Trading System Junkyard

We already outperformed the benchmark , but where's the system

Hello Ajay

Accepted, didn't occur to me as tunnel visioned with trading only a single instrument.

Regards
Sanjay

Hello Abhijit,

To take it further lets say we choose a timeframe and time duration and we carry out this simple exercise.

1. Find the maximum possible applitude extraction if the system catches all the tops and bottoms.
   (One of the ways to do this is by ploting a zig-zag across the entire data set and measure the sum of lengths of all the legs of zigzag)
   
   
2. We divide this sum with number of bars in our dataset (trend/bar).
   
   
3. We get the average range (H-L) of all the bars (range/bar).
   
   
4. We can also get trend noise ratio 2/3 for our choosen TF .

Regards
Sanjay

Theoretically, yes. Another advantage is that we do away with the philosophical discussion about market being efficient or not.

However, during crossovers of returns, slippages while shifting strategies within the trading system occur which pull the returns slightly lower than the benchmark. This problem is worse when the strategies within the system are highly correlated.
That's why I use "~".

Oxy has the brains of a supercomputer. In his lightening fast output, he has already taken a no of stuff as given (which is not the case for ordinary mortals like us) and taken the thread miles ahead. Thanks to him we already have a system to test against the benchmark and decide which one should continue. But for the layman like me, let me see what all subheadings I can put off the cuff before reaching the peak where Oxy is sitting (difficult to catch up with these teens with old legs!).

1. The trading premise
2. Timeframe and Data to be considered
3. Entry, management, Exit
4. Ability to actually put them into PRECISE rules on paper
5. Ability to put them in a computer language to maintain the exact consistency
6. Ability to afford the right software that accomplishes the goals.
7. Actually checking whether the rules are being implemented by the system
8. Checking for initial results

After this we can use some benchmark (I wont use it right now though).

Off to Bahamas now. Hope to catch CV fishing over there. Enough fodder for weeks for you guys (1 post for CV) to post/discuss/tear apart/rebuild.

You missed one crucial point.

The ability to adapt. All systems are useless without their ability to adapt to dynamic market conditions.

my 2 cents ...

Try to be focused ... have a different specialized system for different market conditions.... instead of making a generic one... which might feel elusive.....considering the number of variables will increase exponentially...

But have a good system to select the different one each time... .. then you have a generic system ...
( what I meant to say is build the adaptability layer on the top .... )

Hope I make sense...

I have just given the brief steps till the point where Oxy is sitting. There is no right or wrong. Its your approach. You may try to build an adaptive system, have each system for a specific mkt/system combination or rely completely on 3rd party machine intelligence code w/o knowing whats inside to build an adaptive system. There are many small factors to be taken care at every step that cannot be explained here. Even if you expand the steps that I have given , you will find that you will get at least 100 tactical steps for those 8.

Unfortunately for most people system building is trying to use built in indicators in off the shelf packages, rushing to back test and then optimize 1 entry, 1 exit till some promising results are shown.

Let me try a Narrative process.From my residence to Airport it takes 1.30 Hrs by car on an avg speed of 60 km/Hr.Whatever Set -Up i might have & think that to be a Mechanical thing (to alley the Emotion part out of it) CAN BEST be a compared with a Bicycle traversing to the Airport paddled by a Highly Diabetic non athelete aged person.
Day before yesterday IMF sold 400 tonnes of Gold from their Holdings.This Headsup is in any way going to effect my everyday Nifty Fut trading,probably not but in terms of 'Asset Allocation' of Big Funds this INSTANTLY needs REBALANCING.(Scaling out from 'Precious Metal' & allocating to other asset class.)
My Rs 100.00 of Jan 2008 ,if the inflation is still contained within 8% per annum,needs to fetch me just 8% Only TO REMAIN STATIC Without GROWTH.
Now as the Mkt is at 4700 level Funds will be more invested in Equity (which may be miniscule say 15% )than when in 6000,where they where scaling out from Equity(which at that time may have risen to 45 % )& Reallocate to even 'Hussain's Paintings' .
These are only to put forward an Idea what this System beast is.
Coming back to Abhijeet's point No. 2 (i am well aware that Point No 1 will require lots of attention for thrashing it out),i have only taken TF & Data.Which may seem very innocuous that we will take 5 mins & 60mins & for Data only Close.
The entire Data-Minning ; Time Frame Compression/Expansion should itself require seperate threads for Layman like me.
Layers are above my Heads layers of Cognizence, at this Stage.
Ajay's approach is not even clear to me ,becoz I THINK ONLY A BUY & SELL as that is only visible through my Blinkers.
To Successfully Manage a SINGLE Trade few Auxillary Trades are also nessecary.A dumb fool like me is unable to Fathom the colinearity how can i digest the Huge variables involved.

Lot of us including me ONLY HAVE A SET UP thats all,we cant grasp what actually a even rudimentary basic System is,as we do not know the beast nor have seen one .But in Net have seen 'A Winning System with $6000" & know it requires Back Testing ,so to me a Set Up with back test is a System,not only i am in Fool's paradise but trying to judje the Elephant by it's Tail in a Black Out ambience.

It's by Keynes. To warren buffet, you should ascribe double speak and not such precise statements.