|
PureBytes Links
Trading Reference Links
|
Dennis,
Thank you for your few lines.
>This has been discussed many times on the list. It's probably all in the
archives.
> It is not necessary to do percentage calculations or to buy expensive
> add-ins. All you need is vary your size (number of contracts/shares
> traded) during the backtest so each trade is risking an equal dollar
> amount. Then all the standard TS stats become meaningful.
> Of course, people selling books and/or software advocating percentages
> will no doubt disagree. :-)
I do not sell books or software, nevertheless I disagree.
Because of confidence.
Lot of TS's customers( new ones probably) put their trust in the results
from TS Perf summary - naively.
And naively make back tests with it - The results are completely wrong and
they are fooled . If they trade a system regarding to performance summary
results, they will collapse soon because of unexpected DD or average trade
or ......
It is fist a question of confidence. Wrong results are just non acceptable.
Because of wrong results anyway.
I do not like the idea to artificially vary the number of contract traded
just for compensate the wrong results. Multiple of one future contract will
not represent the exact portion of each trade result if calculated in % with
decimals for one contract , and partial multiple of one contract will result
on wrong ratios for the fields that depends on margin for example. The
resulting approximation/error is growing with the number of trades in the
back test. This way of compensation conduct to approximative results, not
the ones I am looking for.
Because of Comparisons.
The way you advocate you cannot made good comparisons between results of
several systems on a same support or the same system on several supports or
different time frames etc.
Even if you do not do it within a portfolio, you must do it one at a time in
order to test the robustness of your system or anything else you want to
test.
Because of position sizing.
When I want to apply and backtest systems with position sizing, and to
compare the results it is a confusion and unnecessary complexity.
Because it is my need to do it in a different way from your's.
I may be the only one on the list to think this, It doesn't matter.
In this case the tsresearchgroup' RealTime PortfolioAnalyzer add on to TS
was an unnecessary and useless work.
>The whole "TS percentage bug" when trading one market with one system was
perpetuated >by a series of articles, a book and an Excel add-on sold by
Thomas Stridesman. Mr. >Stridesman simply didn't understand how to use TS
correctly to achieve the same result. His >book and his Excel program can be
replaced with a half dozen lines of EL code.
In the same biased way of thinking TS could also have been proposed without
any wrong performance summary to oblige each user to write his own.
> Mr. Stridsman simply didn't understand how to use TS correctly to achieve
the same result
Correct use of TS software could be:
- Apply a strategy to a chart.
- Push the strategy performance report.
- See and analyse the GOOD results.
- Make a decision
This is the Incorrect use :
- Apply a strategy to a chart.
- Push the strategy performance report.
- See and analyse the WRONG results.
- Make a decision
May be YOU simply do not understand or you are not able to accept that what
is suited to you, do not suit every others person's needs.
It is not necessary to debate if I am wrong and you are right and to impose
just one point of view as the only one acceptable. We are all different,
with different needs.
When Help is needed, to be critical is not.
You may help us with great knowledge and intelligence, with respect to human
differences rather than conflict with it.
The prerequisite is open mind to and comprehension of other's.
Sorry for the inconvenience of this post, I will be more positive from now.
Regards
Philippe
> -----Message d'origine-----
> De : DH [mailto:catapult@xxxxxxxxxxxxxxxxxx]
> Envoyé : mardi 24 août 2004 22:32
> À : Omega List
> Objet : Re: Add on Performance summary with Percent values and lot more.
>
> This has been discussed many times on the list. It's probably all in the
> archives.
>
> It is not necessary to do percentage calculations or to buy expensive
> add-ins. All you need is vary your size (number of contracts/shares
> traded) during the backtest so each trade is risking an equal dollar
> amount. Then all the standard TS stats become meaningful.
>
> Of course, people selling books and/or software advocating percentages
> will no doubt disagree. :-)
>
> --
> Dennis
|