|
PureBytes Links
Trading Reference Links
|
Hello Jim,
Thank you for your very interesting e-mail.
> [...]
> In general, I've found that the backtest results can vary greatly
> depending on how these formulas are used. In fact of all the systems
> that have been posted in the list over the last 6 mos. I've yet to
> backtest ANY that give results close to that of BUY&HOLD when tested
> on the stocks comprising the DJIA/NASDAQ100/SP500 US markets over the
> last 10-20 years w/o even including the effect of stops and drawdowns.
Well, US markets are in fact very specific. We had very nice long-term trends
there in a bull market that made Buy&Hold strategy working well.
It is quite the oposite for other markets - in Poland for example B&H strategy
does not work at all (maybe except the years 1993-mid 94).
Therefore it is important to backtest your system on different markets
to see how it performs not in bullish market.
> I'd like to hear from the backtesting and system development
> experts...
>
> Should a system be developed and tested over an entire market or
> tailored to fit an individual stock?
It depends, but generally speaking overfitting the system to one stock
is dangerous because it may never work so well in the future.
> Should a system be developed to fit both bull and bear markets or
> just the recent trading time period?
I would advise to test every system on both because you never
know what the future will come.
> Should a system be able to beat the buy & hold results as AMI
> calculates them or is this just a relative best-can-do benchmark?
Of course it should, otherwise there is no reason to use the system at all.
I have written systems that outperform B&H by quite nice margin.
> Should a system be developed and tested with or without a STOP?
Generally speaking you should use stops (at least stop loss). On the other hand
studies on using *profit target* stops show that using them generally
spoils the performance of the system.
> Should I just give up and buy a good indexed mutual fund? I hope
> not! :-)
Well, it depends on various factors - it is certainly easier and less time consuming
just to invest in funds, but:
- you can use technical analysis also to choose the best fund at a time
- it may happen that there is hard to find the fund that matches your preferences
and objectives
> TJ, you mentioned earlier summarizing the system backtest results in
> the library...
Yes, I am considering several enhancements in the AFL library.
I would like to define some standarized markets that we (as a community)
will use for backtesting so we can very quickly compare the performance
of various systems in various circumstances. This will be a great time-saver.
I hope that AFL Library will grow to become a very valuable resource
for trading system developers.
> I'd like to suggest that the user community establish a standard that
> these systems are backtested to for the purpose of including these
> results in the AFL library. Here's some thoughts...
>
> Standard US and/or non-US market.
> ->30DowStocks
> ->Other Non-US
> Standard time frame to include all market types.
> ->1980-to-Present
> ->Other
> Standard Entry/Exit
> -Next Day Open
> -Other
> Standard results such as...
> ->Return
> ->Return vs. Buy/Hold
> ->Max Drawdown
> ->Win Rate
> ->Profit Factor
> ->Other
Yes, your proposals are in line with my thoughts, I wish other people
on this group speak up about their proposals so we can
find something useful.
Maybe a poll would be useful?
> Finally, it was suggested earlier on the list that the Drawdown from
> the Backtester Report is the entryprice-to-trough rather than the
> peak-to-trough. Is this correct?
Yes, currently back tester reports entryprice-to-trough drawdown,
but peak-to-trough will be added.
Best regards,
Tomasz Janeczko
===============
AmiBroker - the comprehensive share manager.
http://www.amibroker.com
|