Machine learning in trading: theory, models, practice and algo-trading - page 1668
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
The market is 95-99% random, but even 1% determinism can be enough in naughty hands.
MO makes sense only for to some extent "smooth" functions, continuous, in statistical sense, it is necessary to have at least some gradient, if there is no gradient at all, as in PRNG, then MO will not help, we need some heuristics or knowledge about generating functions.
Almost all here in the first place use not enough data (for example one row of clocks for a year), not relevant features and not correctly constructed targets, and then go in search, super-duper classifier, which miraculously makes a lambargini out of garbage. The same thing as with indicators, only there are more settings, it is IMHO some kind of gaming, like you pass all sorts of levels, solve puzzles more and more interesting, no end in sight... But it's not algotrading, it's a fad.
Just like with indices, Momentum, SMA and EMA and a few standard window stats are enough, in the same way, in the context of algotrading the forest or boosting is enough, and if it is not enough, then it is needed elsewhere, not in the MO. But all this is like peas on a pumpkin.
Searched a little, the requirement is only for continuity. A couple of references on optimization
https://www.coursera.org/lecture/mathematics-and-python/optimizatsiia-nieghladkikh-funktsii-0ZFvR
https://tsumathem.elpub.ru/jour/article/download/170/171
Yesterday was kind of a weird Thursday, like a weird Thursday :-( it only happens in real markets.
Show me how you lost equi...
Show me how you've leaked, equi...
Yesterday was a weird Thursday, really weird.
It's okay, what are you risking? Probably a couple hundred bucks at most, and I'm bravely holding a short on the snp500 from 2700 to 2000, at a grand for a hundred points. The way I see it, in the next couple of weeks, there should be a pestilence in the USA like in Italy, only stronger and everything will collapse in spite of the trillions of printed bucks. And goddamn if I give in and close above 2000!
It's okay, what are you risking? Probably a couple hundred bucks at most, and I'm bravely holding a short on the snp500 from 2700 to 2000, at a grand for a hundred points. The way I see it, in the next couple of weeks, there should be a pestilence in the USA like in Italy, only stronger and everything will collapse in spite of the trillions of printed bucks. And damn me if I let my guard down and close above 2,000!
Machine learning on genetic algorithms in neural networks in action))))
It's okay, what are you risking? Probably a couple hundred bucks at most, and I'm bravely holding a short on the snp500 from 2700 to 2000, at a grand for a hundred points. The way I see it, in the next couple of weeks, there should be a pestilence in the USA like in Italy, only stronger and everything will collapse in spite of the trillions of printed bucks. And I'll be damned if I'm gonna slack off and close above 2,000!
You have to be an expert on the pestilence and its consequences to predict the pestilence... not me said....
Here are some interesting ideas in my head
I made a simple graph of the price [i]-price[i-1] increments and simplified it to the limit: if the price increased - +1 if it decreased - -1
I built the cumulative sum of these elementary increments and got the following chart
My question is for the experts, neural network specialists, econometricians and other evil spirits)
Can such a graph be predicted? It is much simpler than the price by all parameters and the most important thing is that it has additive properties
you invented Renko / Range Bars :)
https://www.mql5.com/ru/forum/334178
tradingview[dot]com/chart/SPY/29Tnfldo-SPY-daily-Renko-chart-LR/
you invented Renko / Range Bars :)
https://www.mql5.com/ru/forum/334178
tradingview[dot]com/chart/SPY/29Tnfldo-SPY-daily-Renko-chart-LR/
I didn't invent anything, I asked a question.)