Algorithm Optimisation Championship. - page 119
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
Deal with the traction, you have a problem with it. :)
I think the priorities in evaluating AO are wrong.
In the field of trading AO is a tool used for approximate (not absolutely accurate) calculation of strategy profitability, and the acceptable error can be within a dollar.
However, if you strive for maximum accuracy (which by the way has the same far-fetched meaning as the belief that the values found in the future will bring the mountain of gold), it is possible to excessively overuse computer resources and your time.
Practice requires us to be rational and use working tools effectively.
Yesterday I decided to use Mersen's vortex for my algorithm. Borrowed the one rewritten from C++ by Urain. Urain.
A tip to the participants - I recommend to use this GSF, which will increase the accuracy of your algorithm. Maybe such a step will forgive one of my sins, the generator in the attachment.
No. Nothing is going to change. I've already explained everything I can.
The principle is everything. Practice is nothing!
Don't you want to adopt that slogan? :)
Deal with the traction, you have a problem with it. :)
It's not that you can't, you HAVE to.
You have a competition, not an academic problem. At a competition there should be no uncertainties, no speculation and no ambiguities of any kind. The obtained results of the participants' algorithms shall be clearly (unambiguously) interpreted for the task at hand, i.e. they shall have an unambiguous answer to the question: "Will an extremum be found? Only two answers are either "yes" or "no". The notion of range is introduced in order to treat participants' results in this way. If a participant's result falls into the sought range, it means that his or her algorithm has fulfilled the task. The size of the range depends on the optimization step. For example, if the optimization step is 0.01, then the values received by the participants must differ from each other by the second decimal point. So, if the obtained result coincides with the sought result up to the 1st decimal place, then this result is considered TRUE and the owner of this algorithm is allowed to select the "Winner". With this formulation, there will be no ambiguities, no reasoning about the "goodness" of algorithms, etc. Moreover, by this approach you cut off any negativity that may arise.
You are deluded and talking about things you don't understand. You're also saying something about pulling...
It's simple and straightforward. Everyone is given the same number of calls to the FF, if the search is maximal, then whoever has a bigger result wins. A function may be unknown at all, there are functions with noise and the value of its maximum is unknown in advance.
The step of changing the parameters can't coincide with the step of changing the result, what was your maths class at school? And you're still talking about traction?
The answer to the question - which algorithm performed the optimization better.
The principle is everything. Practice is nothing!
Don't you want to adopt this slogan? :)
Ochochkie, new rule again....:)
When I first took part in the discussion with the proposal to take into account the accuracy of the calculation, the criterion for selecting a winner was the LOWEST number of references to the FF.
I'll revert back to being a spectator. The rules change like gloves on one hand here!
Ochochkie, new rule again....:)
When I first took part in the discussion with the proposal to take into account the accuracy of the calculation, the criterion for selecting a winner was one - the LOWEST number of references to the FF.
I'll revert to the category of spectators. The rules change like the gloves on one hand here!
So it's you making up your own rules as you go along.
There are two criteria for determining winners: accuracy and number of FF calls, accuracy is 3 times more important. The maximum score is fixed, a contestant can use fewer dials than the maximum, thus increasing his standing in the table, and there is a second criterion for that - dials.
I thought that in a day Vasya will come back in the fence and will ask the same questions again, but no - all the old faces!
So you're making up your own rules as you go along.
There are two criteria for determining the winners: accuracy and the number of FF calls, accuracy is 3 times more important. The maximum number of accesses to the FF is fixed, a contestant may use less accesses than the maximum allowed, thus increasing his position in the table, and for this there is exactly the second criterion - accesses to the FF.
I thought that in a day Vasya in the back will come and ask the same questions again, but no - all the old faces!
We need to be clearer. Let's say 100 calls are allowed, the maximum is 10. One contestant gets to 8 for 100 challenges, the other gets to 5 for 10 challenges and stops. It's not like the second one will be the winner.
There needs to be clear, well thought out rules, described fully and in simple language, so that there are no misunderstandings or arguments amongst the participants.
They also need to be accompanied by code examples, also fully thought out and commented upon in great detail.
All this should be done nicely, clearly, in one colourful box with a bow.