Hedging Martingale. - page 23

New comment
 
I think it makes sense to combine this with a fundamental strategy - expensive currency will go up, we will hold the expensive currency
 

oops, the messages about the forbidden grail are gone )))

 

I decided to share one fact that has only now become apparent to me, perhaps it will save time for someone :)

testing was done using two strategies, USDJPY was intentionally chosen as a trend one, since the strategy is flat, it should show the worst case scenario that can occur

1. trading on rebound from indicator levels

1.1. testing with fixed lot 0.01


1.2. testing with fixed lot 0.02 and larger


2. trading on rebound with possible reversal, if we are in the market and there is a reversal signal, we may say, trading on level breakdown

2.1. testing on fixed lot 0.01


2.2. testing on fixed lot 0.02 and larger

The first fact is that regardless of whether a Martin is used or not (it was NOT used here) equity of funds should have a higher change rate than the price change rate, although there may be a mathematical explanation with formulas and stuff like that

 

And immediately fact #2 - in reversal systems on a breakout such as a pallet, a moderate increase of the permissible corridor leads to an improvement of the drawdown-profit ratio. In particular, here is how chart 2.2 from the previous post may look when the channel is doubled:

when the corridor is increased by 5 times, "hang-ups" start to appear

 

strange that 1.1 and 1.2 are so different

no martin was used - a wait and see strategy? or a milder progression?

what was the maximum lot limit reached?

 
artemiusgreat:

And immediately fact #2 - in reversal systems on a breakout, such as a pallet, a moderate increase in the permissible range leads to an improvement of the drawdown-profit ratio. In particular, here is how chart 2.2 from the previous post may look with a 2-fold increase of the channel:

when the corridor is increased by 5 times, "hang-ups" start to appear

The problem is that a modest increase in the corridor leads to improvement, but not always.
 
transcendreamer:

strange that 1.1 and 1.2 are so different

no martin was used - a wait and see strategy? or a milder progression?

what was the maximum limit of lots reached?

Good question, i got interested myself and decided to experiment with different ratios of risk-profit, in our case, the risk is the width of the corridor; i decided to test using beautiful charts from the post above

1. in the first case, the increase in lots was from 0.02 to about 0.8, the higher the desired profit, the greater the increase in lots, variations in this system are only allowed along with an increase in the deposit ...

2. according to the second variant (with reaction to a change of trend and reversals) - from 0.02 to 0.56 with 2-fold increase of the corridor (risk), the profit - any, hence, the fact №3 - in the reversal system the desired profit does not affect the drawdown, which is surprising, with a 3-fold increase in the corridor (risk) the maximum lot size fell to 0.46, which once again confirmed fact №2, in the reversal systems, increasing the acceptable bandwidth is NOT inherently increasing risk, and even the opposite

 

To confirm this data, you need to test as much history as possible and different instruments.

Do you have a lot limiter, or does it just happen?

 
what will the results be on the other story?
 
edutak:

To confirm this data, you need to test as much history as possible and different instruments.

Do you have a lot limiter, or does it just happen?

1. I think we can use the word "you".

2. there are no limits

3. i have already laid out the images and test results for 4 years with the quality of history at 45%, from 2010 to now, the earlier history from the terminal is useless or not available at all

4. The images in the last post are just the last year with the 99% quality of history

So i created another thread on this forum asking about the quality of testing, but as always such threads are ignored at best, possibly deleted, at worst - ban

https://www.mql5.com/ru/forum/38115

for normal testing you need to write your own tester, as close to the real thing as possible, like this, and it takes time, which I do not have

https://www.mql5.com/ru/forum/10920

https://www.mql5.com/ru/code/1661

https://www.mql5.com/ru/forum/5108

1...16171819202122232425
New comment
Reason: