Gathering a team to develop an IO (decision tree/forest) in relation to trend strategies - page 19
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
Been working on predictors lately, writing functions and tweaking/refining things.
At the moment in Deductor Studio program I get the following result on one decision tree, almost out of the box, after training in 2015-2017 on the results of 2018
Been working on predictors lately, writing functions and tweaking/refining things.
At the moment in Deductor Studio I get the following result on one decision tree, almost out of the box, after training in 2015-2017 on the results of 2018
And may I ask why you are so confident that this is a promising direction ? You do realise that 100% per month on a test is firstly very low. And secondly the test correlates differently with the real game. We often see that it works in the test simply because we adapt it to the chart, but in real trade we will be the sinker. In my experience, the test shows tens of times better results than the real game, but it depends on the algorithm. Maybe you should try the strategy on demo? If the demo showed at least 10-50% at acceptable drawdowns and margin level, you'd have much more interested in your community. I don't think you should present the test as a confirmation of the strategy performance at all.
May I ask why you are so confident that this is a promising direction? You do realise that 100% per month on the test is, firstly, very low. And secondly, the test correlates differently with the real game. We often see that it works in the test simply because we adapt it to the chart, but in real trade we will be the sinker. In my experience, the test shows tens of times better results than the real game, but it depends on the algorithm. Maybe you should try the strategy on demo? If the demo showed at least 10-50% at acceptable drawdowns and margin level, you'd have much more interested in your community. In my opinion, you shouldn't present the test as confirmation of the strategy working capacity.
Why it is promising, because it allows us to quickly look for patterns, to find complex correlations. Identification of statistical regularities (frequently recurring events or predictor values) is followed by the process of studying these regularities, including on the chart in different situations. In the standard ATC approach, we only look for patterns by eye, so we can look at the obvious. And to better find the right (useful) patterns, we need to develop machine learning methods as applied to the specific subject area, which is what it is proposed to do.
You are not quite correct in your assessment of the test results provided - this is an area outside of learning, i.e. the area where the patterns identified in the past period worked.
Right now my attention is attracted to the MOEX futures market, and as you know there are no demo accounts with real quotes there, so you can't make a demo account either. However, after the end of the cycle I plan to run a real account with minimal deposit.
Well, about expectations, I consider very good result of 5-10% growth per month with the same drawdown or a little bit more.
Alexei, hire a designer, it's very bad.
Let's discuss my avatar.
What's wrong with it?
At the moment my attention is attracted to the MOEX futures market, and as you know there are no demo accounts with real quotes there, so you can't make a demo account either. However, I plan to launch a real account with a minimum deposit after the end of the cycle.
There are.
There is.
All I've seen are distorted quotes, with rubbish.
Including from the broker Otkritie.All I've seen are distorted quotes, with rubbish.
Also from broker Otkrytie.Have you tried J2T? As far as I remember (I was comparing their stack with Reveal's stack), the trades were the same.
And the demo from MQ does not suit you?
Have you tried J2T? As far as I remember (I compared their vintage with Reveal's vintage) - the trades were the same.
I have not tried it. Has it really broadcast without delay and without distortion of quotes for demo accounts?
If so, I will think about opening a demo account there.
Have you tried J2T? As far as I remember (I compared their stack with Reveal's stack) - the trades were the same.
And the demo from MQ does not suit you?
I don't have any problems with MQ's demo account, it just does not work correctly comparing with Otkritie's one.
I'll be honest with you, I have not done any research on it myself, I use information provided by various sources I trust.