New trends in technical analysis. - page 11
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I have doubts about the adequacy of such a comparison,
but it will still be very interesting to see what happens.
So, sorry for the delay. This weekend I put the optimisation in place. I think I will post the results here next week.
P.s. I rewrote half of the Expert Advisor's code. I have rewritten half of it. + I have also added TP and SL.
OK, sorry for the delay. This weekend I put in some optimisations. I think I will post the results here next week.
P.s. I rewrote half of the EA code. I have rewritten half of it. + I have added TP and SL.
Yes, it's like with designers: if you want to call it "ear", you call it "ear"; if you want to call it "cheek", you call it "cheek".
I remember I've seen a detail in drawings called "Point" ..... And nothing can be done about it, even the chief designer has no right to forbid the creator to name his creation.
Something makes me question the adequacy of such a comparison,
But it will still be very interesting to see what happens.
I'm posting the research.
The approach with "quantum frequencies" is as follows:
We take any TS, which gives the maximum number of signals to enter the market. Then we determine the frequencies of profitable and losing trades using the history (retrospectively) as a result of optimization. Then conditions (filtration) are added to the Expert Advisor: at what frequencies it will need to trade in the future and at what frequencies it will not.
As an example, we took the Expert Advisor from page 8 of this forum. I added TP and SL to the Expert Advisor + optimized the code. Entry into the market is crossing of EMA5 and EMA50. Exit when crossing is opposite or TP = 800 pips (in 5th sign) SL = 400 pips (in 5th sign). Always trade 0.1 lot on M15.
Optimization was made on period 01 Feb 2011 - 31 Apr 2011.
Test with filtering - May 1, 2011 - June 2, 2011.
Optimization.
We have signal results on input - histogram with frequencies on output. Horizontal line - frequencies, vertical line - trade results. Frequencies can be calculated based on various parameters.
Below is a filtered histogram by one of parameters. For example, we can see that trades at frequencies 400х, 441х are positive. We need these frequencies. And we should exclude all deals that were performed at frequencies 257.
Testing.
1. test from May 1, 2011 to June 2, 2011. Without filters.
2. the same period. Applying the filter 1.
3. same period of filter 2.
4. Maximum filtering. Filtering by 3 parameters at once.
Result: as a result of filtering, number of deals has decreased by 80% in the first case, by 60% in the second case and by 90% in the third one. % of profitable trades grew by 2 times. It took 2 days to obtain these results.
The initial strategy had an average of 3 trades per day. Which is not enough! That is, there is nothing to filter out! On the good side, there should be a strategy with 20-30 entries per day.
Full tests in a paperclip.
I would group such "quanta" into "point" too.) It's all from people's desire to look like "great gurus", let's rename bars into quarks, mesons, bosons depending on frame. Let everyone be jealous of how cool we are.
Quantum frequencies is the approach the author of DC2008 called it. You can call it scoring, pushing, the green turtle method, or quantum stuff. Whatever you feel more comfortable with.
Quantum frequencies is the approach the author of DC2008 called it. You can call it scoring, pushing, the green turtle method, or quantum stuff. Whichever you prefer.
So you've been asked for 11 pages. How do you calculate these green turtles ? formula ?
All my life the frequency has been determined by the formula F=1/t
Where F is frequency in Hertz.
t is time, measured in seconds.
How to calculate this "quantum frequency". Suppose I have a TS which performs 300 trades per day.
how to calculate ?
Below is a filtered histogram for one of the parameters. You can see, for example, that the trades around the 400x, 441x frequencies are positive. We need these frequencies. And we need to exclude deals with 257 frequencies.
serler2, take at least an order of magnitude more trades, increasing the testing period - say, from one month to two or three years.
These results are worthless as they are statistically unrepresentative. And remove the mismatches in the charts.
I'm not saying that the concept of frequency should be clarified - otherwise we'll still be talking about mystical things, which everyone understands in their own way.
Here are the studies. ................
Result: As a result of filtering, number of deals has decreased by 80% in the first case, by 60% in the second case and by 90% in the third one. % of profitable deals grew by 2 times. It took me 2 days to get these results.
Sergey, thank you very much for the comprehensive report.
Yes, I can hardly compare two TSs directly using different methods of filtering (as I intended originally),
but this by no means diminishes the power of this method. Even the 16 transactions on the OOS show it ;)
The only disappointment is the high resource intensity of the optimization process.
Even the 16 trades on the CB show that ;)
What do you even see? The fact that a reduction in the number of trades leads to the stability of the TS on history is clear
i have made mistakes in my codes a couple of times, as a result of which EAs traded only in one direction (only Buy/Sell), it increased profitability.