Not the Grail, just a regular one - Bablokos!!! - page 256
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
purely out of sporting interest, I'm working on beating the top-starter's record... %), i.e. to break the record he achieved - to increase the depo by more than 100 times in a month...
Be patient, Uncle Fyodor, there are still two dresses left
Thank you, that's funny ))))
Let me tell you one more truth: working with zero spreads, for example when you found a model with positive values for three weeks, then the parameters of this model have a positive yield in the form of an inverted parabola, which also change its centre over time (i.e. move left or right, from lower values to higher, and vice versa).
Market neutral spreads allow you to slow the change of these parameters, i.e. the speed of movement of these parabolas, and by selecting the optimal values of the trading system per unit of time you have time to positively trade these parameters before they change to unprofitable values. For example, you found a pattern in the past that gave you positive parameters for three weeks or a month. By trading maximum rates of return of this model after this model is built you will never get the same maximum return, which this model has shown, you will get 50 or 70 percents of its parameters because this model moves and it needs to be re-optimized afterwards. I.e. for example during the next week you can trade off this model (if the system model on market-neutral spreads for example shows a return of 500-700%, then you will pay off for example 250-300% maximum during the next week, because the model shifts). Do not pay attention to the numbers 500, 700, 200, 300 - that's for the sky-high extreme risks. In reality I work with models that show approximately 100% on a simulated period with low marginal load on the account, which in practice gives 20-30% return per week. And frankly speaking pioneer - it's enough for me, I'm not going to chase after more %)
Further this model should be regularly re-optimized.
The more instruments in the market-neutral model - the less volatile its optimal parameters are (i.e. the slower moving parabola of optimal parameters), but the longer we need to wait for the profitability of the position.
How does it differ from e.g. usual trading tools? Well, the optimal models that you find in strategy testers work so short time that you do not have time to trade them positively, and the market eats up your positions together with your money. In this case only fast robots with low targets (like scalpers, pipsers and other bells and whistles) that brokers like to rip up work positively.
The market-neutral models don't give a shit what's going on in the market, they are market-neutral.
In short, gentlemen, mathematics, programming and toil, toil, toil...
purely out of sporting interest, I'm working on beating the top-starter's record... %), i.e. to break the record he achieved - to increase the depo by more than 100 times in a month...
Something tells me you can do it %)
So far I have taken the bar 9-fold growth in a fortnight (an example of such a model in the attachment - above) that will give a month 81-fold increase. Models with 12-fold growth in 2 weeks I have now, but they are very unstable (with such rabid risks we obtain a stabilizing model with parameters 12x8), ie theoretically 12*12 = 144-fold theoretical increase in a month is already possible, but the technology superposition should still be stabilized.
The main problem is to catch the sign of the direction of the optimal parameter set and indirect indicators to solve this problem already exist, but they still need to be worked on.
Examples of parabolic TS parameters are below:
this is the thing that walks left and right along the set of optimal parameters with time, which changes the model of the optimal TS. If you work stupidly to the maximum, then there is almost no chance to fail - can only decrease the yield because your TS will be on one of the arms of the parabolas of the optimum parameters of TS and fall only after a long time ( roughly speaking let it drift - you will draw a stop loss).
With the lapse of time we can determine in what direction the set of optimal parameters is moving and acting proactively we can catch the maximum value of this set in the future, thereby maximizing the effectiveness of TS. To do this we use ordinary physics with elementary problem of the 1st class:
A train has left point A in the direction of point B, moving at speed N, the question is: how long before it arrives at point B? or Where will it be in time T?
Only this train is special - it can stop halfway and go in the opposite direction )as the angles of the spreads can change in any direction, but in the long run their angles are linear.
Let's simplify the problem for understanding when solving it - let's introduce some sense of humour )))):
The set of optimum parametres has the form as shown in the figure above. in the figure below I have shown how to identify the direction in which the model is now moving. The identifier of the direction of movement is the bow of the ship with a characteristic cliff. To achieve maximum TC efficiency in this case, we need to use the set of optimal parameters indicated by the red dot, so there is a high probability that we will be on top of the TC efficiency at a later point in time (in the future):
If you remember, in Soviet times there was a popular slot machine in cinemas
Well, it's a similar situation to ))))))))))))))))))). - The torpedo, when it arrives, should hit the centre of the ship. And if it hits the bow or stern, that's okay too. The problem will be solved and the target will be hit (i.e. we will take the profit from the market).
I remember, of course, the most sensible machine was )))
Hi! And thanks for the comments and hints.
It turns out that gStartBar is an important parameter.
And if you take the other two weeks of another month for optimization? Is it also hundreds of percent?
Hi! And thanks for the comments and hints.
It turns out that gStartBar is an important parameter.
And if you take the other two weeks of another month for optimization? Also hundreds of percent?
Yes. In any month in any week. As for the percentage, it depends on what risks you are working with.
In this case gStartBar is distance from normalized model to decision point in time (in my case anyway).
ModelWidth is the length of the model of the normalized channel.
Another key parameter - ModelIDX - is the distance from the axis to the lower boundary of the spread-flow variance channel. Who has entered the decision zone of this model, that is what we trade.
The superposition is formed by trading the set of spreads in the decision point of the current optimal model.
I have basically covered everything that I could.