Discussing the article: "Non-linear regression models on the stock exchange" - page 2

[Andrey Khatimlianskii]

Vitaly Muzichenko #:

Did you find the answer to the question?

This is the most important point of any TS, you can't miss signals, otherwise the whole logic collapses, or is it wrong here?

There are no skips in the posted EA. This is obviously not a code from a real account, there are no filters mentioned here.

Just a demonstration of the idea, which is not bad either.

[Vitaly Muzichenko]

Andrey Khatimlianskii #:

There are no omissions in the posted Expert Advisor. This is obviously not a code from a real account, there are no filters mentioned here.

Just a demonstration of the idea, which is not bad either.

I agree

[Михалыч Трейдинг]

Just kapets (sorry)! During several hours of studying your materials for the 10th time I see that we walk the same roads (thoughts).

I really hope that your formulas will help me to formalise mathematically what I already see/use. It will happen only in one case - if I understand them. My mum used to say, "Study, son." I cry bitter tears in maths. I see that many things are simple, but I don't know HOW. I'm trying to get into parabolas, regressions, deviations.... It's hard to go to 6th grade at 65.

// It is not enough to just throw a bunch of cool features - you need them to complement each other harmoniously, creating a really reliable trading system.

Yes. Both the selection of features and the subsequent optimisation are like straightening the figure eight of a bicycle wheel. Some spokes should be loosened, others should be tightened and it should be done in strict obedience to the laws of this process. Then the wheel will be levelled, but if the wrong approach is taken, if the spokes are tightened in the wrong way, it is possible to make a "ten" out of a normal wheel.

In our business "spokes" should help each other, not pull the blanket on themselves to the detriment of other "spokes".

[Mai Abboud]

I don't think it's effective to predict price based only on the last two data points.

Do you agree?

[HeAic]

To try it out, I added to the original MarketSolver.py a gui-ish interface on Qt5 and a selection of currency pairs (this is for experiments). coefficients are in the console IPython

[HeAic]

Mai Abboud #:

I don't think it's effective to predict price based only on the last two data points.

Wouldn't you agree?

From my personal trading experience, the best way to analyse is zero bar - ticks and first bar. That's kind of enough :)