Are "momentum" strategies possible? That is, catching a continuation of a strong move ... - page 2

New comment
 

I get strange things in backtesting.

For 10 minutes, good results come from strategies with large SL=TP of about 50 pips,

This is very strange to me, as the time in position turns out to be long - on the order of several hours

- it would seem to defy logic - we make a decision for several hours based on the momentum of 10 minutes.

Could this be the case ?

And the results are "stable" - I've plotted the dependence diagram - it turns out that SL=TP = 10-20 - steadily in the negative side, SL=TP = 40-50 - steadily in the positive side.

Stable - for the period of the euro - this year. Time N=10 minutes, the threshold changes and TP=SL changes.

 
Zogman:

Who knows about these simple strategies:

if price has changed more than the "X" threshold in N minutes, then open a position in the direction of the move and wait for take profit or stop loss.

I would be grateful for any comments.

PS

I have modeled this strategy myself.

On history there are such parameters, however, if you continue into the future, nothing particularly good comes out.

For example, for February-April 2015,EURUSD,

for N=8 min, X=8 pips, TP=SL=50 pips

Earn 2,000 pips.

However, the problem is that sitting in a position for a long time is a sign that this is a random earning.

If we run these parameters for a year and a half of history from the summer of 2013 - there will also be plus 1792 pips, but all earnings are mostly at the end of the term.

I have something to say on the subject. I'm on my mobile now. I have no information. I will be at work after the holidays. I got into the subject myself not so long ago. I have already done something in terms of drawing up TC. There are many variants of TC. The code was started by me. There are references to this information and how to interpret it for profitable trading.
 
Zogman:


If there is a desire. I can give you a link to someone who deals with this issue here. I was looking into getting a robot myself... then tested it and came up with mixed results. Went to other links...
So that it is not here as an advertisement.
 

There was an attempt to trace master - slave on correlated pairs .... the idea was abandoned ... perhaps because it is impossible to tell unambiguously who is the slave and who has just momentum, which will immediately choke after the jerk ...

https://www.mql5.com/ru/forum/127925

http://forexsystems.ru/rekomenduemye-ruchnye-torgovye-sistemy/65072-torgovaya-sistema-korzina.html

The vertical red line is the bar where the coefficient conversion ends, next is the OOS. Figures with a difference of 500 minutes.


 
Art:


Thank you very much for the links !

Have you tried arbitrage or market making? And use this idea for similar purposes ?

 
Roman Shiredchenko:
If there is a desire. I can give you a link to someone who deals with this issue here. I was looking into getting a robot myself... then tested it and got mixed results. Went to other links...
So that it is not here as an advertisement.
Have a desire ))

Tested the theme myself. The results are again ambiguous. On passing a certain number of ticks - complete nonsense!!! Maybe also something I do not know and did not take into account ... By the way, now I am too lazy to think about it - maybe I do not know what side to open, but I do not know what side to open. By the way, I'm too lazy to think now - may someone answer my question - why does the spread widen during the news, who exactly widens it and for what purpose?
 
Art:

There was an attempt to trace master - slave on correlated pairs .... the idea was abandoned ... perhaps because it is impossible to tell unambiguously who is the slave and who has just momentum which will immediately choke after the jerk ...

https://www.mql5.com/ru/forum/127925

http://forexsystems.ru/rekomenduemye-ruchnye-torgovye-sistemy/65072-torgovaya-sistema-korzina.html

The vertical red line is the bar where the coefficient conversion ends, next is the OOS. Figures with a difference of 500 minutes.


Dick's thoughts are always interesting - we'll read it. thanks!
 
Daniil Stolnikov:
There's a wish ))

I've been testing the subject myself. Again, the results are mixed. On passing a certain number of ticks - nonsense complete!!! Maybe also something I do not know and did not take into account ... By the way, now I am too lazy to think about it - maybe I do not know what side to open, but I do not know what side to open. By the way, I am too lazy to think now - may someone answer my question - why does the spread widen during the news, who exactly expands it and why?
On the first page I wrote my assumptions about spread widening at the news.
 

Such an experiment (made by mistake, but might be of interest).

Strategy - as described above - but the entry threshold is very small - that is, if not in a position, we enter it immediately.

It turns out that on the same backtesting (with the addition of the spread in pips) we can earn about the same - in all likelihood this is overfitting

PL in pips N - analysis time minutes TP SL Number of TP SL Time in position
1801.4 4 65 -65 54 31 1227.3
1803.4 2 60 -60 56 30 1195.3
1831.1 20 80 -80 36 16 1993.9
1835.6 7 60 -60 59 33 1122.7
1872.9 16 80 -80 37 16 1967.9
1889.6 19 45 -45 92 56 683.44
2038.6 18 45 -45 91 51 673.68
2142.8 1 95 -95 30 9 2457.5
 
forexman77:
On the first page I wrote my assumptions about the widening of the spread on the news.
read the whole post. no response.
12345678910
New comment