Where is the line between fitting and actual patterns? - page 12

New comment
 
paukas: Well, purely empirically. If the drawdown is three times the test drawdown

If the drawdown exceeds the test one by 3 times - then we are screwed )))) How not to let that happen? )))
 
LeoV:

The question is not what patterns, but how to find a training period and an OOS period for the patterns.

))

To be honest, I've been thinking for over a week about creating a thread called

"Correlation of optimization and testing periods. Or optimization of optimization periods...".

Or fitting the second instance... (Sorento).

With screenshots and graphs. To make it more serious. ))

And here's where the question is: What patterns and confirmations of their workability do we want to find? -- comes to the forefront.

 
lasso: To be honest, I've been thinking for over a week about creating a thread called
To be honest, I've been thinking about it for over four years ))))
 
LeoV:

If the drawdown exceeds the test drawdown by a factor of 3, then we are screwed )))) How can we prevent it? )))
For example, a way to reduce the lot at a drawdown. It will be huge in pips, but not so huge in money :)
 
sever30:
I wiped my flub, but can't let you know because I keep deleting this post.


Appreciated your humour)

But the merchants, as you can see, are silent....

Do you think we should ask for a third time? We're not proud....

Or shouldn't we?

 
paukas:
Well for example a way to reduce the lot at a drawdown. It will be too big in pips, but not so big in money :)


This is self-deception.

In the process of developing the TS, the lot must be constant.

(Again, slyly.... )

 
lasso:


1. Appreciated your humor.)

But the merchants, as you can see, are silent....

2. Do you think we should ask for a third time? We're not proud....

Or shouldn't we?

1. Trying.

2. Already asked.

 
paukas:

Here is the equity of one system that exploited volatility. It was calculated based on 2007 volatility increased at the end of 2008 and started to decrease substantially. Unfortunately, DC rolled back the transaction history of old periods quarter by quarter, but still a little bit visible.

This strategy may have worked for several symbols (currencies) with the appropriate settings? If so, then you're the first to find the grabble)))

If I look at the chart for a major, the graphs are very similar to each other, provided that visually we need to account for the profit value per pip and therefore the patterns for the dollar exist, then according to the subtext - if after selecting the parameters for TS, TS can work for several majors (each with its own settings) - so TS works according to patterns, but not adjusted for the historical period

 
sever30:

2. Already asked.


Don't compare yourself to the merchant class ;-)

They have very different chains of logic.

 
IgorM:

This strategy may have worked for several symbols (currencies) with the appropriate settings? If so, then you're the first to find the grabble)))

for the major charts are very similar to each other, provided that visually we need to consider the cost per pip profit, and so the patterns for the dollar exist, then on sub-soot - if, after selecting the parameters for TS, TS can work for several major (each with its own settings) - then the TS works according to patterns, but not adjusted for the historical period


Igor, look, there was an optimisation with three parameters, each with 10 values (agree, this is very modest).

A total of 1000 passes.

How to select the only final set (FN) of values of parameters? What are the selection criteria?

Do you know the answer to this question?

Without answering it, there is no point in going any further!

1...5678910111213141516171819...64
New comment