|
PureBytes Links
Trading Reference Links
|
Thanks rvalue.
I'll follow your advice. Let me be as pointed as possible, and
summarise the whole Theta thread
The Theta model
- similar to Tillson's IE/2
- forecasting model, NOT trading model
- can be used as is for forecasting, NOT trading
- was tested as forecasting model, worked very well
- was tested as trading model, didn't work at all
- slope can be used for trading
- linregslope(log(CLOSE),periods)
- signals a trade when crossing zero line
- use lengthy period, 30 or above, I suggest 50 or more,
try 30 - 250 for end of day data
That is it.
> When people come back with negative back test results with what you
> suggested, you point them towards a different optimized direction..
I gave the Theta code upon request, somebody tested it, found it does
not work, and asked how we use this model specifically to trade. I
replied on why the Theta as is would not work and gave the way in
which we use it as a trading model. Please, this is not trying to
dodge issues nor going off in new directions.
> Try to be more specific with actual systems you have done and
> seen, offering your test results and pls don't lead folks astray..
Well, we use a huge mixture of models, as mentioned before. For
trading we make use a lot of trend lines. We developed some fancy
algo that will fit trendlines for you and this saves so much time.
It runs during the night, takes forever, and fits trendlines to
hundreds of tickers. In Metastock, the Raff regression channel gives
similar lines, although inferior, and I really prefer our own.
We also had good results with neural nets actually, but it is a huge
effort to build and calibrate such a model. The neural nets also
performed much better when we included some fundamental information
in them, not only technical.
The only test results I've submitted were on the T3. Here it is
again, plus, for free, some additional, subsequent results.
We altered the T3 somewhat, optimised it, and got good results. One
system that, interestingly, outperformed it is the Trix. This is
somewhat similar to the T3 and, if you are new to both but want to
toy with them, I suggest you start with the Trix. Hope this does not
lead anybody astray.
Regards
MG Ferreira
TsaTsa EOD Programmer and trading model builder
http://tsatsaeod.ferra4models.com
http://www.ferra4models.com
PS : I enjoyed your Elliott story. After aquiring AGet some years
ago, I've also bumped into some Elliott zealots. Around the
same time, we had a prestigious financial magazine, in a fancy
glossy, subject ALL shares on the local exchange to a fancy
fundamental, accounting model. They categorised and sorted and
came up with the five 'best investment for 19xx' stock picks.
I put the glossy in a drawer and revisited it a year later. All
five shares were down by a lot! At least it did not cost me
any money, and I don't remember specifics, but they were down
with 20 - 50%, all of them. I suspect they had some fundamental
problems at the beginning of the year that the accounting model
did not pick up, but that made them appear cheap.
We build models for a living, so I don't think this approach
is completely wrong. But a trading model should have a good
dose of technicals in it, as well as of probability/uncertainty,
not just cold-blooded accounting. Also, you would not sit on a
share for a year if it just falls I guess, so my evaluation is
not entirely correct.
--- In equismetastock@xxxxxxxxxxxxxxx, "rvalue1" <rvalue1@xxxx> wrote:
>
>
> MG - Thanks for your notes; but you strike me as a brainy individual
> who must not be trading their own money.. you must be trading "smart"
> money.. that is taking risks with other people's money. Your comments
> and answers seem somewhat vague and not pointed. When people come
> back with negative back test results with what you suggested, you
> point them towards a different optimized direction..
> Appreciate your insights; but its not for me.. its better for someone
> who has to work on their PhD, and not have to put their own real
> money on the line..
> I am sticking to the basics, trade long when its obvious the trend is
> up.. and use moving averages to tell the story ala Supefrag style...
> I will certainly keep checking out Elliott waves as I purchased the
> AO software; but am in no hurry to jump in.. I think almost every
> trade that TG recommended using AGet Elliott waves in the Optionetics
> seminar in Detroit last December, went sour. What a rip-off for a
> system touted as 70% winning trades.. Enjoyed the Optionetics
> classes and learned a lot; but you can't believe most of what people
> are preaching when they are also selling.. as they are after all,
> selling.
> MG - you strike me as very sharp.. so do keeop chipping in to this
> forum. Try to be more specific with actual systems you have done and
> seen, offering your test results and pls don't lead folks astray..
> --- In equismetastock@xxxxxxxxxxxxxxx, "MG Ferreira" <quant@xxxx>
> wrote:
> >
> > Just a few comments. I am certainly not trying to deceive or
> > mislead anybody. I have often encouraged you to not take my
> > word for a statement, but also gave references to those who
> > independantly came to the same conclusion. I am not testing
> > the Theta model at the moment, I am submitting it as one technique
> > to this group and am awaiting your and others' observations, and
> > acting on those already given, trying to fine tune it a bit and to
> > see if it makes sense to include in a suite of models. Let us
> > together and objectively see if the Theta model works. I suspect
> > it will work very well, but, don't take my word for it, let us
> see...
> >
> > Results on the latest 'linregslope' refinement anybody?
> >
> > In order to do this, your inputs are appreciated and valuable as
> well.
> > So, if I have in any way offended you, please do not resent me or my
> > inputs, and I apologise for my many hasty comments that may have
> > offended you.
> >
> > Now, in terms of optimisation. Let me not address its problems,
> > which are many, since you have already spend some time addressing
> > those. Let me rather take some time in its defence.
> >
> > Suppose you start trading today, using the simplest of methods. Say
> > you use just a simple moving average and use this crossover of the
> > market as a trade signal. Now, if you see that you are not doing
> > that well, or if you backtest it and see that it does not work, and
> > make even just one adjustment to the length parameter - even just
> > one - it can be taken, no matter what the result of this adjustment
> > is, as one iteration in an optimisation process. If you test even
> > just one alternative model, say using an exponential moving average,
> > then you are performing one step in a cross-method optimisation
> > process. If you ever play around with indicators, or alter
> > parameters, and then 'eye ball' the results, you are busy with
> > optimisation.
> >
> > Suppose you read a certain paper every morning, and look at your
> > moving average model every night, and based on these you make a
> > decision. Then you are working with a suite of models which you
> > aggregate internally, one model quantitative, one qualitative. You
> > dynamically assign weights to them, consult them and use them to
> > make a decision.
> >
> > What mathematics bring to the table is ways to externalise this
> > process and to objectively play around with model parameters until
> > the best one is discovered. There are many pitfalls, but it
> certainly
> > is neither a foreign nor a malicious process.
> >
> > Regards
> > MG Ferreira
> > TsaTsa EOD Programmer and trading model builder
> > http://tsatsaeod.ferra4models.com
> > http://www.ferra4models.com
> >
> >
> >
> >
> >
> > --- In equismetastock@xxxxxxxxxxxxxxx, "Jose" <josesilva22@xxxx>
> wrote:
> > >
> > >
> > > > Mmmmmm, the gloves are coming off it appears.
> > >
> > > MG, please don't take it personally, but unknowingly or
> otherwise, you
> > > are leading potential traders up a garden path with your
> optimization
> > > suggestions. You did it years ago, and you are still doing it
> now.
> > >
> > >
> > > > Interesting to note that the site you refer to was cancelled in
> the
> > > > previous century some time and basically lives on for historical
> > > > purposes. I can assure you the weaponry used today are much
> more
> > > > sophisticated, though it derives from what must have taken you
> an
> > > > awful lot of time to find on the internet, and for which I have
> to
> > > > congratulate you!
> > >
> > > You are probably right - the 25min it took me to find your
> information
> > > and post a reply to your mail, could have probably been better
> spent
> > > on something more productive.
> > >
> > > However, the fact remains that little has changed in your
> thinking
> > > since the "last century", as you put it:
> > >
> > > "...and optimise on periods. When this line goes above zero, buy,
> and
> > > when below, sell. Please let us know the results."
> > >
> > > Optimization did not work in your ancient webpage, and it still
> does
> > > not work in the new millennium.
> > > Optimization favors past outliers, a favourable performance which
> is
> > > very unlikely to repeat in the future.
> > >
> > > Encouraging others to take this path when one should know better,
> is
> > > misleading to say the least. Dressing it up in irrelevant jargon
> does
> > > not help either.
> > >
> > > Enough said on this subject from me.
> > >
> > >
> > > jose '-)
> > > http://www.metastocktools.com
> > >
> > >
> > >
> > > --- In equismetastock@xxxxxxxxxxxxxxx, "MG Ferreira" <quant@xxxx>
> > > wrote:
> > > >
> > > >
> > > > Mmmmmm, the gloves are coming off it appears.
> > > >
> > > > If I look at the site mentioned, I see a spectral chart used to
> > > > illustrate the concept of optimisation of strategies. I see an
> > > > example of the space one has to deal with for a simple, cross
> > > > over strategy. I don't really care about the trading costs at
> > > > this stage, since I am interested in the specific topology I
> > > > have to deal with. Dealing costs will be added once I
> understand
> > > > this, as will proper tests to ensure I am not overfitting and
> so on.
> > > >
> > > > The spectral chart indicates the dangers of just using the
> Fibonacci
> > > > numbers, very popular at that time for that particular market,
> > > > and the benefits that can be obtained by still using a very
> simple
> > > > strategy but just fine tuning it a bit.
> > > >
> > > > Interesting to note that the site you refer to was cancelled in
> the
> > > > previous century some time and basically lives on for historical
> > > > purposes. I can assure you the weaponry used today are much
> more
> > > > sophisticated, though it derives from what must have taken you
> an
> > > > awful lot of time to find on the internet, and for which I have
> to
> > > > congratulate you!
> > > >
> > > > One thing I have to agree with, and, trying to steer this back
> to
> > > > the Theta model somehow, is that overfitting is dangerous. In
> the
> > > > Theta model that we utilise, we do not optimise, we simply use
> it
> > > > with an arbitrary period selection.
> > > >
> > > > I am well aware of the high volatility in higher frequency
> data, but
> > > > this does not mean that it can not be used profitably. In a
> geared
> > > > environment, you often simply do not have the luxury of riding
> out a
> > > > position for a couple of days. Even if you were spot on say a
> week
> > > > from now, market movements in the interim may have wiped out
> your
> > > > margin and closed you out at a huge loss - that will cost you
> plenty
> > > > a shirt. So sometimes you have to use shorter term data. We
> have
> > > > tested a variety of series on 5, 10, 15, 30, 60 and 120 and
> above
> > > > minute bars and, interestingly, many can be traded profitably
> at 30
> > > > minutes. We thus prefer this bar length when working with
> intraday
> > > > data.
> > > >
> > > > Finally, let us try and keep this to the Theta model in
> particular
> > > > and in general to technical indicators of interest.
> > > >
> > > > Regards
> > > > MG Ferreira
> > > > TsaTsa EOD Programmer and trading model builder
> > > > http://tsatsaeod.ferra4models.com
> > > > http://www.ferra4models.com
> > > >
> > > > --- In equismetastock@xxxxxxxxxxxxxxx, "Jose"
> <josesilva22@xxxx>
> > > wrote:
> > > >
> > > > "...and optimise on periods. When this line goes above zero,
> buy,
> > > > and when below, sell. Please let us know the results."
> > > >
> > > > And please don't trade the above, because you are sure to lose
> your
> > > > shirt on it.
> > > >
> > > > Whilst it's taken for granted in trading circles that
> optimization
> > > > to historical data (curve-fitting) very seldom eventuates into
> a
> > > > profitable strategy in the real world of trading, apparently
> this is
> > > > not something yet discovered in the theoretical & jargon-laden
> quant
> > > > world.
> > > >
> > > > Here is a perfect example of theoretical/optimization modeling
> gone
> > > > wrong:
> > > > http://www.quant.co.za/OptStrat2.html
> > > >
> > > > Basically, the author claims to have optimized a profitable
> trading
> > > > strategy based on optimized SMA crossovers (and no transaction
> > > > costs), and comes to the naive conclusion that the shorter-term
> > > > indicators generate a much higher profit:
> > > >
> > > > Daily Data (Four Years) 1,335 points total 334 points a
> year
> > > > Hourly Data (Two Years) 1,380 points total 690 points a
> year!
> > > >
> > > > The problem with the above - as anyone with any basic
> backtesting &
> > > > trading experience will know - is that short-term trades
> generate a
> > > > prodigious amount of whipsaw and brokerage/slippage/spread
> costs,
> > > > turning any theoretical strategy into a real-life losing
> strategy.
> > > > Both examples above are very likely to produce major losses
> once
> > > > transaction costs are factored in.
> > > >
> > > > The above mistaken conclusion could be forgiven if it was the
> work
> > > > of some inexperienced bright-eyed young system developer
> straight
> > > > out of school, but the it actually came from someone that
> claims a
> > > > very impressive CV.
> > > >
> > > >
> > > > In the laboratory-like world of theoretical system optimizing/
> > > > modeling, transaction costs and liquidity issues that overwhelm
> > > > profits in the real world are often ignored. Numerous essays
> and
> > > > white papers are written by academics & pseudo-academics each
> year,
> > > > trying to model a system more complex than the most complex of
> > > > weather patterns, yet offering very little insight into the mass
> > > > psychology and real-life vagaries of the markets.
> > > >
> > > > Caveat Emptor.
> > > >
> > > >
> > > > jose '-)
> > > > http://www.metastocktools.com
> > > >
> > > >
> > > >
> > > > --- In equismetastock@xxxxxxxxxxxxxxx, "MG Ferreira"
> <quant@xxxx>
> > > > wrote:
> > > >
> > > > Hi Andrew,
> > > >
> > > > Let us just backtrack a bit. I noted, when coding the T3 and
> IE/2,
> > > > that the IE/2 appeared to be similar to the Theta model, which I
> > > > know to be a good model. So I did not bother to test it,
> assuming
> > > > it also to be a good model. The Theta model implementation
> provided
> > > > is based on what can be done in Metastock in a very short time
> span,
> > > > and was given on request, and has maybe too many shortcuts in
> it.
> > > > If anybody has ever tested the IE/2, I think we can safely use
> that
> > > > as a proxy for the Theta's performance and vice versa.
> > > >
> > > > Now, since I appear to be the defender of the Theta model. We
> (yes,
> > > > I happen to be part of a team) use the Theta model extensively
> to
> > > > prepare short term forecasts of monthly data, such as M3, CPI,
> > > > wholesale trade and so on. We use it as part of an array of
> models
> > > > and we never use the results of just one model, but the Theta
> model
> > > > shines in this capacity as a good performer and often has a
> fairly
> > > > large dynamic weight allocated to it. Here performance is
> measured
> > > > in forecasting accuracy, which usually is a poor indication of
> > > > whether it will work in a trading environment.
> > > >
> > > > But we also use this model, for end-of-day data, in a trading
> > > > environment as part of yet again a suite of models. This is
> quite
> > > > fashionable and dicated by theory as well - using a suite of
> models,
> > > > and I am in a way recommending this to the group and also
> > > > recommending the inclusion of the Theta in such a suite.
> > > >
> > > > Now, let us not run away from the real point, testing the Theta
> > > > model as a singular trading model. I note your observation, as
> well
> > > > as that of some other members of this group, and can well
> believe it
> > > > - that the Theta did not perform well when you tested it.
> > > >
> > > > This is true of prediction models in general, so allow me to
> expand
> > > > a bit. A good prediction model is supposed to predict where the
> > > > market will be in future, say tomorrow. Now, if it is a good
> > > > projection model, then it will be unbaised, so that the market
> will
> > > > be above it about 50% and below it about 50% of the time. The
> > > > residual or error for a good model will be random. So if we
> use a
> > > > prediction model as is, we are trading white noise, and should
> not
> > > > get good results. So we have to apply our minds a bit. I am
> > > > thinking aloud, why is the Theta not performing as I would
> expect,
> > > > so please bear with me.
> > > >
> > > > In our trading model, we do use the Theta model's prediction as
> well
> > > > as its slope. So we extrapolate the model and note the slope of
> > > > this extrapolation and we use both in the model. We have noted
> that
> > > > when the Theta long term line (theta = 0) turns, it often
> indicates
> > > > a turnaround in trend. This could be a better way to build a
> > > > trading model, using the slope of the long term component. The
> > > > slope of the extrapolation is in fact half the slope of the long
> > > > term component, since the extrapolated short term is constant
> and
> > > > the Theta is
> > > >
> > > > ( lt + st ) / 2
> > > >
> > > > so
> > > >
> > > > d( lt + st ) / 2 = dlt / 2
> > > >
> > > > since
> > > >
> > > > dst = 0
> > > >
> > > > Another note, we often take the log of the series before we
> > > > calculate the slope, but this should not make a big difference
> in
> > > > many cases.
> > > >
> > > > Anyhow, try the following test
> > > >
> > > > linregslope(log(CLOSE),periods)
> > > >
> > > > and optimise on periods. When this line goes above zero, buy,
> and
> > > > when below, sell. Please let us know the results.
> > > >
> > > > Note that the parameter should be on the long side. It should
> > > > ideally be above 30 for a number of statistical reasons that I'd
> > > > rather avoid for now. I think a good starting point would be 50
> > > > days and test up to at least 250.
> > > >
> > > > Regards
> > > > MG Ferreira
> > > > TsaTsa EOD Programmer and trading model builder
> > > > http://tsatsaeod.ferra4models.com
> > > > http://www.ferra4models.com
> > > >
> > > > PS : I *really* appreciate your opening sentence.
------------------------ Yahoo! Groups Sponsor --------------------~-->
In low income neighborhoods, 84% do not own computers.
At Network for Good, help bridge the Digital Divide!
http://us.click.yahoo.com/EpW3eD/3MnJAA/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/
|