Machine learning in trading: theory, models, practice and algo-trading - page 3419

[Maxim Dmitrievsky]

fxsaber #:
Is it unsolvable?

Temporarily, maybe.

[mytarmailS]

Maxim Dmitrievsky #:

There's a bot at the top of the marketplace right now that only buys.

A TS that opens and closes trades only in longs is one thing, but a buy&hold on one brokerage house and a sell&hold on another brokerage house is quite another.

buy&hold on one brokerage house and sell&hold on another brokerage house is quite another.

[Forester]

fxsaber #:

Intuitively to me it looks like a fit.

Optimising with a tester is also a fit. And MO is a fit. But if there's a pattern, it's a fit. IO is better because it can do 1,000 features. Manually it will be incredibly long, like 1st bar is 100pts higher, and 5th bar is 200pts lower, 15th bar is higher ..........

[fxsaber]

Forester #:
Optimising with a tester is also a fit. And MO is a fit. But if there is a pattern, it will fit.

To the point that if there is a pattern, it is the same both ways.

[mytarmailS]

fxsaber #:
To the point that if there's a pattern, it's the same both ways.

you're dead wrong.

or you're wrong.

here you go:

Maxim Dmitrievsky#:

There in the marketplace on the first place now a bot that only buys.

[Maxim Dmitrievsky]

mytarmailS #:

you're dead wrong.

or you're wrong

here you go:

Maxim Dmitrievsky#:

There in the marketplace on the first place now a bot that only buys.

It's hard to even call it a pattern there... it's more like a well-adjusted average without stops :)

we don't care about regularities when we can sit out losses :)

[mytarmailS]

Maxim Dmitrievsky #:

It's hard to even call it a pattern there... more like a well fitted average with no stops :)

Well, if it's a stupid averaging, the authors didn't bother to add sells? )))

Well, I know personally on Forex one pattern that only in longs works like clockwork every day.

The same inflation is also a pattern, the stock market always grows thanks to it.

[Maxim Dmitrievsky]

Here you go

The algorithm is very nice (in terms of how implemented, via python dictionaries and MO). I was very happy with it.

[fxsaber]

mytarmailS #:

you're dead wrong.

or you're wrong

I'm correct. I don't understand why the pattern is not common to both sides, and what in the original price series should indicate that.

[fxsaber]

We have three optimisation results.

1. OnTester for Long&Short.
2. OnTester for onlyLong.
3. OnTester for onlyShort.

Assertion: combining n.2 and n.3 is better on Sample than n.1.

Scenario: if n.2. is much better than n.3 (automatically better than n.1), then leave only n.2. with the belief that the pattern is one-sided.