From theory to practice - page 7
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
I will double-check it myself tomorrow. Once again, please note that I am looking at ONLY non-parametric statistics. I don't care about variance in its classical sense. Whether the non-parametric coefficient of scale s changes for the t2-distribution Student's is the question.
The increment(first difference) across tests will be stationary. Things will change slightly, at different sites, as the window shifts.
Put in a tsv or txt file - date, time, cotier, cotier increments, ts signals.
The increment (first difference) on tests will be stationary. Everything will change slightly, in different sections, as the window shifts.
Put in csv or txt file - date, time, cotier, cotier increment, cc signals.
Not a single mention here of the two sources of tick archives I discovered in November, http://truefx.com/?page=downloads and http://advancetools.net/index.php/instrumenty/tikovye-ob-emy/istoriya-tikov.
The first one (it's not a DC, and I haven't figured out how to read its data yet) with month-by-month (the unpacking used by Alexander_K) archives of 15 pairs from 05.2009 onwards, the registration is unobtrusive. The format is .csv.
The second gives archives of 24 pairs from 05.2014 in chunks about a year long, received from three known DCs, it is possible without registration. The format is .tks, they enclose a script to convert to .csv ticks, minutes and something else.
In both cases it is possible to download the archives for free.
Saving of computing power is VERY important to me - I've already tried to run TC with 18 pairs, for each of which the algorithm was calculated separately for Bid and separately for Ask. CPU load is 100%. And I need to work with 36 pairs at the same time. If working with the average between Bid and Ask does not violate statistical regularities, then it is a definite saving and I'm very glad about it.
Run your methods on samples from 23:59 to 01:00 server time and you will be unpleasantly surprised, that stats for Bid and Ask are very different.
Run your methods on samples from 23:59 to 01:00 server time and you will be unpleasantly surprised that stats for Bid and Ask are very different.
The increment (first difference) by tests will be stationary. Everything will change slightly, at different sections, as the window shifts.
Here's what I was thinking, my friends. If I have to try my best to prove my every move to people like dear SanSanych, you will get bored, and I won't have time to tell you everything I want to.
So let's do this - whatVizard_ has writtenis simply obvious to me.
Let's leave it as a working hypothesis, the proof of which will be done by the youth :)
For each currency pair we just have to determine TABLE scale coefficients s of probability density function and use them when calculating weights w of price values for WMA moving average.
Once again - the scale factor s is, in general, NOT equal to standard deviation
How is it calculated? - You may ask. Well, that's the secret :)))
Here's what I was thinking, my friends. If I have to try my best to prove my every move to people like dear SanSanych, you will get bored, and I won't have time to tell you everything I want to.
So let's do this - whatVizard_ has writtenis simply obvious to me.
Let's leave it as a working hypothesis, the proof of which will be done by the youth :)
For each currency pair we just have to determine TABLE scale coefficients s of probability density function and use them when calculating the weights w of price values for the WMA moving average.
Once again - the scale factor s is, in general, NOT equal to standard deviation
How is it calculated? - You may ask. Well, that's the secret:)))
Yes, Yuri - somewhere in my heart, I understand that we should work separately with Bid and separately with Ask. Working about the average isn't quite right and it makes me extremely frustrated. With my computer I have to forget about working on 36 pairs at the same time...
If you are going to work with ticks, there is no point in working with 36 pairs. Trading costs are only acceptable on a few instruments.
On the rest, no mathematics in principle will give you a return exceeding trading costs (spread+s+swap+commission+slippage_losses+transaction_holdings+broker_remuneration+resource_payment+....).
That is, I recommend to work only with instruments, there may be only 4 or 5 of them, which have minimal relative costs.
Accordingly, averaging Bid with Ask, can only increase costs.
I love secrets, they remind me of a woman who is not naked)
I will double-check it myself tomorrow. Once again, please note that I am looking at ONLY non-parametric statistics. I don't care about variance in its classical sense. Whether the non-parametric coefficient of scale s changes for the t2-distribution Student's is the question. San Sanych, I kind of know a little bit about physics and mathematics - maybe you will have a more constructive dialogue?
My dialogue is very constructive, it's just that you don't understand the meaning of my words, you are not at all familiar with the literature and the results in the field of financial rows.
Nevertheless, success.
You are not the first inventor of the bicycle here - it is a very interesting and exciting hobby
My dialogue is very constructive, it's just that you don't understand the meaning of my words, you are not at all familiar with the literature and the results in the field of financial rows.
Nevertheless, success.
You are not the first inventor of the bicycle here - it is a very interesting and exciting hobby
I am familiar with the more serious literature.
Should I also attach my educational credentials here to give more credibility to what I'm writing about?