Where is the line between fitting and actual patterns? - page 16

[Vladimir Paukas]

Debugger:

Regularisation algorithms are algorithms that prevent the over-learning effect of NS. If you google it, you can find quite a few implemented and working.

Thus with these algorithms, the very topic as 'the line between fit and regularity' simply disappears.

Although the question of network architecture remains open.

And don't confuse normalisation and regularisation.

Normalisation is about bringing differently scaled data to the same scale.

And lastly, not all types of NS work well in financial markets.

What types go well?

[Serqey Nikitin]

"Where is the line between fitting and real patterns?"

For myself, I have solved this question very simply: the same TS settings are bound to give approximately the same profitability, of course,

corrected for the volatility of the currency pair, while working on multiple currencies, at least 3-4 major pairs. If this condition

it is possible to fulfill this condition - the adjustment is excluded.

[Владимир Тезис]

I think that if the code works according to real patterns, it will work in profit even without any tweaking (tester optimization of ranges of using code parameters). This is probably what makes it different.

[Vladimir Paukas]

VNIK:

"Where is the line between fitting and real patterns?"

For myself, I have solved this question very simply: the same TS settings are bound to give approximately the same profitability, of course,

corrected for the volatility of the currency pair, while working on multiple currencies, at least 3-4 major pairs. If this condition

it is possible to fulfill this condition - the adjustment is excluded.

Not a fact. Dollar pairs run parallel - it doesn't help there.

[Yury Reshetov]

drknn:
I think that if the code works according to real patterns, it will work profitably without any tweaking (tester optimization of ranges of using code parameters). This is probably what makes the difference.

This is the hallmark of the grail. Financial instruments are non-stationary and therefore do not have stable patterns: dense today, empty tomorrow, etc.

[Andrei01]

Reshetov:
Financial instruments are non-stationary and therefore do not have stable patterns

Isn't non-stationarity a stable pattern? :)

[Yury Reshetov]

Andrei01:
Isn't non-stationarity a stable pattern? :)

For the very gifted: non-stationarity is the absence of statistical regularities such as expectation and variance.

Put Bollinger envelopes on the chart and you can see what the "patterns" of nonstationarity are, because the centre of the indicator is expectation, and the distance from the centre to the envelopes is dispersion.

[Леонид]

Andrei01: Isn't non-stationarity a stable pattern? :)

Of course, non-stationarity is also a kind of regularity. But you can't make any money from it ))))

[Vladimir Paukas]

LeoV:

Of course, non-stationarity is also a kind of regularity. But you can't make any money from it ))))

It depends on where you work. ;)

[Avals]

LeoV:

Of course, non-stationarity is also a kind of regularity. But you can't make money on it ))))

For example, stock indices of prosperous countries have a constant upward trend. And the dumb rule of "let profits rise and cut losses" is using non-stationarity with positive mo for longs. True that's also why dips happen much faster than periods of growth))) Profits are composed of several very unsteady trades :)