Experiments in Technical Analysis

[murthymsr]

dear sanjay,

thanks for the draft proposal for establishing the 'template' for the development of a trading plan. i share/supplement some of my thoughts as below.

Objective: (Mission Statement)
To try to earn a profit of n% per month consistently on the capital deployed while keeping the drawdown periods to minimum. (plz propose a value for n)

i feel 5 % profit per month is the minimum to consider a system as acceptable.

break-up:
safe interest earned on capital deployed: 1%
additional iterest due to the risk involved in the very nature of trading : 1 %
expenses like system, internet, electricity & accommodation : 1.5 %
wage component of time spent on the trading : 1.5 %

Maximum loss per trade should be kept below 3.33 % of available capital.

i feel,this may be limited to 2 %

Trade Execution will be done the next day at opening.

this is more hypothetical in nature. may be, we can think of the average of the high & low of the following day, which is much nearer to reality.

The entry/exit signals will be available at EoD after the data is downloaded and the scan/explorations run (will use the one uploaded by Murthy, few posts previous to this).

while the template prosed by me removes the low value / low volume stocks, the vales for each of the parameters may be given a final thought.

close>20 & volume>10000 & turnover> 100 lakhs selects about 340 stocks.
close>20 & volume>20000 & turnover> 200 lakhs selects about 270 stocks.
close>30 & volume>30000 & turnover> 300 lakhs selects about 230 stocks.
close>40 & volume>40000 & turnover> 400 lakhs selects about 200 stocks.
close>50 & volume>50000 & turnover> 500 lakhs selects about 180 stocks.
close>100 & volume>100000 & turnover> 1000 lakhs selects about 120 stocks.

so, the preferred limits that may be set are:
close> 50 & volume> 50000 & turnover> 500 Lakhs.
this gives a tradeable set of about 180 stocks, which will be large enough & practicable too.

Currently all the systems proposed are based on Weekly and/or Daily timeframe. If any member would like to propose something that considers intraday data (JDMs Idea), it would be a great learning experience for all.

any trader would normally place buy/sell orders as per market conditions prevailing during the trading hours. the actual trading is done intraday only, i.e., during market hours. but this consideration may be postponed to the next stage of analysis/ testing and for the time being, we may limit our study based on buy/sell signals are generated based on EOD data.

hope, we will be able to close-in on the various open ends and evolve a common system by which. we can asses on equal footing the different systems presently proposed and to facilitate improvement.

but one contrarian thought. variety is the spice of life. should we hammer it down?

thanks for the initive.

murthymsr

 

[karthikmarar]

Hi friends


Sanjay , thanks for the template. Great work.

Maybe we should consider a strategy for the short side also like murthy suggested... Afterall it does look like we are near a top...

I wish we had more active participation in this venture... COME ON GUYS... join in... We need more inputs .. promise you, it is going to be a lot of fun and learning...

regards

Karthik

 

[SGM]

Maybe we should consider a strategy for the short side also like murthy suggested... Afterall it does look like we are near a top...

Hello Karthik,

Yes, it does look like the markets are toping out, but that is what it will make the exercise more exiting. It will be quite a learning process to see how the systems perform in the current conditions.

We definately need to also consider the short side of the trade. Having decided to go the FnO route, that would be an advantage we should not miss out on.

The current systems are not good enf for the short side, that is what some basic back testing shows up. So now we will be awaiting an afl/system from you and maybe Murthy for taking the short trades.

I think we should give it some more time for inputs from other members and then finalize the template.

Warm Regards
Sanjay

 

[see3d]

We definately need to also consider the short side of the trade. Having decided to go the FnO route, that would be an advantage we should not miss out on.

The current systems are not good enf for the short side, that is what some basic back testing shows up. So now we will be awaiting an afl/system from you and maybe Murthy for taking the short trades.

My own opinion is that if a system is not symetrical (i.e., the opposite conditions work for shorting profitabily), then good system results are just due to the market conditions for the stock (bull or bear) being favorable during the testing period. Besides, it is a very good strategy to be able to have both long and short positions at the same time in the portfolio to protect capital against overall market shocks. If you have long positions with no margin, you are allowed to have an equal value of short positions (which double your leverage), without having to pay any margin interest --but you will have to pay out for dividends on the short shares. The log side will do better in a bull market, and the short side will do better in a bear market.

Regards,
Dennis

 

[karthikmarar]

Hi Sanjay, Ananth


Here is a experimental afl of Mabiuts-K with some short side strategy. Can you check it. Now also included a Min volume parameter to exclude penny stocks.

Like dennis pointed out it seems to work ok during bull and bear phase but miserable in sideways market..

regards

Karthik

 

[asnavale]

Hi Karthik,

I have downloaded the AFL for MABIUTS-KS and will try. Let you know the results. Good to have people like you around who are innovating and sharing with others.

You had earlier replied to my query regarding an AFL that you would try with beginvalue qnd endvalue. As far as I remember these are useful when you mark certain range on the chart by double clicking. That means we have to take individual chart and apply the AFL every time. What I am interested in is just explore one or more stocks at a time without going to the charts. The range should be selected based on the dates/number of bars/number of days in the exploration window. Probably a FOR LOOP with a test of given criteria may help. But I am still unable to come up with a solution. I will keep trying. Will let you know if I can get something which works.

One of the applications I am thinking of to utilise such an AFL is to draw a regression line from a particular date/bar to any other date/bar specified. Another is to calculate values such as exponential moving averages starting from a particular date/bar such that the data prior to that date/bar is completely ignored. Let us see what we can do about these problems.

Regards

-Anant

 

[asnavale]

Hi Sanjay,

The plan you have chalked out is very good and the suggestions of Murthy are worth considering. Let us wait to see if any more suggestions come from other members.

I would request Anant to take up the implementation of MABITUS H.

I will surely do it. As you know, in the current form MABIUTS-H gives too many signals. Need to eliminate the loss making signals as well as those making insignificant gains. A couple of changes I tried but the profits decreased in line with the decrease in number of trades. One change I tried was to follow a BUY only if the EMA is upwards and SELL only if EMA is downwards. Though it reduced the signals to about a third of original number, the profits also reduced similarly. This means although the strategy picks up many signals, it does give mostly profitable signals. Need to fine tune it for more improved performance. Still Trying. Will post when some progress is made.

Meanwhile we can proceed with the trade plan as suggested and work out as a team. I feel we WILL come out with some useful strategy.

Regards.

-Anant

 

[SGM]

I will surely do it. As you know, in the current form MABIUTS-H gives too many signals. Need to eliminate the loss making signals as well as those making insignificant gains. A couple of changes I tried but the profits decreased in line with the decrease in number of trades.....

Dear Anant

Try using average size of profit-loss/trade and/or profit/ bar (day), as one of the criteria to judge the performance of the changes you are trying in MABITUS-H.

Regards
Sanjay

 

[SGM]

Hello Dennis,

My own opinion is that if a system is not symetrical then good system results are just due to the market conditions for the stock being favorable during the testing period.

I agree with you on that; Point Noted, thanks.

It is a very good strategy to be able to have both long and short positions at the same time in the portfolio to protect capital against overall market shocks.

I guess, this strategy will work best when the overall market is in a sideway phase. A delta neutral portfolio will protect us from the breakouts against our position, but at the same time it will also limit our gains. Anyway wouldn't know much about it, just shared my thoughts.

Looking forwards to more inputs from you

Warm regards
Sanjay

 

[ragh_ash]

HI Anant,
first of congrats on gr8 work being done!
i have followed this thread for a long time and literally thousands of my doubts have been cleared just bcose of this thread.
i wanted to add my opinon (for what it counts?)
donot wory abt loss making trades, whipsaws are integral part of Trend trading(that's our theme)
instead let's try trailing stops. also we trade whenever system gives a signal. only then we will be profitable.
to quote ED Seyota
"To avoid whipsaws, don't trade"
regards
PS : it is just my opinion , i may be wrong also!