Metastock 10 Performance Systems

[sens_c]

Dear Memebers,

I have done some basic analysis and back testing (1500 bars) with the metastock performance systems on NSE historical data. The results are attached as excel sheet.
Sheet 2 contains the list of securities used for back testing (the top 269securities, TTQwise).
Sheet 1 contains the result of the explorations.
1. Hit Ratio - %of scrips showing positive trade efficiency.
2. Scrip & Perf - Top five scrips for the particular system and its trade efficiency. (Trade efficiency is defined as the no. of points gained/lost per day, counting only the days when the system ran a position).

Now, I have some queries regarding the results. I would really be happy if senior members could throw some light on these -

In my results, top five systems show more than 75% hit ratio, meaning they were successful in trading more than 75% scrips out of my list of 269. These systems (Landis reversal, MESA Sine waves, Swing Index, Adaptive moving average and Exponential moving average) are very different in nature except for the two MA systems.
I wish to use the principle of "Confluence of Indicators" in selecting my trades.
That means I wish to carry out these explorations stage wise, i.e. every exploration will use the results of previous exploration.

Now, my doubt is if my method is alright or if I am making some major mistake. As such it looks fine, but "Confluence of absolutely different kinds of indicators" may not be what was implied in the statement.

I understand that this is a silly doubt, but I still have it!

Please help!

 

[Dev Mookerji]

were is the excel sheet?

 

[sens_c]

Sorry Dev,
Please find the attachement now.

 

[minnamor]

I do not see what securities you tried the systems on but I suppose that you tested long only positions by the look of things. Your suggested procedure is sound but you should test the final system (including stop losses, trailing stops, profit targets, filters etc.) with TradeSim if you want to test a portfolio and also because that is the only way of including money management measures (amount of securities, amount of risk per transaction, etc) and to find various statistical measures relating to portfolio performance.

 

[sens_c]

Dear Minnamor,
Thanks for replying. I have tried in tradesim as you suggested.
Actually what i have done is that I have created a composite system (Landis Reversal, MESA Sine Waves, Swing Index, Adaptive Moving Average and Exponential Moving Average).
The composite system will generate a buy signal if all the indicators say so (connected by AND) and will generate a sell signal if any one indicator says so (connected by OR). Other than that I have used the true range formula for stop loss {20DMA-ATR(20)-1}.
Using this system tradesim posts about 13% annual return if I select Limit or Market as Entry Confition. But it reports abject losses if I use "Worst case" slippage, i.e. all entries taken at days high and all exits taken at days low.
Hence is my doubt about the effectiveness of my system.

If you feel interested, then I shall give you all reqyuired codes/ data for setting up and testing the system.
And thanks again for replying!

 

[oxusmorouz]

Using this system tradesim posts about 13% annual return if I select Limit or Market as Entry Confition. But it reports abject losses if I use "Worst case" slippage, i.e. all entries taken at days high and all exits taken at days low.
Hence is my doubt about the effectiveness of my system.

Worst Case Slippage :
Buy signals are executed at the high point of the day and sell signals are executed at the low point of the day. So much of slippage is rare in real trading, which perhaps explains why your system reported a loss.
Please go through the tradesim user manual for more details.

 

[oxusmorouz]

Since you are using tradesim to backtest, make sure you delay entry/exit signals by 1 bar ---> open price, in case you haven't already. Procedure for this is explained in the user manual.

 

[minnamor]

Composite systems are worse than binary wave systems. The appropriate solution would be to test for best long system (or if you prefer the best oscillator system when adx shows weak or no trend) and the best trend following system (in all other situations). Then do the same for shorts rembering that default parameters value might have to be modified from long entry values. Once you have done this, have a system with the best long system(s) (remember to use OR instead of AND for signals) and the best short system(s). This composite system is much better than composite signals. Try it.

 

[sens_c]

Dear Minnamor and OXUS,
Thanks for your valuable inputs. I shall try these out and report my results here. I am sorry for this delayed response, PC Sucked out because of some software without company technical support.

 

[sens_c]

HI Minnamor,

Could you suggest some effort efficient method of selecting a ood long system among the 90 odd which are there in Metastock PS?
Translating all formulas in tradesim, then running each exploration and then running tradesim!.. It's been three days I have started and I am only done with 15.

I have interpreted you method like this -

Begin()
If ADX = Strong
Long = Adaptive MA Long Signal
Short = Adaptive MA Short Signal OR RSI Short Signal
Else
Long = RSI Long Signal
Short = RSI Short Signal OR Adaptive MA Short Signal

End()


I am using "OR" for Sell conditions as the ADX situation may change while I am in the trade, so I must consider both short signals for Trade Exits. Whereas, for Buy signals this is not necessaty as I shall initiate a trade with the Trend Followings system only if ADX is strong, otherwise I shall initiate trades with oscillators.

Is this what you have indicated?

Coding forn this is gonna be quite complicated, I guess.. :-(
Meanwhile, have anyone tried the Trading Solutions (the neural network thing)?