Machine learning in trading: theory, models, practice and algo-trading - page 2336
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
If the average position holding time for the TS is 10 minutes. And the current position hangs for 10 hours, then its result is a fluke (completely non-systemic)?
What is the reason for the practice of turning off the TC (if not news) before the news? My answer is to get away from the luck.
If, say, the TS makes most trades in the morning. Then the positions that were taken in the evening are non-systemic. Their result is a chance. Because the TS was not tuned to the patterns in the evening, because the backtests showed morning inefficiencies.
If the average position holding time for the TS is 10 minutes. And the current position has been hanging for 10 hours, then its result is a gambit (completely non-systemic)?
What is a "system result"?
If it is a system result, then any trade it generates will be systemic. And the holding time doesn't matter at all.
Non-systematic in this case, we can only call it a situation where a trade is hung for 10 hours because of force majeure (missed withdrawal).
Moreover, I don't see the difference between testers and real. As we adjusted, so we go. What is the sense of setting up without regard to some individual trades, if then they will be in the real? You won't be able to cut them out there.
The only sensible thing to filter is the random super profits on the studs that would simply not take place in the real account.
There are systems with 95% of trades that are losing or losing-low profitable (break-even or short trailing), while the remaining 5% have huge profits. To cut them out just because they are few, and configure the system to maximize profits on the low profitable 95%? That's just stupid.
What is the reason for the practice of disabling the TS (if it is not news) before the news?
If the news is taken into account when testing, then the decision - to disable or not - is made at the same time. But the answer becomes obvious - you disable it (or not), if it gives more profit.
The same can be said about closing positions before the weekend.
And why is this practiced by those who have not checked the influence of these factors on the history? Herd mentality, I guess.
Like a spread filter that pokes around wherever it's needed and not needed.
If, say, the TC makes most of the transactions in the morning. Then the flying positions, which lasted until the evening, are non-systemic. Their result is a fluke. Because the TS was not tuned to the patterns in the evening, because the backtests showed morning inefficiencies.
If the TS does not imply closing trades before dinner, the positions that have survived until the evening are quite systematic.
Or the TS is not finished, and the morning pattern is not fully exploited.
Moreover, I do not see any difference between the testers and the real. As we set up, so we go. What is the point of setting up without regard to some individual trades, if then they will be in the real? You can't cut them out there.
It is ambiguous.
Googled what they write about "market inefficiencies" - 99% of the information is the usual sophistry, written to order from the copywriters' exchanges
but there is a more or less definite concept of "an efficient market" - in general they give the definition that "the price takes into account everything" and "the market price reflects the real value of the asset.
but if we go in the opposite direction - to add "not" prefix to "efficient market" and get some antonym using the definition "efficient market" then in my opinion we won't be able to get a formalized description of "market inefficiency" - in general it's another riddle with formalization of the problem
I'm not talking about a general concept, but about that specific concept which was the basis of a particular TS. If it is well defined enough to write an Expert Advisor based on it, then it shouldn't be difficult to develop a trade tracking algorithm based on it.
What is the reason for the practice of turning off the TC (if not news) before the news? My answer is to get away from the luck.
If, say, the TS makes most trades in the morning. Then the positions that were taken in the evening are non-systemic. Their result is a chance. Because the TS has not been adjusted for the patterns in the evening, because the backtests have shown morning inefficiencies.
From the practical point of view the filter in the long positions may be the profit growth rate.
Has anyone started reading Prado? He has an interesting topic on how to clean the covariance matrix from noise
We should clean the data from the noise.