Experiments in Technical Analysis

Hi Kalyan, you have put it in a short and simple way. That is correct.

However for Uma, I will explain it in detail:

If we restrict to the bands we are restricting our gains. If we try to widen the bands to 2 SD level, 95% of time we protect ourselves but it leaves only 5% of opportunity to make profits or cut losses. It is true that within the Bands you are with the trend and safe. But we gain from the breakouts which go out of these ranges. When the bars move above the upper band it means that the prices are rising sharply and faster than the average. When the bars return to the zone between the bands it means the price rise has a reached a maximum and we are on the other side of the peak. So, we sell at this re-entry point which gives us a price near to the peak. When the prices fall further, the bars eventually penetrate below the lower band and the fall is shrper than the average. Since we had exited when the bars came below the upper band, this sharp fall will not affect our position. When the prices recover they start rising and crossover above the lower band. This signals the end of fall and we buy. This ensure that our buying price is as close to the lowest price as possible. By making the bands at 1 SD level we have a sfety of 65% along with the 35% chances of making extra profits. In a way the upper band acts like a support line, when it is broken down we exit. The lower band acts like a resistance line, when it is surpassed we enter.

It is true that TREND IS OUR FRIEND and we should follow the rule by restricting ourselves to the trend. But we should remember:

One who breaks the rules is a fool.
One who follows the rules is a wiseman.
But ONE WHO KNOWS WHEN AND HOW TO BREAK THE RULES IS A GENIUS

I hope I have cleared your doubt. If you still have doubts please free to ask.

The MABIUTS-H(B) which I posted is not yet foolproof. It has some limitations and drawbacks. But for the present excercise it is adequate. I will list out the shortcomings later. I am working on how to improve upon them.

Regards.

-Anant

dear sanjay & friends,

our approach has been to identify the trend and follow it. so, the basic issue is trading in a clearly trending markets. but how trending are the markets now? what strategy one shall take when the trend up or down is not clear?



yes, steep drawdowns are likely but may be reduced to some extent by increasing the number of stocks held to 10 or may be even 20, from the present typical value of 5. the number 'five' is relevant in intraday trading, which calls for multiple entries and exits and a very close monitoring. it may not be prudent to ask 'to keep away from the markets' when the trend is not clear. may be we are getting one of the rare oppurtunities to test the system in turbulant markets.

however, system tester may if he so chooses, reduce the exposure and keep minimally exposed rather than being away from the markets. this will help in evaluating the system performance in turbulant / falling markets also.

the MM rules, is one which we could have given enough weightage. 1% loss in trading capital per stock is one that may have to be adopted strictly. but here, as we are basically trading on EOD data, a stock may be sold whenever the EOD price falls to more than 1% capital loss at the next day's open market. is my interpretation of SELL criteria on capital loss correct? sanjay and others amy please clarify.

in a narmal trading situation, a trader will go by a few considerations other than the buy / sell signals given by the system, namely MM, market condition and historical gain / loss of his portfolio till date. how can we incorporate them into the system? to the extant, we do not consider these factors, our test will be hypothetical.

we may have a low value of profit target in an uncertain market, even if the system has the potential otherwise.

we may have to have a well defined guidelines for money management, which are near to the best and also common to all system testers.

i have located a MM calculator (freeware) from stator-afm.com. this is very basic, but will sureky throw some light into the MM principles.

Happy Experimenting !

Regards
murthymsr

Hello Murthy

We couldn't have choosen a better time to test these systems. The real test for the systems would be that they should perform even if we see 10 K or 15 K on Sensex.

Regards
Sanjay

Hello


The back testing of the MABIUTS family shows that, having a small SL will result in losses. The bigger the % SL the better is the over all gain. The best performance is when there is NO PRE-FIXED % STOP LOSS and the systems uses reverse CO as Exit.

Ideally we should use only the reversal (cross over in the opposite direction) as a stop and not use any % Stop Loss. The problem in this approach is that we have to calculate the position size to keep the risk at specified level. For this we need to calculate the price at which next CO will be triggered. Once we can do that, the position size can be determined thus managing the risk involved.

Now the question is how to calculate the crossover trigger (Stop Loss Price) for, let’s say

EMA(C,13) and EMA(EMA(C,13), 13);


The following post gives a way to find the cross over of 2 EMA's


If the price point of reversal is know at the time of taking the trade, we can include it in the MM rules.

Regards
Sanjay

Hi Sanjay

In a simple moving average cross over system there will be a definite adverse crossover signal (Technical Stop) before the % stoploss is hit unless the stop is too tight. So if your position size is based on the allowed risk which is reflected by the stoploss you will be quite comfortable.

Moreover it is not possible to have a stop based on the CO signal mentioned by you. If you plot the CO signal along with the EMAs then you will notice that after a crossover the EMAs and CO signal move in the opposite direction before they converge again. A sample chart enclosed for your quick reference.

But in systems which are not purely based on crossover a technical stop may not appear before the stops are hit. For example in the Mabiuts system which has an additional condition of price crossing above the previous peak , the BUY signal is above the actual crossover point and any adverse CO may occur at points much below your stoploss.

The problem gets a it complicated when we consider positions in Futures. Here we cannot have a position sizing based on % stop loss (like 1% of equity in our exercise) as the position size is in lots. I am enclosing an example where a technical stop makes one lose more than the maximum stoploss. In such cases what we do…

Hope I have not confused you

warm regards

Karthik

Thank you Sanjay for giving consideration to my Spoil Sport .

Dear Murthygaru,

Here I'd like to say one thing. Markets are always uncertain. This we all know. And just to remove the confusion arising out of this uncertainity only we try adopting mechanical trading systems. And in such cases, confining ourselves with lower profit targets might not be a correct proposition as it defeats the very purpose of a mechanical trading system. I am sorry I am not fully aware how the system is designed. But in case the system is designed based on this, I suggest to have a relook.

If I understood the system or your post on wrong side, excuse me and ignore this. And if my understanding is right, kindly consider this.

Thanks and Happy Experimenting

AJAYKUMAR

Yes the real challenge is trading ranging mkts when trend following systems (& indicators) do not work.

Regards,
Kalyan.

Hello Karthik & friends

I agree on that, but what stop would be too tight, is 5% stop too tight? I think yes. In that case do we consider smaller position size with a bigger SL? Maybe that will also not help, since we are trading the same market, the diversification is not really a diversification,

I have also noticed that the CO point moves away from the price (rapidly nearest to the point of entry). Logically (traditional wisdom) the stop should not move down, once it is set.

You would also notice that the CO point seems to be defining the risk. As we initiate the trade, the risk increases, then it becomes flat and then eventually starts to reduce. The question is how to quantify it and use it, maybe it cannot be used, just asking.

In this case, can we consider the CO as just a filter and crossing of the peak as the trigger? If yes, then the SL would be the peak - noise.

The problem is not the lot size but the total equity (risk capital). Typically a trader is undercapitalized to trade in futures. Average position size is 3.5 L, so to implement the system with proper MM rules one needs an investment of 15L (so that we can use 5% stop on position size, equating to 1% of equity)

Confusion is not neccessarly always bad, at times it is good to be confused. Specially when you have friends to clear the confusion and eventually understanding would emerge out of it.

Or may be further confusion


Warm Regards
Sanjay

I had bought few shares of Engineers India Ltd before May 2006. Can anyone forecast its future movements as per technical charts
Thanks

Hi Karthik,

Since U have knowledge of vast number of indicators, was wondering if u have ever studied KST. Was wondering how useful it is.

Rgds

Rahul