Optimal values of SL and TP orders for an arbitrary TS. - page 17

 
aZtec >>:

...Я понимаю, что на первом этапе вы пытаетесь максимально упростить задачу, и в результате получаете условие необходимости постоянного нахождения в рынке. Однако, это условие всё же вряд ли можно считать оптимальным для реальной торговой системы.


You are certainly right to point out the problem. But that doesn't make it any easier...

I will translate what you said into terms that define the system of interaction between the TSO and the Market. Analyzing the general view of the FR, we can distinguish only two stable and absolute (not relative to anything) states of the market - trend and flat. All studies show that the main condition in which the market is almost always is a flat condition, and only with rare exceptions, the market shows herd properties (trend character). Of course, you can configure TS to take profits from the main state and hope that the parts of the opposite mood are rare enough to cause a noticeable damage to the profitability of TS. Maybe so, then your concerns are valid, but not fatal for the TS. Otherwise, we would need to come up with a mechanism to change the TS algorithm to an inverse one (relative to the underlying condition) just before the expected news release. It's troublesome, but probably necessary. The topic needs to be explored.

So, I don't see the need to abandon the paradigm of always being in the Market, but I agree with the possible need to "invert" the OTC Front End before the news release if it is localized in time. If there is no definite temporal determinism, then the problem remains the problem of the nervousness of market processes. Anyway, without having completed the identification of the optimal TS for the stationary market and finding the optimal MM, expand the area of parameterization of the object should not be otherwise get bogged down in details, albeit defining.

 
The assumption of optimality of continuous staying in the market (in terms of profit maximisation), imho, remains only an assumption. In fact, let us suppose that TS has the positive expected payoff only in certain sectors of the CR corresponding to certain market conditions. Then, as Sergey has shown earlier, expectation of profit (not considering the spread) in other segments will be zero and there will be no need to exit the position. But we are interested not only in profit but in risk as well. The situation with the risks is different, in the moment of uncertainty we will exit the market to avoid drawdowns. Correspondingly, we can increase the share of capital in the game, thereby increasing the mass of profit. Although the relative profit (pips per trade) can decrease in this case.
 

For example, if we have a number of signals (let's assume all profitable), Buy, Out, Buy, Out, etc., the proportion of f increases sootv deposit also increases, in your version f remains constant.

P.S. . My aim is to show how easy it is to spot f on the map, because I have no idea what to do with it.

 
Neutron писал(а) >>

Let us also say straight away that

1. no TS is able to get MM different from zero on martingale (integrated random variable (IC) with zero MM - in first approximation analog of price series) if the history is long enough.

i'm sorry, if a stop is really rare (50 Bets before it), 50 Bets behind it, after the stop, you can go to the market hoping it will work out. or else the statistics will lie to me this time ... if not, i'll double my deposit 30 times for a 100 pips profit for the whole lot and leverage 1 to 100. and wait for the stop, then again ..... . i don't think it happens that way, or the system as such will become obsolete in this month? i don't think so either. i think it's better to take risks for small stops than to take profits seldom, but for manual, not mechanical systems. i completely agree with you.

 
Neutron:

Let's go! Let's take it slowly...

Let's start with algorithmization of the simplest arbitrary TS with reinvestment of capital f. Recall that in our case the capital fraction f is defined as a relative and dimensionless value of funds per point of price movement. Suppose at the initial time we had capital K[0] and as a result of the first transaction we have earned (lost) from the market h[1] points where h can take any natural values, i.e. h can be equal to 5 points (and we have won the bribe) or -51 points and we have lost (returned to the market) 51 points. Then the monetary gain of our capital as a result of the first transaction will be determined by K[1]=K[0]+h[1]*f*K[0], it may be both gain and loss of capital, everything is determined by the sign before h[1] and its absolute value. For the second transaction the expression looks similar to the one already written: K[2]=K[1]+h[2]*f*K[1]. Let me remind you that the fraction f of the capital involved in trading is fixed. In general, after i transactions our deposit size will be determined by K[i]= K[i-1]* (1+h[i]*f), considering that we have already got an expression for K[i-1] we can substitute it in the last formula and get K[i]= K[i-2]* (1+h[i-1]*f)* (1+h[i]*f). Continuing along the chain we get:

We've got the expression that shows the relative value of the increment of our deposit K[n] to its starting value K[n] through n transactions for an arbitrary TS that is determined by the value of its bribes h[i]. The symbol P stands for the product of brackets by each other. That's all for now. The point is that we cannot go further with expression for deposit growth presented in this form. But we can try a trick, in particular, recall the fact that values of point bribes h[i] are integers, and in case of large number of transactions we can always find groups of bribes with the same number of points in each bribe. Thus, regroup the terms in the product into a "product of interest piles" and take advantage of the fact that by rearranging the terms in the product, the product does not change.

I'll continue later...

With some assumptions, f can be calculated this way:

 
hrenfx:

With some assumptions f can be calculated like this:

Some very strange reasoning in this thread.

Doesn't the TS itself define the SL and TP levels?

Do you seriously expect anyone to do these calculations?

To consider an "arbitrary" TS is to completely disconnect from reality and move into a fantasy world of utopias,

which, in fact, the author successfully demonstrates.


 
more:

Some very strange reasoning in this thread.

Doesn't the TS itself define SL and TP levels?

To consider an "arbitrary" TS is to completely disconnect from reality and move into a fantasy world of utopias,

Don't stop people from having fun!!!!111!!!

Comrades, please continue, very interesting...

 

The use of f makes the most practical sense.

f shows which "leverage" to use for your (real) TS.

You can calculate it through optimization methods (e.g. optimization of MM in a tester), or as I wrote above.

 
hrenfx:

The use of f makes the most practical sense.

f indicates which "leverage" to use for your (real) TS.

You can calculate it through optimization methods (e.g. optimization of MM in a tester), or as I wrote above.


fiction this optimal f is a fitting of historical statistics. The performance of the system is not constant. The main thing is the criterion of system failure - when to scrap it. And these criteria are individual for each system. MM should be such a way to earn maximum money up to the moment of abandoning the system or to drain no more than a certain amount of money :)
 
There is much talk about how it should ideally be.