[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[EquisMetaStock Group] Re: Rules Management for Trading System



PureBytes Links

Trading Reference Links


This is a big question - and I guess it does not have an easy
answer.

For one exercise, we use optimisation, per ticker, to solve this
problem.  We test numerous different systems, your 1 ... 100, and
optimise each one, and then take the one, say #20, that performs the
best, where each one is optimised.  We do all sorts of tricks to
ensure we do not overfit, but, again, it is no simple answer.

But to get quick and general answers, we do not use 1 ... 100, but
just one of them, optimised, that our sorter thinks performs best.
Otherwise it becomes very complex and in practise you also start
overfitting.  Often, if #10 and #20 are good systems, and you use
both at the same time, you either seriously overfit or add no real
additional value, ie #10 and #20 fire at the same time or fire so
far apart you get confused!

One thing we have also used, for a more thorough, automated solution,
is to take a number of 'good' systems, say what we feel are the best
30 of the 1 ... 100 systems.  Then we do not optimise, but
standardise all of these 30 inputs.  Again, this is not trivial and
there is no simple answer, but we try to generate a value between -1
and +1 for each of the 30 systems.  So in stead of taking the say
RSI (between 0 and 100), we divide by 50 and subtract 1 to get it to
between -1 and +1, and do something similar with all the rest.  So it
is easy to compare and easy to optimise.

Also, we try to include some variation, say system 1 is a moving
average crossover, we use 10, 20 and 35 period and so get 3 inputs.
Sometimes we also mix and match these, say C <> 10 SMA, C <> 20 SMA,
10 SMA <> 20 SMA and so on.

Anyhow, we end up with this huge amount of technical and often also
fundamental inputs and throw it to a neural network, which then
optimises it.  Once again, the optimisation is not trivial.  We have
tried many things, such as getting the best equity line, getting the
best fit, getting the best turning point indicator and so on.  Once
we have an optimised net, we calibrate it - which is another big
topic - and then we use it or throw it away and start again!

Maybe I should just make a long story short, by mentioning that neural
nets are supposed to be good at combining lots of inputs from say
system 1 ... 100 into a single buy/sell answer.  We have found them
indeed to be good, as has others, but we had to address all sorts of
practical issues to get something that finally works robustly enough
for trading.

So, back to your specific question.  Either simplify and pick the
best of the lot, or take a huge lot and use something like a neural
net to combine the multiple inputs into a single output.

There is an Australian guy - Woodes Rogers or something like that -
that has written a book on his system, giving his 1 ... 5 criteria.
We use this as one of our systems and I know one trader who thinks
it is the best thing since sliced bread, but it really does not
perform well in our tests.  But you can google for his book to get
literature and some other opinions on this.

Regards
MG Ferreira
TsaTsa EOD Programmer and trading model builder
http://tsatsaeod.ferra4models.com
http://www.ferra4models.com


--- In equismetastock@xxxxxxxxxxxxxxx, chichungchoi <no_reply@xxxx> wrote:
> 
> 
> Does anyone know any topic related to rules management for trading 
> system? such as following:
> 
> 1) Closing above 50 days moving average
> 2) Closing above the highest high from the last 20 day
> 3) ...
> ...
> 100) ...
> 
> If there are so many rules for trading strategy, how can I do the mix 
> and match searching for the most reliable and profitable combination. 
> Once the best combination is found, and how to assign some threshold 
> before triggering any signal?
> 
> Does anyone have any idea/
> Thank you
> Eric





------------------------ Yahoo! Groups Sponsor --------------------~--> 
Has someone you know been affected by illness or disease?
Network for Good is THE place to support health awareness efforts!
http://us.click.yahoo.com/Rcy2bD/UOnJAA/cosFAA/BefplB/TM
--------------------------------------------------------------------~-> 

 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/equismetastock/

<*> To unsubscribe from this group, send an email to:
    equismetastock-unsubscribe@xxxxxxxxxxxxxxx

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/