System Implementation Exercise

mTrader, read from the biginning... A lot of threads in this forum are like 50 to 80 pages long.. but they contain a wealth of knowledge...

IF the seniors need any hand in testing any of the systems, I am happy to jump in and give a hand where ever needed.

Dear friends,

I last posted in this thread on April 06, 2007. Due to my work schedule I had to be away from home on tour and had to suspend posting in this thread. Meanwhile all other participants also stopped posting their results due to various reasons.

Now I am back but I think there is no point in posting the results during this past period of absence. Also, it may not be worth continuing the exercise alone. However, for the sake of completion I would like to post the present position.

When I posted last, the portfolio had the following stocks:
Based on the MABIUTSHB (along with RSI) trading system the present position is as follows:

WOCKPHARMA gave SELL Signal on 18-06-07 and sold on 19-06-07 @Rs 403 giving 16.4% profit. RPL and UTVSOF have not given SELL signal till today. the latest closing price of RPL is 114.2 (unrealised profit is about 65%) and the latest closing price of UTVSOF is 508.7 (unrealised profit is about 69%).

One of these days I will post my views on the whole exercise.

Regards

-Anant

Welcome back,Anant!

Saint

Thank you Saint for your interest. I look forward to receiving your critical and expert opinions and advice.

Regards

-Anant

Dear friends,

Today I am writing about my views about the 'System Implementation Exercise' mainly with respect to the system I implemented. As on today, RPL and UTVSOF are still running without giving SELL signal yet.

The exercise was conducted with different systems and right at the beginning we faced a substantial crash in market. Almost all systems suffered because of the crash. However, we continued the exercise. Later on we again faced market crash but the exercise continued. As we were dealing with virtual money, there was no need for panic. Each system behaved differently even after the market recovered. Each of us had to figure out what is going wrong. May be it was incorrect Stop loss, May be it was stock selection criteria or some other reason. I am writing here about what I figured out of the whole exercise. Some of these views may be similar to what others feel and some views may contradict their views. We can discuss about these once others also express their views on the whole exercise.

The whole experiment can be divided into three parts:

1. List of stock selected for the exercise.
2. Stocks selected for trading and the criteria used to select them.
3. Stop loss.

I will discuss each one here:

1) List of Stocks selected for the exercise: Each day the list of stocks was selected using an AFL given by Murthy. It mainly depended on price, Turnover, Average Volume and Total Volume. For each of these a lower limit was defined.
My Views: As the list was generated each day, the stocks figuring in the list varied each day. The number of stocks was about 200 but it varied each day. Though a specific group of stocks did not change each day, there were a few which would get included one day and excluded on some other day. In fact, if we consider all the stocks that qualified during the period the total is about 400 stocks. I feel proper attention was not paid to this problems and I feel this affected the outcome. In my opinion we should have selected a list of stocks prior to the exercise and stick to this list throughout.

2) Stocks Selected for Trade and the selection criteria: Each day the AFL was run on the selected list and BUY-SELL signals were generated. All SELL signals were implemented the next day but for buying there were restrictions. The rules stipulated a maximum number of stocks to be purchased subject to a maximum prescribed amount that can be spent on a day. There used to be quite a good number of Buy signals every day and we had select only a few of them for buying so that the rules are not violated. The criteria for selecting was mainly left to the individuals though there were some general guidelines.
My Views: Many times the stocks selected for trade turned out to be poor candidates while those excluded went on to give handsome returns. It is difficult to set particular rules to select a few stocks for trading. It would be better to trade all the qualifying stocks. But this is also not easy to follow as we can not trade about twenty stocks every day with limited resources. It would be better to get a smaller main list of stocks in step 1 above. Also some specific criteria have to be formulated to select really strong signals.

3) Stop loss: This was one of the hot items of discussion prior to and also during the exercise. Some advocated a fixed stop loss, some favored a trailing stop loss. There was also mention about no stop loss. In the last part of my tests I was not using any stop loss.
My views: I think this subject needs thorough discussion and only then we can apply stops intelligently to maximize the profits and minimize the losses.

In addition, I have one question which needs answer from the knowledgeable members in this forum: SHOULD WE HAVE A STOP PROFIT ALSO ?

These are some of the thoughts I wanted to share. You are all requested to give your views, comments and suggestions.

Regards

-Anant

Welcome Mr Asnawale.
I Was Regularly Reading This Thread It Was Definately A KNOWLEDGE Prvider For Me. In Certain Situation(like Intraday) Stop Profit May Be Pequired Like Stoploss.
During Intraday Trading Many Times My Stoploss Is Triggered. Had I Decided To Have A Stop Profit Idea I Would Have Covered The Profit(fot 1-3 Point Up) Instead Of Someloss (due To S/l Triggering)
As There Are Many Knowledgeable Memberes Here, They May Please Guide. But I Do Agree With Your Stop Profit Idea

Hi Simple

Yes indeed designing is sexy and implementation is rather difficult. While designing you put lot of creativity and knowledge to use and Implementation destroys all that …

The best thing with implementation is that it exposes the positive aspects and weakness of the system designed by you. It helps one to understand the Indicators used by you very well. The weakness of the indicators used would be fully exposed.
Implementation gets to back to the design table. I think this a process of refinement and goes on till finally we get our holy grail.( mind you, I said our Holy Grail )

Of course there was no comparison with buy and hold, nifty stocks etc. Moreover the period was too short to compare with buy and hold.

It would have been wonderful if we could test with different ranking systems, different MM styles and different RM styles. But for commonality sake we had set common rules and parameters for the implementation. We did have a ranking system basically based on volume though each participant had a variation of his own keeping volume as the base.

Overall it was a great experience. I truly wished that more newcomers would have participated as this exercise would have given them a good insight into the nuisances of trading. Tough it is not the real thing the group participation would have been much better than paper trading alone.

Warm regards

Karthik

I wont necessarily agree with that, the key to smoother implementation would be rigourous testing as well as accounting for various execution scenarios which could be deceptive for daily data based systems.

I agree with the refinement point and for a technical reference here are the general steps involved for trading models

1. Translation – Express observations of the markets in some form of mathematical language or implementation.
2. Execution –Use this model for backtesting or real-time data to make new predictions or conclusions about the markets.
3 Validation – Interpret and test the resulting predictions to assess the extent to which the model reflects reality.

Dear CV

Nice to have your views here..

I think I made the wrong choice of words.. Pardon my lack of dexterity in expressions…. When I said, “Implementation destroys all that…” I meant many a times the creativity we express in Translation proves to be lacking when actually we try implementing (Execution)…

Here we had translated our ideas and tried execution on a real life situation rather than backtesting…

However I fully agree that the execution was rather limited. Somehow we tried had put a set of rules to achieve commonality. On hindsight I feel that was not right. Maybe we should have allowed a larger degree of freedom to the participant with their execution.
Maybe we can try that too provided with have the required quorum.

Warm regards

karthik

Points noted Karthik.

Another idea would be add additional triggers for intraday execution and rating them individually, this could iron out few of the tricky issues when executing on different timeframes