Experiments in Technical Analysis

Hi Friends

Hope you all had a wonderful joyous Diwali celebration. With a truncated trading week and the fno expiry the coming week may not be very eventful. The week after could be exciting. Meanwhile we will continue with our journey into the exciting world of Technical analysis.

Trending or ranging….. That is a question that often bothers us in our trading. This decides our approach. We already saw that ADX and RAVI helps us to understand whether the stock is trending or ranging. We saw that RAVI does a better job many a times compared to ADX.

Now let us also look at other indicators useful for the same purpose. One indicator widely used is the linear regression slope.

Let us have a brief look at the linear regression slope. We will not go into the details of the mathematics as it is a wee bit complex.

The Linear regression method uses a “least square method” for finding the best-fit line for a series of prices. We use the slope of the best-fit line to see if the prices are moving up or down. The slope can tell us if the stock is trending up or down. The regression slope also depends on the period considered.

Another calculation from the linear regression analysis is the Correlation Coefficient otherwise known as r2 (R square). This value varies between 0 and 1 and provides an indication of how reliable or strong the trend is. When the price is moving listlessly sideways the r2 value is near zero and when the trend is very strong near one. So the r2 can be used as confirmatory Indicator for the trend strength. The r2 value depends on the number of periods used for a regression analysis. For each period we can calculate a critical value beyond which the trend can be said to be reliable.

Luckily for us we need not go into complex calculations as most software have inbuilt indicators for the linear regression slope. The r2 values are also not difficult to calculate and the critical values for various periods are already available.

Before we go further on this let us just have a look at the linear regression slope indicator with the r2 values plotted as histogram. Chart enclosed...

More on this a little later….

This is a useful and good SIGNALLING thread .. Keep the work on ..

Hi Friends

Hope everybody is enjoying their long holidays. And thanks, Jataved for the encouraging words.

Let me continue with the Regression slope Indicator. The Regression slope Indicator (RI) should be always read in combination with the r2 indicator. The slope gives the direction and r2 gives the strength of the move

The RI has some advantages over ADX and RAVI. The RI is more response to the price moves. So many times the RI shows the start of the trend much earlier compared to both RAVI and ADX. See example 1.

Also at time when prices rise unevenly the ADX does not respond well showing that the trend is not tradable. But the RI indicator responds better showing that the up trend still exists. See example 2.

I will post the codes soon so that you can test it . I am still working on the metastock codes lest the MS guys will take me to task

Warm regards

karthik

Hi Karthik,

I was going through guppys site and I came across a expert advisor. I needed your help in the same.

Under Trends add the Bullish formula :

ADX(14) > 20 AND ( Mov(C,15,S) > Mov(C,30,S)) AND ( Mov(C,5,S) > Mov(C,30,S))

and the Bearish formula :

ADX(14) > 20 AND ( Mov(C,15,S) < Mov(C,30,S)) AND ( Mov(C,5,S) < Mov(C,30,S))
In Symbols add this condition

Now what I understand from the formula i have put in the box above is that if adx is more than 20, and mov avg of 15 and 5 day is more than 30. Also, when stoch is less than 30 and one day before it was more than or equal to 30. This is the main criteria and if this is met then the signals will come as per the bullish and bearish formulae which are just based on ADX and mov averages.

I am attaching the chart also where i have put this advisor. Please correct me if I have understood this wrong.

Rgds

Rahul

Attached Images

Amara.png (31.6 KB, 68 views)

In my search for the PERFECT trading system (holy grail!), i backtested MABIUTS from karthik. It generated 125% profit on backtesting for a period between Jan 2006 to Oct 2006.

Its a simple cross over system, using EMA.
Backtest report enclosed.

Regards
Siva

Note:-
BTW, Karthik's latest beta generated a loss of -44%

Attached Images

MABIUTS- Backtest Report.PNG (36.1 KB, 117 views)

Dear Siva

Thanks for Backtesting my MABIUTS system. It is a very simple yet robust system and has always given me good money. Very good one for the beginners.

Seems like I am wasting my time looking for a more complex system

Your footnote seems to prove it

Warm regards

Karthik

dear Siva,

Is the back test coducted on a single stock or a number of stocks? I am working on 30 stocks. The cross over technique is one of the strategies I have included. I am about to complete the tests and shortly write about them. It should have been over by now but I had to go out of station for some time. Also ceratin events at home kept me away from working on the system and so I could not complete. By this week end I hope to finish it.

-Anant

Dear Rahul

I went through your system.

First I don’t understand the purpose of adding the stochastic condition.

Stoch(5,3) < 30 AND Ref(Stoch(5,3) ,-1) >=30 would mean that the stochastic indicator today is less than yesterday’s and is crossing down below 30. That would mean than you would be buying when the stochastic is crossing down into the over sold region. This is contrary to the normal method of buying when the stochastic leaves the oversold region and cross above 30. The problem in this is that the stock could remain in the oversold region for a long time. Please do review.

Also marrying stochastic with trend following indicators would be quite ineffective. They make strange bedfellows. Even if you use a condition of stochastic crossing above 30 along with the MA cross over conditions, most of the time you will miss the entry or you entry would be delayed as the MA cross over and stochastic crossing above 30 would rarely be synchronous. IMHO stochastic is very ineffective in detecting trends. Of course they are quite useful in providing re-entry points in a already trending market.

Also please consider one thing when you combine different indicators. The parameters should be reflecting similar time frames. For example the stochastic (5,3) would refer to a short term of a week. At the same time you are using the ADX(14) which would refer to almost three weeks data. You could consider using stochastic of different parameters say 15 , 7.

Also using ADX in isolation (without DI+ and DI-) in a mechanical system could prove faulty as ADX > 20 is also true for a down trend.

Also check your code for the “trend” in your exploration. Maybe you do not have the stochastic condition there. That is why you are getting that bullish signal. I am enclosing my chart with the conditions given by you. (amra 1)

The system without stochastic condition gives much better results. Just to give you some ideas look a the same system with slight modifications as follows

See chart enclosed (amra 2)

Regards

Karthik

Dear Karthik,

thank u So much. I will check my condition again in the explorer. Actually i am not planning on using it. I found it somewhere so I was just trying to understand the formulae and yes also see how it works on different charts and I did not find it very useful. And your explanation on what to keep in mind while designing a system is very useful.

Ok karthik since this method topic has come up and what goes with what. How will these go together and what would be the reasons they would or not go together. Moving Avgs, MACD Crossover, And RSI crossover with its own Moving avg. I am using this and finding it useful in silver trading on a 15 minute chart.

WIll we have to change the time period of the indicators if we use it on daily chart. for a 15 min chart I am using normal macd, rsi 14( its ma of 42) and moving avg crossover of 30 days and 15 days. This is being used for the 15 mins chart and working good so far. Will i have to change it for daily chart.

Rgds

Rahul

[quote=sivavkm;62676]In my search for the PERFECT trading system (holy grail!), i backtested MABIUTS from karthik. It generated 125% profit on backtesting for a period between Jan 2006 to Oct 2006.
.......................

thanks siva for the back test report.

enouraged by your post, i too checked the karthik's MABIUTS and another 26 trading systems, i could get from different sources and tested the same. i present the summary of my observations as below.

each of the trading system is developed for different holding periods, different trading methods and a comparision may not be all that perfect. but with the SHORT TERM TRADER in view, i have backtested nSE stocks with reasonable turnover during the period 1 jan 2006 to 30 oct 2006. the results of the systems which gave more than 50 % profits only is given below in descending order of profitability.

the profits varied from as high as 99 % to as low as a LOSS of 25 %. so, selection of the trading system is very important for any trader.

note: no two tests are likely to give the same results as the test 'settings' may be different.

MABIUTS : 99 %
T3 : 93 %
3d Reversal : 91 %
ATR Trading system : 61 %
Bull Bear Fear : 58 %
GANN : 53 %

amomg the remaining 21 systems, 12 made profits ranging between 49 % & 0 % and 9 made losses, the max LOSS being 25 %.

this post is made as an acadamic exercise only, and NOT to promote some body or to criticise any other's trading system.

what comes out important is that one should identify his trading style and identify the trading system that gives the best results for that trading style.

i congratulate karthik, whose system stood on top of the report.

all the best.

murthymsr