Using artificial intelligence at MTS - page 2
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
we take the value from 4 points, multiply each value by a coefficient, sum up - what is not smoothing by a filter?
Have you heard of linear-weighted moving averages?
Why not? - Alt+F4
You have already invented MACDZeroCrooss that is just MACross. Now you have invented a digital filter;-)
If you add more points to this expression, you can make a profit on any symbol in the tester after optimization.
If you add more points to this expression, you can make a profit on any symbol in the tester after optimization.
By the way, Integer correctly pointed out that proceptron looks like a digital filter. I'm an expert in this field and I know what I'm saying. That is to say, a preceptron is a filtered AC. The meaning of this filtering (or linear combination of speakers) is not clear.
I'm not an expert in neural networks, but it's a natural question. What exactly does this EA have in common with artificial intelligence? Optimization of x1-x4 is done in a very natural, manual way. So, any Expert Advisor can be called artificial intelligence, because any EA needs optimization of input parameters, and the fact that preceptron is calculated as a linear combination of present and past values of AC, is not a determining factor either. If the optimization is performed automatically inside the Expert Advisor, I would agree with the name.
Translated into Russian, any problem the solution of which may be obtained by a complete or simplified search of variants is classified by the bourgeoisie as artificial intelligence. In the USSR such problems were classified as a branch of applied mathematics and a subsection of algorithms for finding optimal solutions.
That is, the preceptron is a filtered speaker. The meaning of this filtration (or linear AC combination) is not clear.
It is clear that everything that passed through the filter is identified as a signal to open a long position. Everything that was filtered out is considered a short position. In the same way already opened positions are filtered for a trend reversal.
Yuri, why get so emotional? The question was almost certainly about the hidden meaning of this filtering, not the interpretation of the result... Roughly speaking: why this particular filtering? When applied to a trading system, it may not be the most adequate question, but you can try to reason why AC and not some MACD...
AC has been implemented due to these considerations only as it has no external settings other than symbols and timeframe.
So do not look for a hidden meaning : it was not there at all since everything was done following the principle the simpler, the faster optimization would be. This is just an example of a primitive neural network using the built-in MT4 genetic optimizer of strategies on historical data instead of the standard training algorithm. Nothing more, nothing less.
Incidentally, takeprofit of the strategy is missing due to the same reason - it is an additional setting parameter. Although, if it is implemented, it is quite possible that trading will become either more profitable or more stable?
It makes more sense now. Whatever you call it - perceptron, gene, filter or devil with five horns - the essence of the advisor does not change. Your clever words have misled people and until you explain what perceptron is, there's no hope for understanding and constructive discussion... In general, the idea is really very interesting: all optimizable parameters are packed into one simple function.
But the best neuronet is the one in one's head - especially when one only needs to recognize a couple of situations, e.g. buy/sell. But human brain hardly recognizes market entry/exit signals, so it is hardly of any help. Not to mention the XOR error inherent to the perceptron...
And the thing is that there are no patterns in Forex. There is a well-defined pattern - movement in the support and resistance channels and breakdowns of these equations. Everything else is a random movement.
Training of neural networks is a classic tweaking. And if there really are some regularities, ideally neuronet will catch them. But so far there are no results. Because there are no regularities at all, except for the model described above. Rosenblat invented the perceptron back in the 60's and tried to use it in the market then.
Of course my words don't mean that Reshetov should stop all of a sudden. You just need to work in a more relaxed way, without bombast and resentment.
It makes more sense now. Whatever you call it - perceptron, gene, filter or devil with five horns - the essence of the advisor does not change. Your clever words have misled people and until you explain what perceptron is, there's no hope for understanding and constructive discussion... In general, the idea is really very interesting: all optimizable parameters are packed into one simple function.