|
PureBytes Links
Trading Reference Links
|
Well first let me say that I don't trade so anybody who wants to discount my
opinion is perfectly welcome to ignore it for that reason.
I think systems should be tested in a way such that no one highly correlated
group makes up a large part of that system's trading. So maybe a concept of
maximum "at-risk" per correlated group could be introduced so that when this
level is met no more trades from that group can be accepted until another
one is closed out. This can be tricky - for example if your system goes
short SF but is short the DX (dollar index) you are both long and short the
dollar. So what is your "Currencies" at-risk then? Probably you have to
monitor both gross and net at-risk and set limits for each.
I also think it's very worthwhile looking at European and Asian markets. A
great "well-chosen" example is the behaviour of Brent Crude v Nymex Crude
over the past year. If you traded Aberration on both, they both entered the
last big uptrend on 11-Mar-1999. CL exited on 11-Oct-1999 and didn't
re-enter until 17-Nov-1999. Brent Crude remained long. Both are still long.
But you could look at Japanese markets (TGE,TCE,TIFFE,OSE,YCE,KANEX,KCE)
other Asians (SIMEX, SICOM, COMMEX Malaysia, KOFEX, KSE, KLFE, SFE) or
Europeans (LCE,LME,EUREX,LIFFE,MATIF,IPE etc) for your extra
diversification.
About trading illiquid markets: I have to admit that I don't know how much
of a problem this is for a long term trend follower. I expect that as long
as the average net gain per trade is large enough you can stand a lot of
slippage (a al Mark Johnson's post). For my own trading I'd much prefer to
enter and exit on the open and not use stops - I'm hoping the benefits of
simplicity for me and my broker, the positive impact any order bucketing
will have and the certainty of being filled outweigh any negatives...but I'd
like to hear from you if you disagree.
Regards,
Robert
-----Original Message-----
From: Owen Davies [mailto:owen@xxxxxxxxxxxxx]
Sent: 27 March 2000 17:03
To: Robert Hodge; omega-list@xxxxxxxxxx
Subject: Re: Long Term Systems
Robert Hodge brought up a significant issue when he pointed out:
> While I am a long term trading kind of guy one thing does upset me when I
> see results for simple long term systems like channel breakout and
> aberration: people tend to choose the "usual suspects" (currencies,
energy,
> interest rates) to test their ideas on. (etc.)
Robert, ignoring the S&P, which contracts would you like to see the
system tested against? Grains? Meats? Once you get out of the
standard test subjects, it's easy to run out of contracts with open
interests
big enough to be comfortable trading.
There's also the argument that it's kind of futile testing a trend following
system on markets that you already know don't tend to trend. Do you
see any validity in using trend following for some markets and different
methods on others, based on the market's historical character?
Owen Davies
|