Market theory - page 133

New comment
 
Sergey Petruk:
here , Otkritie Broker

The state of the RTS Index is apparently about to be hit by the Price of the Market and the market is about to go upwards:

The market is competitive with two breakeven levels: Price1 = 91870 and Price2 = 107361.07:


 
Алексей:

So you are saying that you were able to identify a strong pattern leading to profitable trading? Generally speaking, if you do a frequency analysis of the direction of movement a day ahead, it turns out that, at least, this parameter is independent of the nearest past prices, there will always be about 50% of guesses. What percentage of profitable trades were you making in 2010-2011?

406 profitable trades out of 501 = 81.03%.
 
Awl Writer:

An analogue EQ is a good example. But a live person generally won't hear a difference if the bass is 10 milliseconds behind.

10 milliseconds? Is it a lag? )) It can be ignored without a second thought. Then a reasonable question arises - why is there such a tremendous delay on the same mouwing? Or it is calculated using another principle?

Awl Writer:

Analogue EQ is a good example. ... Note that here the filter core is 'one-way'.

Why? What do you mean one-way? What's stopping you from taking a filter with a buffer and running it through?

Awl Writer:
you can put an SMA(20) on any chart with a shift of -10, there's your "unstretched" filter.
Yes, it's more or less true, but the shift is very large. We need to get rid of it.
 
Yousufkhodja Sultonov:
406 profitable out of 501 trades = 81.03%.

Cool. Such a percentage on a sample size of 501 trade opportunities is quite difficult to match.

Nobel to Yusuf, definitely.

 

Daniil Stolnikov: 

...
Let's take the same picture , let Vin be the raw data, Vout - after filtering. For sound there is no difference. For price data, there is a big difference. // In digital equalizers, sound is run through a buffer. // Imagine an SMA core, a rectangle. Logically, an average value of several consecutive numbers in a row should be located in the middle of this interval, it belongs there, not at the end. I.e. SMA(20) with a shift of -10 is correct, while a simple SMA(20) is wrong and useless. In fact, SMA is a bad and primitive method of analysis, and it hasn't been used in statistics for a long time, but never mind. If you have created a working TS on moving averages, then you are lucky. This discussion needs to be moved to another thread
 
Awl Writer:
Take the same picture, let Vin be the raw data, Vout - after filtering. For sound, there is no difference. For the price data there is a big difference.
What's the difference? I don't understand - both are oscillating. The only thing is that the sound oscillates around zero, the price is always above zero. But I'm sure it can all be converted to the right form.
 
Daniil Stolnikov:
What's the difference? I don't understand - they both oscillate. The only thing is that the sound oscillates around zero, the price is always above zero. But I'm sure it can all be converted to the right kind of sound.
The difference is that you just listen to the music after filtering. And the price after filtering you must somehow use it to make a profit. And this is another task and it has nothing to do with filtering.
 
Daniil Stolnikov:
What's the difference? I don't understand - they both oscillate. The only thing is that the sound oscillates around zero, the price is always above zero. But I'm sure it can all be converted to the right form.
A sound wave is a sine wave. Cotiers are not stationary and subtracting the trend doesn't really change the picture,
although econometricians will say that the increments are stationary ))))
And MA2 (simple MA2) for example is calculated as kloz+lag1cloz/2... ...i.e. the past values are used.
You can try to remove the lag by prediction... with a neural network or the like... but it still won't work properly...
 
Vizard_:
A sound wave is a sine wave. Cotiers are not stationary and subtracting the trend does not really change the picture,
Although econometricians will say that increments are stationary ))))
MA2 (simple MA2) is calculated as kloz+lag1cloz/2... ...i.e. the past values are used.
You can try to remove the lag by prediction... with a neural network or the like... but it still won't work properly...
Yes, sine, if the input of the audio device is a sine wave. But what about music or speech? As with price, it's a set of harmonics with different amplitudes, frequencies and phases.
 
Vizard_:
Yes, it's more complicated than that. But first we need to find out what will bring us a profit)))) We could play around with Fourier or something. We could do a lot of math and build it up without the HF and so on.
I have nothing at hand, well, for example from the player music. You'll get a yellow line, though. If the voice, you can recognize it, if we know it,

that a short after the letter A would bring in a profit...


Did you duplicate Yusuf's calculations there, does it really work out the way Yusuf showed in the graphs? I still have doubts about the correctness of the profit calculation, the graph of increasing profits is too monotonous with daily fixing, there are no significant downward deviations. I cannot believe it, it seems to me something is wrong in the calculations. If there are no errors, then it really is a grail.
1...126127128129130131132133134135136137138139140...288
New comment
Reason: