Discussion of article "The algorithm of ticks’ generation within the strategy tester of the MetaTrader 5 terminal" - page 5
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
....
Dear Prival,
Unfortunately, you perceive everything only from your point of view. Statements about "why we are not given raw ticks, average temperature" clearly indicate this. You forget that the broker is responsible for the prices he gives. And he cannot give out all kinds of rubbish (and in reality there is such blatant rubbish) and then be responsible for it.
You persist in beating against reality without noticing the technicalities and possibilities. Reality is in experts' robustness (loading, filters, etc.), not in perfect modelling or tick representation.
Also, you have a strangely low level of criticality even with the results you get. You have twice got completely wrong results with ticks, never self-checked, but went straight to demanding something. I suspect it is exactly the same pattern in the other tic scores.
It's very interesting to read the comments on this article, and in general - interest in the site is increasing every day. But I think there is a struggle with windmills. If real ticks are needed and there are suppliers of them, it is enough to insert several lines into the Expert Advisor code that will replace the tester's ticks with real ones. Or am I wrong?
It is advisable to read the discussions of 2006 - all this has already been discussed in MetaTrader 4 discussions:
- Recommendations for not using the MetaTrader 4 Strategy Tester
- Standard misconceptions in attempts to trade in noise (was "Nightmare on MT4 Street")
- Comparison of real tick flow and generated by the strategy tester in "all ticks" mode
In MetaTrader 5, the tester has become much better, both due to new algorithms and the basic one-minute history.Dear Prival,
Unfortunately, you perceive everything only from your point of view. ....
. I suspect that in the other evaluations of ticks exactly the same picture.
Everyone has their own point of view. It's bad enough when a person doesn't have a point of view at all. I try to show the point of view from the trader's side, as I see it, you see this process from the terminal creator's point of view, if a representative of a brokerage centre appears here, he will have his own point of view. This is normal. We just need to decide whose vision is the main thing, for whom the terminal is created? For the developer? I think not. For a brokerage centre? Yes, partly.
But I think that for the trader, we are the end users of this product. If we are not there, no development will be needed. That's why I think that our opinion should be treated very carefully.
As for my conclusions, you have an opportunity to refute me.
Show me that the generally accepted indicator of the quality of modelling of the minimum RMS of modelling error, here is the formula.
You have 0.
Here x is time, y is price. You can generalise this indicator to n-dimensional space (multi-currency).
Post the data on which it will be possible to double-check your assertions. Although personally I already understand from the article you have not managed to achieve this.
I argue that:
- thetiming of modelled ticks within a minute is irrelevant
- theerrors are minimal - the article shows this clearly.
And you are playing with idealisation of ticks. Although you are talking to a person who has personally written adaptive filters for MetaTrader, which work on hundreds of brokers. And these filters automatically (there are no external settings) change their parameters for tens and hundreds of symbols every day so that few people can predict all the parameters for each symbol.Besides, there are manual settings of filtering in the server itself, there are a lot of datafeeds written by brokers themselves, there is hot swapping of feeds - all this completely kills the very idea of building trading strategies on the analysis of micro peculiarities of ticks interaction.
Our vision unites the points of view of: traders, brokers, developers and technical capabilities. It is impossible to create a balanced information and trading platform without taking them into account. And we are doing it for the fifth time already.
You have 0.
Here x is time, y is price. We can generalise this indicator to n-dimensional space (multicurrency).
Post the data on which it will be possible to double-check your assertions. Although personally I already realise from the article that you have not managed to achieve this.
How are you going to measure time and price with the same yardstick? I can still allow you to measure the RMS of a tick sequence, or to measure the difference between the historical andmodelled tick (and, accordingly, the RMS and other derived parameters from this difference), but how can this difference be a measure of quality?
Try using the same formula to compare tick sequences from different brokers, or, more precisely, try to guess by the values of these parameters where the broker's ticks are in front of you and where the tester's ticks are.
I argue that:
- thetiming of modelled ticks within a minute is irrelevant
- theerrors are minimal - the article shows this clearly.
And you are playing with idealisation of ticks. Although you are talking to a person who has personally written adaptive filters for MetaTrader, which work on hundreds of brokers. And these filters automatically (no external settings) change their parameters for tens and hundreds of symbols every day so that few people can predict all the parameters for each symbol.What's that got to do with it. You can write a million more filters, I will be only glad for you. You either do not understand what I am asking (asking) you, or pretend that you do not understand.
I will try again.
Why when a trader sits in front of the terminal, working, he receives ticks to enter. Everything is normal. Everything is fine. But as soon as he wants to get the history. He is given not ticks, but OHLC (minutes). And from that OHLC for testing, again we try to simulate ticks.
"We create our own difficulties and deal with them successfully."
Renate, why do you create your own difficulty, feed the story as ticks and don't do modelling at all.
Submit, PLEASE history in the form of ticks, not the average temperature of the hospital in the form of OHLC.
How are you going to measure time and price with the same yardstick? I can still allow you to measure the RMS of a tick sequence, or to measure the difference between the historical and modelled tick (and, accordingly, the RMS and other derived parameters from this difference), but how can this difference be a measure of quality?
It is this difference that is the measure of quality. Let me try to explain it with the help of a figure.
There is a true value, in our case a tick. It has a price ( Үist axis) and a certain position on the time axis ( Үist axis). In the figure, this is the red dot. Suppose there are two algorithms that model this point. We want to compare these two algorithms and answer which one models it better.
The measure of quality is the distance to the red dot. It's like a shooter, who shoots better, the one who hits the 10 or the milk. It's as simple as Pythagoras' theorem. The sum of the squares of the cathetes equals the square of the hypotenuse.
The one with the smaller hypotenuse is the better modeller. Perfect modelling is when the distance is zero. We never miss, we hit exactly ten.
H.Y. Why model ticks with OHLC, when we can hit the ten at once?
just the frequency of ticks is a little bit higher, even during rush hours it is much higher.
и наверно чтобы сразу ещё стакан при тестерирвании был =)
After all, the minute history is generated so evenly, so there is a difference of 5 pips at five digits.....
Yes, in the early days people complained that the spread was 4 points.
and now half a point during testing is already hard?
trailing will not beat it and it's even =)
Moreover, please note that this is a simulation.
and has nothing to do with real trading, it's just a test.