|
|
 |
| The Breakout
Bulletin
The following article was originally published in the August 2002 issue of
The Breakout Bulletin.
Answers to Your
Questions
Q: I ran
an optimization test, but am not sure I did it correctly. At first I tried
to optimize all variables in increments of 0.1, but that would have produced 4.8
million tests! So I instead started to optimize one variable at a time,
working my way down the list of variables, until the variables held steady. I
set the equity input equal to $100,000. Please comment on my optimization
technique.
A: I
recommend setting the equity parameter equal to zero when optimizing. That way,
it will default to one contract per trade and you don't have to worry about the
account size settings. I usually either do one variable at a time or 2 at a
time. For example, you might try optimizing the entry and exit parameters for
long trades together, then the 2 parameters for the short trades, followed by
the trend parameter.
I like to look at several performance statistics
when optimizing. I generally optimize on the profit factor, but always check the
equity curve to make sure it is as straight as possible. I also like to have at
least 100 trades in my sample. I like to see the profits for the most recent
period strong, a low number of max consecutive losers, and a high net profit. I
also like to see a small max loser and a good balance between long and short
trade profits. Finally, I like to make sure that the parameters I end up with
produce reasonable trades. By this I mean that the stops can't be
unrealistically tight or so loose that I'm not comfortable with that much risk.
Similarly, the entry fractions shouldn't be so small that I enter on every false
breakout or so large that I get too few trades.
Of course, there's no one
correct way to choose parameter settings. To a large extent, it depends on what
kind of trading you like to do. With a little searching, you will probably find
completely different parameter sets that give equally good, although different,
results.
Q:
I have looked at the new Minimax version but cannot get the parameter
settings quoted on page 4 of the new manual to give me anything like the
performance summary results shown on pages 13 and 16. Could you please confirm
the setting used for these results were as per page 4 or, if not, could you let
me know what they are?
A:
The parameters on p. 4 of the new manual (v2) are the ones used
to generate the results listed in the manual. Your data may be different than
mine. I used TS 6 with the full session (23 1/2 hour) contract e-mini data
series to generate these results. In TS 6, these are symbols @ES and @NQ. You
should confirm that your data covers the full session and not only the day
session. You should also confirm that it covers the same time period
as in the user's guide. If you're not getting the same results but you
believe your data is good, I would recommend reoptimizing. I recommend doing
this in any case so that you get to choose the parameters that give the kind of
trading results you prefer. You may prefer to minimize your drawdown at the
expense of net profit, or maybe you'd rather tighten up the stops even though it
might reduce the percent winners. There's no one best set of parameter values.
Also, you'll probably be more confident in the system if you explore different
parameter sets and choose your own.
Q:
What should I do in the spreadsheet when a new contract starts
trading? I use CSI for my data, so my data backadjusts after the
rollover date. Does this mean I need to backadjust the data in the
spreadsheet manually?
A:
After a rollover, the easiest way to handle it in the spreadsheet is just to
copy over all your existing data with the newly adjusted data from CSI. That
way, all your data will be back-adjusted properly. For example, let's say the
rollover date for the Sep contract is Sep 18th. On that date, your CSI data
series will back-adjust. This may change the prices prior to Sep 18th slightly
based on the difference between the closes of the Sep and Dec contracts on
Sep 18; this is the essence of the back-adjusting process. To make sure all the
data in the spreadsheet is adjusted accordingly, you could copy your CSI price
data from the beginning of the date in your spreadsheet to Sep 18th, and paste
this data into your spreadsheet over the existing data. This way, all the data
in your spreadsheet will be up-to-date. The system only looks back about 14
days, however, so this will only make a difference for some of the calculations
for up to two weeks following the rollover date. I haven't tested this, but
based on how the system works I suspect the difference between back-adjusting
and not back-adjusting (i.e., using the nearest futures) will be fairly
small.
That's all for
now. Good luck with your trading.
Mike Bryant
Breakout Futures
|
|
|
