Discussion of article "A scientific approach to the development of trading algorithms" - page 8
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
Has the new article been published yet in English?
Hello, thank you very much for the article you wrote, it gave me a better understanding of the market. But the attachment doesn't seem to work properly because of the missing indicator file?
If you don't use the function responsible for interacting with the indicator, the bot will work fine without the indicator. I didn't attach the indicator to the article because it's not distributed for free
But I think you tested the wrong hypothesis with the test EA.
Also not the same logic as in the basic study.
You based your price distributions on what it will be in 40 blocks. So why would you close a position at the first block in the opposite direction? Follow the hypothesis and close after 40 blocks. Then the financial result of 100000 trades will be like in the diagrams. I.e. the trend will be stronger than at random, but its direction is impossible to choose.
Interesting article and method. showing that there is trendiness in the market. That the price is more likely to go to the area of 10-20 blocks than at random (in both directions).
But I think you tested the wrong hypothesis with the test EA. You built price distributions based on what it will be in 40 blocks, regardless of the direction of the previous block. If you were building depending on the previous block, you could choose the direction according to this rule. I assume that the charts will be without significant differences, i.e. the direction of the previous block has very little influence on where the price will be in 40 blocks. With 300 pt blocks, 10 blocks = 3000pts, which is a lot of days.
Not the same logic as the basic study either.
You were basing your price allocations on what it will be in 40 blocks. So why would you close a position at the first block of the opposite direction? Follow the hypothesis and close after 40 blocks. Then the financial result of 100000 trades will be like in the diagrams. I.e. the trend will be stronger than at random, but its direction is impossible to choose.
In this case it is not important. There are 2 types of strategy: trend and flat. If the probability of trend continuation is 0.5, both will not be profitable. If the probability of trend continuation is >0.5, the trend strategy will be profitable and vice versa. If the probability of trend continuation is lower than 0.5, the flat strategy will be profitable. This EA is made for example, not for trading. The Expert Advisor only shows that this or that strategy is profitable.
But I can say that the dependencies of the direction of the next block on the previous block do exist and these dependencies are much stronger than shown in the article. A lot of time has passed since the article was written, I have changed the algorithm of building blocks, I have switched to measuring the probability of continuation of individual blocks, I have taken into account slippages of closing blocks, I have started to analyse the probabilities of falling/growing movements separately, I have developed an algorithm of multiscale analysis, I have found the signs of market switching from trend to flat mode and I have found how to track this switching. In this article only the base
But I can say that dependencies of the direction of the next block on the previous block do exist and these dependencies are much stronger than shown in the article.
Yes, this is more useful than what the price will be in 40 blocks. That's too long to wait... especially if the blocks are 200-300 pts
A lot of time has passed since the article was written, I have changed the algorithm of block building, changed to measuring the probability of continuation of separate blocks, took into account slippages of block closing, started to analyse probabilities on falling/growing movements separately, developed an algorithm of multiscale analysis, found the signs of market switching from trend to flat mode and found how to track this switching. This article contains only the base
Impressive... Where can I read about the new blocks?
On the blocks: very similar to the grid. What is the difference between the new blocks? I would get away from slippage bars on ticks. Maybe just run the decisive part of the EA on ticks when crossing pre-calculated grid levels?
Yes, it's more useful than what the price will be in 40 blocks. It's too long to wait... especially if the blocks are 200-300 pts.
Impressive... where can i read about the new one?
On the blocks: very similar to the grid. what is the difference between the new blocks? I would get away from slippage bars on ticks. Maybe just run the decisive part of the EA on ticks when crossing pre-calculated grid levels?
It is not a grid, the size of blocks is dynamic and changes when the price changes. It turns out that the closing price of each next block rarely coincides with the prices of the previous blocks.
Yes, already the approach has changed a lot in these 3 years. According to the article they were the same.
Thanks for the article, now it is clear that the market is different from SB.
Thanks for the article, now it is definitely clear that the market is different from SB after all.
Thought about the reasons. At blocks of 200-300pts we see the strongest differences from the SB after 10-15 blocks, i.e. after 2000-4000pts of movement in one direction. Such price shifts are most likely formed by news. The price shifts and already fluctuates at a different level.
If we cut out news movements/blocks (or just blocks from 13 to 20 Moscow time, when the main news is released) from those 100000 examples, the probabilities of outcomes of 40 blocks will most likely become more similar to the SB.
Bottom line: one of the main differences from the SB is the influence of news.