How much is needed.... - page 17

 
Korey:

to Mathemat to Figar0

From a source in 1998, a programmer who went to work for 300K/year, artificial intelligence for hedging,
it is known that everything written here by locals has been in a complete set of purchased smart stations for about 10 years,
which have always been programmed at least as well as in a C++ environment,
and again in a 3000 c.u. purchase box there are + about 10 ready-made robots, including neuro.
(It is even easier with training artificial brains - a workstation is purchased or the time is stolen at work).
In other words, everything that can be squeezed out of slippery machines has already been done.
Locals are doing homemade stuff, starting with LRMA, graph rendering, etc.
I.e. MT-X is the 4th wave-generation autotrade. So to speak, the people's, nationwide wave, for the people's wave.
Will there be successful robust solutions in the 4th generation?
It is the same as to ask if it is possible to go round the globe on a moped.

P.S. I posted there the link to S.S. Belyakov's dissertation, so the file was banned on the server, and the funny thing about this diligent vigil is,
S.S. Belyakov did not bring anything new, but he explained to the Russian-speaking audience how purchased software works.

Just recently in the news, hedge companies (USA) lost more than 5,000,000,000 not rubles in 2007.
"Good algorithms and programmers at least as good as c++". :))
 
Andy_Kon:
It was in the news recently that hedge companies (USA) lost more than 5,000,000,000,000 not rubles in 2007.
"Good algorithms and programmers are at least as good as c++". :))
It is precisely this news that confirms the credibility of robust solutions.
 
Korey:

no lower than in a c++ environment,

...

I.e. everything that can be squeezed out of slippery mash-ups is already done.

...

P.S. I posted a link to S.S. Belyakov's dissertation there

1. Yes, if it was all about coding algorithm... It's about the algorithm of preparing the initial data (just like in neural networks!), you understand. Even "best" mashup on some data with Cauchy type distribution (no m.o., much less s.k.o.) will not work well.

2. I don't think so. The best slippery quasi mashups have already been created (already JMA from Code Base is a very good product. Competitors like LRMA or TEMA aren't bad either, but they lack smoothness). That is, I'm saying that the Blue Dream of a beginner trader (slippery, non-lagging and simultaneously fast on strong movements) is implemented quite well. But it alone is not enough to create a robust trading robot. Something else has to be done.

3. Do you have the dissertation itself?

 
Mathemat:
2. I don't think so. The best slippery quasimats have already been created (already JMA from Code Base is a very good product. Competitors like LRMA or TEMA aren't bad either, but they lack smoothness). That is, I'm saying that the Blue Dream of a beginner trader (slippery, non-lagging and simultaneously fast on strong movements) is implemented quite well. But it alone is not enough to create a robust trading robot. Something else has to be done.
Extrapolation has to be done
 
Prival:
Mathemat:
2. I don't think so. The best slippery quasimats have already been created (already JMA from Code Base is a very good product. Competitors like LRMA or TEMA aren't bad either, but they lack smoothness). That is, I'm saying that the Blue Dream of a beginner trader (slippery, non-lagging and simultaneously fast on strong movements) is implemented quite well. But it alone is not enough to create a robust trading robot. Something else has to be done.
Extrapolate.
I finally figured it out!
 
Korey:

Korey, good afternoon to you. I take it you are no longer interested in comparing VTE and TA forecasts? At least point out the success of the two VTE forecasts that have already come true.
I'm working my ass off for you - and you just ignore me. Of course, I am interested in your opinion as a professional.

The yen forecast has not yet come true. VTE says 104.00 is just around the corner...
What does the TA tell us on this issue?
 
Reshetov:
Prival:
Extrapolation has to be done
Finally figured it out!
I don't see why not. I've known for a long time. And in the Kalman filter it is one of the procedures. But there is 1 nuance: there cannot be qualitative extrapolation without qualitative interpolation, if our current estimation is +- lapot accurate, then the movement forecast will be like in Pikul's "three laps to the right of the sun and three days travel through taiga, you won't miss the gatehouse :-)".
 
Mathemat:

3. Do you have the dissertation itself?

I have the dissertation itself, 1.5 metres. I had no problems downloading it, until you put up the link.
Now the interesting situation is whether it can be uploaded if the URL of the file remains on the source's website, but the file itself does not exist?
 

to Sar

How could it not be interesting?
You are showing a master class, one should not miss it.
I think I've got a good chance to fix it with a few clicks,
I have no account with Alpari but I have to use demo version. I have a look therefore indirectly, post more often and with screens, it is interesting.

 
Korey:

to Sar

How could it not be interesting?
You are showing a master class, one should not miss it.
I think I've got a good chance to fix it with a few clicks,
I have no account with Alpari but I have to use demo version. I have a look therefore indirectly, post more often and with screens, it is interesting.

Well, here's another thing. But more specifically, could you take the time to make a TA USDJPY. I would be very interested to see it.
Reason: