Market patterns - page 23
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
And post lots of pictures. I'm intrigued.
Better in a new thread -- if anything it would be easier to tear down :)
There are a lot of pictures, but they don't differ much from those presented above. Name an open source time series, or provide a ready-made series to avoid bugs, I'll retransmit, get equity and post here, preferably something with a trick, because the number of attempts is not infinite because it's still not my set-up))
I can suggest a series to check. For credibility, too, it can be shredded and lagged.
The main problem in this situation is that it is enough to look one step ahead to create a grail, myself many times so cheated, so the cut series if only the cut points do not coincide with the orders, can not show fake.
If i have a good effect on the speed of response of the TS, i.e. if the TS quantifies logic by candlesticks then by candlesticks, if it uses tick data then by ticks.
In general, we can say that if we cut the TP at the point where there is increase in equity and nothing changes on that slice, it is good. But how can we make a cut without running the whole cut first? But all the same, the probability of seeing magic on a specially prepared cut, with a different number of interconnections, is high. If the result is ambiguous, then there is a suspicion of genuine graality. But no one usually sells such systems or tells you about their existence when they have been battle-tested. So I doubt that "someone" has honest motives to give you such data.
Of course! I'd be happy to! Either openly or in person as you like. On the spider they suggested doing a mix of rows, FI and SB, in random order. I'll try it that way some more.
If in doubt, I strongly recommend to ask for investor's password from his real account, where there are at least 500 trades (if pipsarian - 5000). See what's real and what the risks are.
If that's not there, it's not even worth bothering with these "black boxes".
I don't know if there is any interaction with any trading platform and if there is trading on it, I doubt that if the system was tried out on a real account, there would be any motivation left to try to fake it on the side. But I do not intend to cut it so close, because firstly, the man from whom the content comes, is very interesting, and secondly, I have not met so difficult to falsify "grails".
And in general I was interested in the general methodology of such checks with such scarce access to data. And considering the fact that it is unlikely that someone possessing a profitable system will give out demo versions with unlimited lifetime that allows copying the signal to a real account, this is an important question.
If the results on the SB are null and indistinguishable from the SB itself, then there is no peek, as peeking into the future on the SB necessarily draws a grail.
That is why I decided to ask the community for other ideas, but not because of SB, but because the equity curves of different FI normalized by MO differentials are quite similar. As if all FIs have a single very strong inefficiency. Which you'll agree is very strange.
Of course! I'd be happy to! Either openly or in person as you like. On the spider they suggested doing a mix of rows, FI and SB, in random order. I'll try that again.
By the way the option -- half SB, half normal. And watch. But if the real pictures are similar, then the dude should only scratch the SB, not sell it or something like that.
We should figure out how to generate such a SB to mix "seamlessly" with FI and it would be unnoticeable by both cumulative graph and final differences. Especially interesting are smooth mixes between such rows. Well and by tradition requested sequence, this time for full control take some piece and cut by adding 1 step value to understand how the system would behave in real time. It's a hassle, but it's a sure thing.
We need to figure out how to generate such an SB so that it blends "seamlessly" with the FI and it looks inconspicuous
I don't understand how the verification is done now, if you givethe dealer a series of numbers - a quote, he also answers with a series of numbers - equity, or something else. If it's a verification, it's a little... naive.
If you cannot check it on a real (micro) account, try to negotiate a successive verification: you provide a quote[0] to the seller, he replies with action[0] (buy/sell/nothing and its max/minimum volume), you provide another quote[1], he replies with action[1]. Long, tedious, but you can rely on this kind of verification.
So, take initial value and generate it in reverse. And play with amplitude to make it imperceptible.
Logically, I have an approximate idea of how to do it, all that's left is to implement it.
Here's what I'm thinking: A noise is generated, for example with a constant probability density of distribution values vs time, with MO =0. Then some differential (C(t)-C(t-1) or O(t)-C( t) or average) is calculated, then noise is fitted to a series of differentials from FI, by MO and probability density depending on amplitude, which is approximately normal for FI. A mixture is made, sequentially with random lengths of individual sections, and then a cumulative plot is calculated from those increments. Then there will be no "seams" and it will look like one instrument. But there will be patterns in some places and not in others.
I do not understand how the verification is performed: the "grailer" receives a number series -quote, while the "grailer" replies with a number series - equity or some other. If we checked it that way, it would look worse.
If you cannot check it on a real (micro) account, try to negotiate a successive verification: you provide a quote[0] to the seller, he replies with action[0] (buy/sell/nothing and its max/minimum volume), you provide another quote[1], he replies with action[1]. Long, tedious ... But you can rely on this kind of verification.
Yes, spider also advised to ask the current state, in addition to equity. That makes sense, thank you. The next check will be with a sequence of rows that grow by 1 value. It's like realtime, it's troublesome but sure.
Here's what I'm thinking: noise is generated, for example with a constant probability density distribution of values vs time, with MO =0. Then a series with FI is calculated with some differential (C(t)-C(t-1) or O(t)-C(t)) or the average) then the noise is fitted to a series of differentials from FI, by MO and probability density depending on amplitude, it is approximately normal for FI. A mixture is made, sequentially with random lengths of individual sections, and then a cumulative plot from those increments is calculated. Then there will be no "seams" and it will look like one instrument. But there will be patterns in some places and not in others.
This is the first thing that comes to mind when such a task is set.
But I can make a more cumbersome logic. In principle, you can arrange a series in such a way that by putting the main types of inefficiencies sequentially, you can determine the Expert Advisor's logic by the test results. So your 'ne'er-do-well' has to be wary, otherwise you can give away its grail, not just verify it.
Try what you get from such a semi-synthetic range. This is a warm up for now. Get the result and post it here please, if this test is passed there will be another final one.
But it will be a miracle if it does make it through.