Market etiquette or good manners in a minefield - page 4
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
I feel like I'm doing it wrong. I tried to model this process and play with parameters. I don't get such nice surfaces. I get planes and linear dependencies. I do not understand anything. Out of despair, I tried to simulate "optimal f" from Vince's book - exactly the same result. It turns out not at all like it says in the book. Took his crazy game from Vince's book, when he wins $2 and loses $1 - very good and fair game. Anyway, still trying to get the same as Vince's.
I built surfaces in Mathcad. I used an expression for the rate of return per unit of time (the unit is the characteristic time of price change by one pip):
Because the obtained value is small, about 10^-4 (it's not a mistake, because it's a rate of return for the time of price change of one pip), the result was multiplied by 10^5 and I obtained these surfaces, which you can see on the fig.
S - 10^4
dS - changed in the range of 1-100
L ranges from 1-300
p - changed in range 0-0.03
Spread - changed from 1-10
Should converge
Great. Is this methodology for calculating leverage applicable only to NS? Or does it work for any MTS?
The NS has absolutely nothing to do with it. It has nothing to do with anyone. The only requirement is to know the probability of the prediction.
which shoulder is the tester using?
The one that was on the last one logged in to the terminal
What leverage is the tester using? And how can you change it?
You have to re-login to another account/ different brokerage company.
That's the thing, some people can't calculate the prediction ;0) . I've got a system for catching extremums of the day. Now I'm trying to improve it, it's unstable. So, after reading this I've been thinking for the first time - what kind of leverage does the tester use? And how can I change it?
stLot *Lot=K*Lever - connects open position size (Lot) with leverage and deposit size (see the very first post of the topic). Knowing the size of the open position, you can find the leverage used, and vice versa. And another thing, the tester doesn't use any leverage, it's your TS's prerogative, so the answer like "...switch to another account / other brokerage company..." does not really correspond to reality. Well, or maybe I'm misunderstanding something.
As for calculating the prediction - 1/2+p for "correct" transactions, you can get this number in the following way:
We run the TS so that there would be a few hundred transactions (the more, the more reliable the estimate) with a minimum lot. Further, we send tester's report to a math application (for example, Excel) and convert each bribe (expressed in $) to points by adding to each one the value of brokerage company's commission in points. Now, count how many transactions are in + and divide it by the total number of transactions. If the result is <0, we unambiguously "reverse" the TS. If now everything is OK, then we subtract 1/2 from the obtained ratio, this is the required value of p.
stLot *Lot=K*Lever - connects open position size (Lot) with leverage and deposit size (see the very first post of the topic). Knowing the size of the open position, you can find the leverage used, and vice versa. And another thing, the tester doesn't use any leverage, it's your TS's prerogative, so the answer like "...switch to another account / other brokerage company..." does not really correspond to reality. Well, or maybe I'm misunderstanding something.
As for calculating the prediction - 1/2+p for "correct" transactions, you can get this number in the following way:
We run the TS so that there would be a few hundred transactions (the more, the more reliable the estimate) with a minimum lot. Further, we send tester's report to a math application (for example, Excel) and convert each bribe (expressed in $) to points by adding to each one the value of brokerage company's commission in points. Now, count how many transactions are in + and divide it by the total number of transactions. If the result is <0, we unambiguously "reverse" the TS. If now everything is OK, then we subtract 1/2 from the obtained ratio, this is the required value of p.
inaccurate, missed it.
.... got in + and divide the resulting number by the total number of transactions-1/2. It is clear though. Except we do not get the probability, but the frequency of positive transactions. Under certain assumptions it can be considered a probability.
It seems to me that the probability of a deal is not constant and may vary from deal to deal. You have to be able to calculate it (probability) before each trade. But it is much more difficult.
Of course, Sergey, I agree with you. It would be nice to be able to estimate parameter p locally, in real time, but I'm afraid this task has the same order of difficulty as constructing a non lagging indicator, with all its consequences...
By the way, if we draw in double logarithmic scale the dependence of the optimal size of the take on the optimal size of the trading leverage, here's what we get:
Parameter p is included here in an implicit form. The accuracy of prediction of p increases from left to right for two fixed values of brokerage companies' commission - 2 and 8 points (red and blue lines correspondingly). In the calculations we used simplified expressions for the dependences dS(p) and Lever(p) assuming that p is small. The difference from the exact solution did not exceed 10% in the whole presented range.