Public discussion of the formula for calculating the cost of resources in the MQL5 Cloud Network - page 17
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
There are four cores available. Presumably willing to charge from $0.08 per hour per core.
At the same time, 8 cents is not the price for which I would specifically turn the machine on or leave it on overnight. Rather, the price is just for surrendering cores in the background while you work, and you'd probably get three out of four. The computer is on for at least 12 hours a day.
Then from 2 cents an hour (that's more accurate)... :)
In this case, these are my current calculations
Costs have not yet been calculated.For the seller this is a super profitable business. This is more than 100% net profit per year.
Firstly, why is it a super profitable business? Secondly, why is it 100% a year (unless you're talking about buying a PC for like $300)? Thirdly, would you personally rent out your whole PC for one year for $0.01 per core?
Let's first define what we're dealing with, so:
1. There is an average PC with 4 cores and 4 Gb of RAM. Let's say it costs $1000 together with all the extras.
2. Suppose a person has a real cent account for $ 1000 and mechanical trading tools or semi-automatic devices, which are very likely to bring 100-200% per annum by managing this set.
Thus, investing $2000, a person in a good deal a year may well recoup the amount invested. Even if he manages to earn only 100%, consider that the PC was bought at almost 100% discount.
3. Now suppose he wants to optimize his PC by leasing 2 cores out of 4 thinking it will not harm his performance too much. Suppose he rents cores around the clock at $ 0.02 per core.
So if everything will be good and the whole year will be in demand, this man will earn about 346 $
4. So investing $ 2,000 in a conservative trade + a bit of luck this person can get about $ 1346. Which is quite good when you consider the cost of the PC and the amount of internet and electricity costs.
5. Now let's imagine that all 4 cores are rented out, with no trading. That is, a person has invested only $ 1000 for the purchase of PCs
This result is not ideal, but it is good (assuming that the cost of the Internet and electricity people are satisfied).
Now imagine how profitable to rent the core at $ 0.01. Surely no one who knows the other prospects in his right mind would not rent their resources for such a price (as for someone else's resources, such as resources firms - well, the profits should share, and no one has canceled the cost).
If such a price is set, there will be many times more people willing to offer their resources than users of those resources. Naturally, the law of supply and demand will cause the price to drop over time.
As for supply and demand, let's think logically:
1. Let's assume that an average optimisation run lasts 20 minutes.
2. The maximum effective time of optimization is outside working hours, mostly weekends. I.e. this is the time when most sane people will try to perform the fullest optimization task.
3. A simple calculations shows that with different PC configurations we have about the following
Let's assume that an average PC has 2 cores, so the critical value is 288 passes. If the number of passes is higher, either optimization won't be performed in time over the weekend or you will have to sacrifice something (both variants are unacceptable from a normal viewpoint).
4. Now let's assume the following:
a) Average number of passes per test per Expert Advisor = 1000 (let's assume a not very complex Expert Advisor). Hence, it turns out that it takes approximately 20000 minutes for full optimization (which is almost 14 days, or 13.8 days to be exact).
b) Suppose we have two cores of our own, thus the optimization will take about a week. More precisely, about 6.9 days.
c) - let's make a decision to lease cores (for calculation we will take 100). As a result the optimization will not take more than 3.5 hours. In the end the trader will pay about 7 bakinets for such optimization.
Of course, there will be much more cores than 100 in the cloud. But the number of passes will surely exceed 1000, and many of them will last for more than 20 minutes.
PS
And now let's imagine the number of people who want to optimize exactly at weekend (and I understand them quite well). So how much will it cost per core at that time?
In my opinion to unload a cloud for comfort state (when supply obviously exceeds demand) the number of cores at least by an order of magnitude (or even two) higher than the number of people who want to use them.
Interesting:
Let's assume that a person has [...] means of mechanical trading or semi-automatic machines which with high probability will bring 100-200% interest per annum by managing this set.
[...]
Now let's imagine that all 4 cores are rented out and no trade is conducted.
A person who is able to "very likely" make 100-200% per annum, would not trade in a cent account and rent kernels.
And anyway, how does trading relate to the Cloud Network? [...]
A person who knows how to make 100-200% p.a. "with a high probability" is not going to trade in a cent account and rent out cores.
In fact, how does trading relate to Cloud Network? [...]
1. I think that on MICEX or NYSE $1000 is not enough even for a warm-up. And for you?
3. How many traders do you think that they make 200% p.a. on Forex?
3) OK. Let us assume that the trader receives stable 10% per week (2% per day). Thus, the number of 100 traders will be 100, of which 10-20 spend on optimization. Isn't it an option?
4. Who manages a large deposit is unlikely to stumble in the difference between 1 cent and 2. If the result still pays off (and we assume that it pays off), then personally I do not see a big difference.
5. Of course trading is not directly related to Cloud Network, but indirectly it is. We are not all looking for Martians there. :)
Yeah, one more thing, to the question of super profitability - What is super profitable renting out nukes or Forex trading?
1. I don't think that on, say, the MICEX or NYSE, $1,000 is even enough for a warm-up. What about you?
2. How many traders do you think, how many of them make 200% per annum on Forex?
3) OK. Let us assume that the trader receives stable 10% per week (2% per day). Thus, there are 100 Bakuites, of which let's say 10-20 spend on optimization. Isn't it an option?
4. Who manages a large deposit is unlikely to stumble in the difference between 1 cent and 2. If the result still pays off (and we assume that it pays off), then personally I do not see a big difference.
5. Of course trading is not directly related to Cloud Network, but indirectly it is. We are not all looking for Martians there. :)
Yeah, one more thing, to the question of super profitability - What is super profitable renting out nukes or Forex trading?
Vladimir, are you serious?
Of course, I've tried not to express my personal attitude to you, and just noticed that the post was off-topic. But to be honest, I hoped that I wouldn't get an answer.
A trader who makes a stable 50-100% a year will find investors for such sums, that is enough for any market.
And you can start demonstrating your stability with a demo (continue on micro).
And this is all a load of rubbish. Thinking about "how much a trader earns" is completely irrelevant to the topic of this thread.
Vladimir, are you serious?
No less seriously than those who claim that $0.02 per hour is super profitable. Also as serious as those who believe that $0.01 is the optimal price.
As for stability - it does come to an end, and those who once earned an average of 2% a day can lose everything.
And so as a conclusion - I'm not saying that the price of a cent per hour will not bude, but that's who will rent for such a price?
PS
There Mischek tried recently to make a table wishing to lease cores. And where are they all?
Probably not a payback, but a breakeven, plus a small profit. Plus the ability to quickly optimise your system when required, giving up, in due course, the idle part of your resources and spending what you have accumulated.
Imho, a more serious profit is unlikely to be obtained.
So still - why don't you ask how much the consumer is prepared to pay??? I, for example, am willing to give up a dozen cores for 2 quid an hour. Dig in :)
I was being sarcastic, in case anyone didn't get it.
Still, the number of testers and the number of those who want to rent, in my IMHO, will differ at least by 2 orders of magnitude. In these circumstances, no profit or even approximate payback is out of the question, because a good part of the network will be idle or make one run in a couple of hours.
Also, due to the nature of genetic testing, my 18 agents are only 2 to 5 times slower to optimise than Cloud (on my EA). If this time is within 24 hours, I'm willing to wait.
Ready to pay $ 1 - $ 5 per day for network rent (when I need it).
Let those willing to buy resources, as the point of view of those willing to sell is clear - self-sufficiency (although, why should - as if resources are not spent on anything else, but only on Cloud).
P. S. I'm willing to give up my resources for free, and it would be great if the developers would do it too