New trends in technical analysis. - page 12

New comment
 
IgorM:

What do you even see? The fact that a reduction in the number of trades leads to stability of the TS on the history is clear.

If I've made mistakes in my codes a couple of times, as a result of which EAs traded only in one direction (only Buy/Sell), then the profitability increased.

In that way one can trade with two EAs, one with an error in buy direction only, and the other with an error in sell direction only. And it will be a grail in total:)))
 
khorosh:
So you can trade with two EAs, one with a buy only error, the other with a sell only error. And it will be a grail in total:)))
Several times the situation has repeated itself when correction of a silly mistake in the code led to a catastrophic fall of profitability of the "promising" Expert Advisor.
But I could not learn to deliberately make such errors :))
 
khorosh:
So you can trade with two EAs, one with a buy only error, the other with a sell only error. And it will be a grail in total:)))
The idea is noteworthy, i.e. for Sell positions there are some conditions for opening/closing, and for Buy - others
 
serler2:
......................

Optimisation.

The input is signal results - the output is a histogram with frequencies. Horizontal axis is frequencies, vertical axis - results of trades. Frequencies can be calculated using different parameters.

.....................................

Did you mean to say " .... ... with different methodologies"?

Or is the methodology the same, but different parameters, i.e. input vectors?

..............

And another question concerning the histogram:

It turns out that on the period studied to get the AFC (3 months) the TC showed a highly profitable result?

How many trades were there in three months?

And could you attach a picture of the activity spectrum by number of trades (like DC2008)...

 
ZZZEROXXX:
I also don't quite understand what the frequency of 257 is in the case of crossover. Also, what is the difference between filters - filter1, filter2...

Frequency is a kind of price fluctuation in the market. The price swing can be up or down. (hence filter 1, filter 2) Every tick the frequency can change.

My market is divided into 512 frequencies. i.e. the minimum frequency of oscillation is 1 and the maximum 512.

Frequency 257, from the histogram above, is the frequency at which most of the strategy's trades are losing.

IgorM:

What can you see? The fact that the reduction of the number of deals leads to the stability of TS on history is clear.

ZS: I made mistakes in my codes a couple of times, as a result of which EA traded only in one direction (Buy/Sell only) the same profitability increased.

Of course, I do my tests on history. The frequencies are selected on one period, while trading at given frequencies is performed on another timeframe. There is no result fitting. The results show that the number of losing trades has decreased! (I.e. out of 98 trades, the system left only 16 profitable ones)

During the test, the trading is both buy and sell. There are periods when the EA only starts to buy, this is due to the fact that there are no necessary frequencies for sell in the market.

lasso:

Do you mean ". .. ... with different methods"?

Or the methodology is the same but different parameters, i.e. input vectors?

And another question about the histogram:

Does it turn out that the TS shows the strongly profitable result at the studied period (3 months)?

How many deals were there in three months?

And could you attach the activity spectrum by number of deals (like in DC2008)...

All the same, different frequency calculation vectors. As I wrote above, in one case we are looking for frequency fluctuations upwards, in the other case we are looking for frequency fluctuations downwards. In the third case we have both of them.

With the same strategy I took a longer period for the test. I'll post it now.

 

Here is another result of optimising the same strategy. As Mathemat recommended. I took a longer period. (I should say right away: it does not make sense to take a year or two or 5 for optimization. The frequencies change from time to time. Profitable ones become profitable, profitable ones - unprofitable.)

This time.

Optimization was performed for the period: November 1, 2010 - March 1, 2011. (4 months) Select frequencies.

Test with application of filtering: 1 March 2011 - 5 June 2011. (3 months) Test selected frequencies on new data.

Test without filtering

With filtering

Out of 265 trades 147 remained. The percentage value of profitable deals has risen from 24 to 35. (It is not as high as in the previous test), but the result is higher. (probably due to the long period of testing)

AFC spectrum (up to 74 frequencies) 512 frequencies do not fit into the screen horizontally =) . Red are highs, green are lows. Blue - activity.


In the staple are full stats with transactions.

Files:
qcfqlpyozmi3.zip  44 kb
 

Or you can take a publicly available indicator without using mysterious quantum frequencies and obtain this balance curve for the period 08.08.08 to today.


 
khorosh:

Or you can take a publicly available indicator without using mysterious quantum frequencies and obtain this balance curve for the period 08.08.08 to today.


With a little more effort, you will have a RIGHT balance curve
 
serler2: Frequency is a kind of price fluctuation in the market. The price fluctuation can be either up or down. (hence filter 1, filter 2)
So frequency is simply the difference in Close on neighbouring bars?
 
Mathemat:
I.e. the frequency is simply the difference of Close on neighbouring bars?

It is the amplitude of price fluctuations, broken down on a scale from 1 to 512. The Ask price is taken into account (the calculation takes 7 lines of code) is more complicated than Close[i] - Close[i+1]
1...56789101112131415161718
New comment