The Holly Grail Vs Optimization

[walterw 10]

Hello dear fellow traders ¡¡

 

I was thinking this weekend as I am preparing a technical update, why do we traders have certain tendency to study in new frontiers of technical analisis ?

 

Is it the look for the Holly Grail ? mmm... or is it that technical optimization is really posible ?

 

First of all let me make clear that the Holly Grail does not exist... you will never find it... and chimps dont drink water with grails, less sacred ones...

 

NOW, optimization DOES EXIST... and this is as true as you can change the way you trade with very simple optimizations...

 

Evolution of a method its posible, it takes time and normally, its funny, but the optimization comes in a simple presentation... too complex things most probably are not too much optimized...

 

The other day I drove on a friends BMW, pure pleasure, very simple presentation of the pannel, not to much buttons, not too much lights, simple presentation but a great optimized piece of automotive engineering ¡¡ now BMW has been optimizing their cars since some time now... lots of years of getting their product optimized, are their cars perfect ? no, not perfect, but yes optimized... same happens with Technical Analisis... can it be perfect ? NEVER.... can it be optimized ? thats for sure...

 

I am a strong beleiver that practical experience is the best teacher in order to produce optimization on Analisis... trading has attached a series of sensations, joys, frustrations, discipline, undiscipline, patience, non patience, etc... all this must observed in the process of optimization...

 

for example, my friends BMW has some nice lights outside on the doors when you aproach it from outside... mmmm maybe some BMW engineers thought thru experience that some times having light when aproaching a car may be very usefull on a dark night... you see... as time goes by, we find new things that could be revisited on our technical methods...

 

Thats why some traders drive on BMW`s (technically speaking) and other ride an old Oldsmovile year 60... you see, it depends specially on your actitude towards accepting what can be optimized...

 

An open mind is needed for this... the right actitude to accept when you are not totally correct on your performance and offcourse you must beleive on the fact that optimization is possible...

 

Are you looking for the Holly Grail ? stop doing so.... better use your time for optimization.... cheers The Chimp.

 

PD: BMW X6 aint that nice ?

[darthtrader3.0beta 10]

I've posted before, I just think the whole concept of the "Holly Grail" should be disgarded. Its just a poor metaphor for a fruitless path as far as trading research goes.

The biggest danger with the holly grail mindset IMO is that it ignores your opponent. We aren't really trading against eachother here, we are trading against hundreds of millions of dollars in data analysis and infrastructure trying to analize as much data as possible and find the smallest relationships between the data sets.

To me if you want to sharpen the edge of your analysis, the trail has already been cut for you. Its obviously not in throwing 40 indicators all doing basically the same thing, then using the tradestation optimizer to dance through different settings find patterns in the clouds.

The way to do it is store massive amounts of tick data of related instruments and then use the well studied field of data mining algorithms to find non random patterns. The barriers to entry are nothing financially, the software is all there ready to use for free basically. The real barrier to entry to the next level of analysis is that the knowledge and background needed to take advantage of this stuff is substantial. Just because something is hard to do, to write it off as the search for something that doesn't exist, when your literally trading against the thing you don't believe exists is foolish.

[walterw 10]

  darthtrader3.0beta said:

I've posted before, I just think the whole concept of the "Holly Grail" should be disgarded. Its just a poor metaphor for a fruitless path as far as trading research goes.

The biggest danger with the holly grail mindset IMO is that it ignores your opponent. We aren't really trading against eachother here, we are trading against hundreds of millions of dollars in data analysis and infrastructure trying to analize as much data as possible and find the smallest relationships between the data sets.

To me if you want to sharpen the edge of your analysis, the trail has already been cut for you. Its obviously not in throwing 40 indicators all doing basically the same thing, then using the tradestation optimizer to dance through different settings find patterns in the clouds.

The way to do it is store massive amounts of tick data of related instruments and then use the well studied field of data mining algorithms to find non random patterns. The barriers to entry are nothing financially, the software is all there ready to use for free basically. The real barrier to entry to the next level of analysis is that the knowledge and background needed to take advantage of this stuff is substantial. Just because something is hard to do, to write it off as the search for something that doesn't exist, when your literally trading against the thing you don't believe exists is foolish.

 

Are you referring to neural networks ?

[darthtrader3.0beta 10]

  walterw said:

Are you referring to neural networks ?

 

Well neural networks are one of many well studied data mining algorithms, the perceptron model is one of the oldest. The "holly grail" meme drives me so nuts because I truely believe it short changes us as retail traders as far as what software is brought to market. It just seems odd to me that we can buy software that can plot every absurd idea known to man, ships with every useless indicator ever devised but you can't buy off the shelf retail software that stores the thousands of dollars of TICK data that we are all streaming to our machines daily in a useable format like SQL. You basically have to do it yourself.

I installed some poker software the otherday that installed SQL light for a database to keep track of players betting and it has simple tools for manageing the database. I guess my point is I've never been interested in something and part of online communities that shun technological innovation like retail trading does. Music technology is probly a better analogy. Of course you can still record music with a 4 track recorder but the music world jumped all over the advancing computing power in the 90s like there was no tomarrow to the point that you can do things with an off the shelf computer today that was totally unimaginable 15 years ago for any amount of money.

Retail trading software is basically the same as it was 15 years ago with some minor tweaks. The only guys who have truely embraced technological innovation are the big guys.

[januson 10]

Before one talks about The Holy Grail we need a clear and concise description, can anybody provide that?

 

I can share some of my thoughts in the meantime...

I believe that much of the money spent in algorithm trading is in developing user-friendly interfaces and also in developing methods for smooth and seamless distribution of large orders. I think it is extremely hard to keep as low profile as possible during distribution, the idea is to influence the market at least possible to obtain the best price.

Likewise the hardware storage costs alot of money - storing tick data can be extremely difficult both in hardware and software.

 

My guess would be that an individual can build a trading system and store data at a price around $ 100,000, this would cover 20-25 shares divided into 2 different markets. This is very detailed I know, purely estimations build from my experience. Now we can do some math... try to cover 50.000 different equities spanning 20 different markets - and this is only equities, take forex, options etc. Wuaaaaa what a bill

 

I am 100% sure that these things here cost huge amounts of money:

http://www.microsoft.com/casestudies/casestudy.aspx?casestudyid=4000002610

Edited December 9, 2008 by januson