From theory to practice - page 611

New comment
 
Unicornis:

but nice )

increments not in points

69 pips 4-digit maximum

average - 14

for the euras

the signal is unstable and varies if we look at the observation windows at different time intervals

so...

 
danminin:

Trade manually and you'll be happy.

I've been thinking...
about the difficulty of testing trading systems manually.
When you test a system manually, what do you expect from it? that all trades will be profitable. and if there are losing trades, we'll think the system isn't working.
There are some systems with only 53% of profitable deals, but these systems are nevertheless good.
Even if you trade on it manually for a month, it won't show a nice equity chart.

 
danminin:
That's it, trade, what more do you need? )
The best is the enemy of the good! :-)
 
danminin:

I've been thinking...
about the difficulty of manually testing trading systems.
If we're testing a system manually, what do we expect from it? we expect all trades to be profitable. and if there are losing trades, we think the system isn't working.
There are some systems with only 53% of profitable deals, but these systems are nevertheless good.
Even if you trade on it manually for a month, it won't show a nice equity chart.

Yeah

and don't forget to set a 2,000-second delay,

and make sure the broker's giving us a good tick history for the year,

And the spread in this story corresponds to the real one, not the drawn one...

And a dozen other similar little things.

 
Natalja Romancheva:

Yeah

and don't forget to set a 2,000-second delay,

and make sure the broker's giving us a good tick history for the year,

And that the spread in this history corresponds to the real one and not the drawn one...

And a dozen more such little things.

ok, check manually for the year. ))

 
danminin:

ok, check manually for the year. ))

No, no way!

It's just that there are a lot of nuances to consider when testing!

 
Evgeniy Chumakov:
Alexander! Make a histogram from this file, I think it'll be interesting. Although who knows.

Eugene, still, was this bimodal distribution obtained for some sliding window or just 30.000 OPEN M1 taken and cunningly grouped?

It seems to me that if the work was done in a sliding window = 24 hours, then the correct time window of observation = 12 hours.

I remember in my studies, too, a normal distribution for the sum of increments was obtained just for 12 hours, and as the window size was further increased, this distribution "blurred" and the kurtosis became <0.

 
Alexander_K2:

Eugene, still, was this bimodal distribution obtained for some kind of sliding window or just took 30,000 OPEN M1 and cunningly grouped?



These are the sums of the increments in the dynamic observation window.

 
Evgeniy Chumakov:


These are the sums of the increments in the dynamic observation window.

!!! Holy crap.

 
Alexander_K2:

!!! That's amazing.

It's a question of pulling those ears to make it work. It's just a pretty picture. I've put it in a drawer for now.
1...604605606607608609610611612613614615616617618...1981
New comment
Reason: