Machine learning in trading: theory, models, practice and algo-trading - page 313
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
It's not logical, they show them and take them to DU... And those who can not take them elsewhere :)
They show, but you haven't seen? But they show? But you haven't seen?
Showing secretly from you? But shown? But you have not seen?
Shown but you haven't seen? But they're showing? But you haven't seen?
Are they showing it secretly from you? But shown? But you haven't seen?
I'm too old for this d... If I had I would have seen it, and if I had I would have seen it, I'm sure it would have been shown, I'm 100% sure.
That's the thing, the leader.
How the people who are lower in the ranking deal with sabotage is very well written here.
Well, yes, looked at his profile on google, it seems he knows the little guy, but this particular report about the contest from Two Sigma, not about anything, they have compiled two chips into one, shoved in XGB and did not understand how the result is formed. As a post on the forum will do, as for the report - dz, but maybe I'm behind the times and people now gather preconference for the sake of a message in a sentence or two, in general, I will not complain, so that young people do not start to cock...
"The ones who are lower in the rankings" are more interesting.
I saw his profile on Google, he seems to be good at it, but this report about Two Sigma contest is about nothing, they have two chips in one, put them in XGB and have no idea how the result is formed. As a post on the forum will do, as for the report - dz, but maybe I'm behind the times and people now gather preconference for the sake of a message in a sentence or two, in general, I will not complain, so that young people do not start cocking ...
In that video, it is not at all interesting how the rapporteur's team solved the problem set by the $40 billion hedge fund. What was interesting was the problem itself, the reasons it was not set, the kernel version of the competition and the results of the CONCURRENT, not the rapporteur's team. And it was the results that made it clear that the first places were distributed randomly. And the hedge fund just screwed up if it hoped to squeeze something out of its data by giving it to the strongest MO-teams, who have eaten their cats and dogs on other MO-tasks.
I'm surprised they didn't see that. Apparently you were skeptical of this video from the first seconds, so it was hard to include an analysis of the words spoken.
If the hedge fund wanted to confirm its hypothesis that you can't squeeze shit out of the D1 data, then it did an excellent job.
There is only one good piece in the video that shows how quickly and easily the input data was deanonymized by building stdev(timestamps).
And the speaker himself was interesting not only for his acknowledged expertise in MO, but also for his practical experience in HFT. That is, his reserches work on HFT, but on larger timestamps resemble a lottery. And the man knows how to research price BPs. HFT is a well-known company.
And the speaker himself was interesting not only for his acknowledged experience in MO, but also for his practical experience in HFT. That is, his reserches work on HFT, but on larger timestamps resemble a lottery. And the man knows how to research price BPs. HFT is a well known company.
what kind of office, where to look at what? maybe there are other people there
I recommend watching a short video of HFT-quantum, which is one of the world leaders in subj.
Of course they will be sound.The luck factor, if not the main one, is certainly essential in this case.
It's not clear they were able to make any money at all? Or again fairy tales for grown-ups and all the same eyewash, with abstruse speeches.
It is not clear whether they were able to make money? Or again, it's a fairy tale for grown-ups and a lot of whitewashing, with abstruse speeches.
Go to the link - we see the name of the company. Googling the name, we see some of its results. I see that some people are getting too lazy for their own good. You can't push three buttons.
"Smart talk" and "adult" are complexes, apparently.
ZS I regretted that I posted the video. ''''' life is like that.
Go to the link - we see the name of the company. Googling the name, we see some of its results. I see that some people are getting too lazy for their own good. You can't push three buttons.
"Smart talk" and "adult" are complexes, apparently.
ZS I regretted posting the video. ''''' life is like that.
Maybe I'm wrong. But, haven't found where they made money in trading on this. They may very well have good methods for recognition, but life has taught me that in seminars, they sell what usually doesn't work well.
Well, here are guys really making money using MO, quantitative methods.https://en.wikipedia.org/wiki/Renaissance_Technologies
In that video, it is not at all interesting how the rapporteur's team solved the problem set by the $40 billion hedge fund. What was interesting was the problem itself, the reasons it was not put out, the kernel version of the contest and the results of the CONCURRENT, not the rapporteur's team. And it was the results that made it clear that the first places were distributed randomly. And the hedge fund just screwed up if it hoped to squeeze something out of its data by giving it to the strongest MO-teams, who have eaten their cats and dogs on other MO-tasks.
I'm surprised they didn't see that. Apparently you were skeptical of this video from the first seconds, so it was hard to include an analysis of the words spoken.
If the hedge fund wanted to confirm its hypothesis that you can't squeeze shit out of the D1 data, then it did an excellent job.
There is only one good piece in the video that shows how quickly and easily the input data was deanonymized by building stdev(timestamps).
And the speaker himself was interesting not only for his acknowledged expertise in MO, but also for his practical experience in HFT. That is, his reserches work on HFT, but on larger timestamps resemble a lottery. And the man knows how to research price BPs. HFT is a well-known company.
Yes, you are right about the unaccounted for market information on D1.
Well, the reason for the competition for the fund, most likely - headhunting, this area is quite specific, probably look for those who "ate their dogs and cats on other MO tasks" or made their own MO-algotrader ecosystem, probably more effective than economics graduates take and all sorts of traders handicrafts, who got lucky a couple times at the last candor.
And it was clear from the results that the first places were distributed randomly. And the hedge fund simply screwed up if it hoped to squeeze something out of its data by giving it to the strongest MO teams, who had eaten their way through other MO tasks.
Let me rephrase the assignment.
Give us the code of your model to trade on D1, a trained model is also not allowed, your code should run and train the model on our server, we need to know and understand the entire strategy.
As a result, we will pay you a little for your perfect strategy, and we will go on taking profits without you.
It's a trap.
All the adequate people got away from this, only the "let's teach gbm optimizing random grain, and maybe it will work" team is left.
There is an adequate hedge fund built on the same principle - numer.ai
they give some of their secret predictors, obtained by their specialists and paid subscriptions, and encrypt them so no one guessed. IO experts in turn do not send them their model, but only a forecast, the hedge fund further uses them for forecasts on new data. In the end, the hedge fund does not post its secret predictors, and MO experts do not post their secret models. Everybody's happy.
Except their payouts are idiotic, the first place gets twice as much money as the second, although the real profitability of their models may differ by 0.0000001%. I, for example, once in the top 100 lasted 36 hours out of 168, and got a pittance of $2, even spent electricity is not recouped.