Discussion of article "Applying OLAP in trading (part 1): Online analysis of multidimensional data" - page 3

New comment
 
Aleksandr Masterskikh:

I just wrote that recently articles not about trading, but about programming of secondary things (services for processing ready-made trading results and so on), not related to market dynamics analysis, have become prevalent.

Of course, you need that too, but the main task is trading itself, finding algorithms for entering the market (and programming these algorithms).

Just write it. What's getting in the way? Or is it more comfortable to judge?
 
Artyom Trishkin:
Just write it down. What's in the way? Or is it more comfortable to judge?

I do write - you can look at my articles. But I don't see your articles (about trading itself, not about processing secondary data in the form of libraries).

 
Aleksandr Masterskikh:

I do write - you can look at my articles. But I don't see your articles (about trading itself, not about processing secondary data in the form of libraries).

Year 15 and 17. Two articles. And you resent that not enough people write about trading. I'm telling you to fill the gap, if there is demand and desire.

What's stopping you from doing it? That's the question.

 

For apologists of trading in its purest form, I will repeat once again - write specifically what is missing. Abstract arguments are not accepted. Before suggesting a topic, please make sure that there are no articles on it yet. Maybe "trading" means information on strategies, indicators, money management, processing of trading transactions, grids, multicurrency analysis, testing and optimisation, integration with external analytical software and so on? So all this has been discussed many times. I write on topics for which there are no materials available, and I have personally made such tools for myself. In particular, OLAP as a means of analysis of trading system indicators in different sections supplements a trade or optimisation report with information that is obviously missing there. In a good way, all this should be inbuilt. All this has a direct relation to trading. If someone does not agree with it, it is his/her problem. Write in the publications thread, not in the discussion of this article.

 
Good article. What is missing is an assessment of the impact of EA parameters on the results in the case of more than 3 parameters. Or optimal combinations of parameters. Multidimensionality is very far from being understood. 2 parameters for entry or exit usually do not give results, 3 is already difficult to estimate, and 4-dimensional saddle is difficult at all. Tunable optimisation is a good thing. And it is closer to trading))))
[Deleted]  

nothing is missing, i.e. it is not clear at all what the article is about. Here, one should have a certain understanding of the laws of emptiness that, having given up abstract reasoning, would formulate his concrete claims to emptiness.

topics directly related to TC are interesting, research. Personally.

 
The discussion doesn't match the technical resource. The article is excellent!
 
Artyom Trishkin:

Year 15 and 17. Two articles. And you complain that not enough people write about trading. I say fill the gap, if there is demand and desire.

What's stopping you from doing it? That's the question.

Yes, I have two articles about trading.

By the way, according to English-speaking readers, my article "How to reduce risks..." is among the top ten (at least 60 thousand readers in several languages is not a bad result).

I mean that it is better to write 2 articles that will help many people in developing a trading system, than 100 articles about libraries that give almost nothing for algorithmic trading.

Market dynamics is extremely complex (it is a non-stationary process), that's why I am surprised by programmes, in which 1 line is market analysis and 1000 lines are code of dubious services.

I am sure that the goal of the resource ( www.mql5.com, which is definitely the #1 resource in the industry) is to popularise algorithmic trading, not programming for programming's sake.

 

I'm attaching an example of simple wrapper for OLAP classes. The wrapper can be embedded into your EA to analyse instantly trading history at the end of a single pass in the tester.

To choose required analytical sections (selectors) and type of aggregation, the wrapper can be used in EA's OnDeinit, something like this:

void OnDeinit(const int)
{
  OLAPStats stats(SELECTOR_SYMBOL, FIELD_NONE, SELECTOR_PROFITABLE); // choose selectors and fields as required
  stats.setAggregator(AGGREGATOR_COUNT); // choose aggregator, could be e.g.: stats.setAggregator(AGGREGATOR_PROFITFACTOR, FIELD_PROFIT_POINTS);
  stats.setSorting(SORT_BY_VALUE_DESCENDING); // optionally choose sorting order
  // MyOLAPStats callback;      // optional custom implementation of 'display'
  stats.process(/*&callback*/);
}

OLAP is usefull for splitting data by some attributes, which are not provided by standard tester report (for example, profits by symbols, duration, etc).

All dependencies (required header files) can be found in the article. Slightly updated OLAPcube.mqh and Converter.mqh are attached as well.

Files:
OLAPstat.mqh  12 kb
OLAPcube.mqh  42 kb
Converter.mqh  1 kb
123
New comment