are in the midst of it right
now, and see how you feel about that. Then double the 'badness'. Now
how do you feel? Can you keep trading it? Do you feel confident
about the logicalness of the underlying premise? Is the market
telling you that you are crazy?

Number one, you simply have to find a system with real promise. A
marginal system is not good enough, because by the time you realize
it is actually broken, so are you -- emotionally if not financially.
You need to find a system so good that, when it's bad, it's not
*terribly* bad, and one that doesn't go soft for very long periods of
time. Then, if it gets pretty bad for any substantially lengthy
period, two things will be likely: 1) You will come to realize what
is happening, and 2) Because the system had good enough expectation
to begin with, you won't blow your capital or your emotional well
being before that realization sets in.

At that time, you have to scale back or sit aside. Meanwhile, you
keep looking for other edges.

The most important metric for me is drawdown. You have to be able to
stand a multiple of the maximum over the testing periods. I mean
*comfortably* stand it. Otherwise, you might as well stay
discretionary, because you don't really have a system you can
emotionally trade, in all likelihood.

It also really helps to be a winner out of the gate. With
discretionary trading, one will often hear that this is a curse,
dooming the trader to becoming overconfident. That indeed may be
true. But for a systematic trader, it's a blessing, because you pile
up enough capital to allow you to more comfortably tolerate that
multiple of the max drawdown that is probably lurking out there. But
when you get into a patch that is beginning to look unlike any other
patch in your sample or OOS periods, you have to be on guard. A
simple dip below the equity curve is not cause for alarm, and as I
mentioned may be cause to become *slightly* more aggressive. By
definition a dip below an equity MA *must* happen. What would be
alarming would be a relatively long period of substantial
underperformance versus both backtesting expectations and previous
live performance. I'm not sure I can put precise numbers on it, but
I'm sure I would know it if it happened.

Yuki