A little surprised :) Thought I'd share and ask a NOT rhetorical question. - page 20
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
That is, the original statement "integer mathematics saves/accelerates in financial calculations" has completely crumbled.
With libraries, constant recoding int <-> double and reducing the level of confidence in the resulting code by two orders of magnitude (no one in their right mind will believe in integer version of complex calculations because of potential overflows and loss of accuracy) you will lose to usual real maths on SSE2.
Everything is good with our optimizer and will be even better - it was made for real practical tasks. Proven by practice and supported by our great experience in development.
It was clear from the very beginning that you and Prival are from the same field.
Theorists are visible from afar, especially when they give up their positions with a smile. If one statement doesn't work out, it's easy to replace the next one and so on.
BUT-BUT. ! I didn't surrender anything. :)
If you don't get it - then alas - you have to understand that with such volumes it's no longer realistic to drive. :)
Why are you playing dumb - you should understand ( although ... dunno ... ) that dubs are also computed in integers. :) After bringing in the mantissa. In general - such concept as calculation with REQUIRED accuracy for performance improvement you are not familiar at all. So keep beating yourselves up. :)
Time will tell - will your tester be good or not. IMHO - so just - at the price of opening drive on the clock. :)
:)
...
It was absolutely clear from the start that you and Prival are at odds with each other.
The theorists are visible from afar, especially when they give up their positions with a smile. If one statement doesn't work out, it's easy to replace the next one and so on.
It is not correct to compare them, at least Prival is doing well with theory.
Please don't equate Academic and Prival.
It is not correct to compare them, Prival's theory is fine.
Please do not equate Academic with Prival.
Prival and I don't think so. :)
Prival and I don't think so. :)
BUT-BUT. ! I didn't give up anything. :)
It's the practitioner who doesn't pass, defending what he or she then coaxed out with his or her own.
And you really passed when the tale/theory of acceleration on integer calculations failed. It was enough to get SMA/EMA/LWMA to make it all fall apart.
Don't play dumb - you must understand ( though ... dunno . . . ) that dubs are also computed in integers. :) After the mantissa. In general - such concept as calculation with REQUIRED accuracy for performance improvement you are not familiar at all. So keep beating yourselves up. :)
You are engaging in outright stupidity, trying to attribute my ignorance of real numbers.
Your level of "given accuracy" was perfectly demonstrated by an example of integer muving calculation. Even the level of accuracy double is not enough, although some world programs like MetaStock (at least version 7.xx, I haven't checked for a long time) calculate everything in float for the sake of economy, which leads to serious discrepancies with other programs. For example, FX Charts (1999-2001) was calculating the indicators more accurately (double vs float mathematics) than MetaStock.
It is not correct to compare them, at least Prival's theory is all right.
Please don't compare Academic and Prival.
He is exactly the same theoretician, not even willing to calculate figures and volumes (to him - your approach will require NN gb of tick data! - it won't! - so what!), and then compare it with physical feasibility.
The theory doesn't matter (no use) to others, when a person hasn't tested it in practice, and even more, assessing it one-sidedly (from the point of view of one/the trader side, instead of assessment from 3-5 involved sides). On the contrary, he misleads people with his "theory", replacing the potential beneficial effect with only verbal exercises without evidence.
If only this were done in an environment where there is no one knowledgeable about the issue.
It is the practitioner who does not surrender, defending what he has coaxed out with his own sweat.
And you really gave it all up when the tale/theory of acceleration on integer calculations failed. It was enough to get SMA/EMA/LWMA to make everything fall apart.
You are engaging in outright stupidity by trying to attribute my ignorance of real numbers.
Your level of "given accuracy" was perfectly illustrated by the example of integer muwng calculation. Even the level of accuracy double is not enough, although some world programs like MetaStock (at least version 7.xx, I haven't checked for a long time) for the sake of economy, they calculated everything in float, which caused serious discrepancies with other programs. For example, FX Charts (1999-2001) was calculating the indicators more accurately (double vs float mathematics) than MetaStock.
Ehhh - Rational numbers, these are numbers in the form of fractions.
Calculation in these numbers does not know such a thing as rounding errors. By definition. And you say you know what I'm talking about?
Numbers like A/B+C/D.
Wolfram Mathematica EXAMPLE does such calculations more than once.
Don't pull any ears, conversion to dubbles for computation. There's no need for any conversions there. At all. It's as simple as that.
Ehhh - Rational numbers are numbers in the form of fractions.
It's easy to jump from one failed statement to another.
When you have a processor with rational data types and when the next set of SSExxx commands can handle them faster than double, then you can link rational numbers to the discussion about speeding up calculations. When you publish tests of SMA calculation with different methods and show winning over double, then it will be a speech of practice.
In the meantime, the original statement about acceleration of real mathematical calculations by switching to integers has failed.
Calculus in these numbers knows no such thing as rounding errors at all.
Actually, it is obligatory to know -- the case of an overflow denominator.
That's one. And second, I doubt very much that arithmetic with double is slower than with rational numbers.
Actually, it is obligatory to know -- a case of denominator overflow.
That's one. And second, I doubt very much that arithmetic with double is slower than with rational numbers.