Shall I get some tips about useing Machine Learning on trading?

[algofy]

I will try to be really brief here.

If you know programming, go and generate some random numbers (example using rand() in stdlib.h in C). If you are completely math & programming idiot (as most of the ET here are although most of them will strongly denial this is the case). You can download the free random number generator in your Apple/Android Apps store and it will do the same job.


Plot those number in the chart. You will be surprise, they look exactly the same as your normal daily price chart in your indices, forex, commodities or whatever things you have in your day trading chart. The chart has support, resistance, trend line, channel, head and shoulder and the whole BS price actions features that the current trading industry try to sell.

You can even show this chart in ET and ask the "ET Expert" for opinion of the next price move, for the sake of fun of the year end celebration.

Machine learning on stock market which is random (especially the short term day trading) ? Give me a break.

Ahhhhh it all makes sense now the trouble you have had with me and the superiority and name calling. You're an acedemic random walker type. You believe everything is random therefore no edge even exsists for me to exploit.

 

[O(1)]

LOL. In this case you can continue using machining learning to interpret the random price, looking at your day trading chart and continue to make money as you claimed.

do your accusations ever end? I don't recall ever saying that I use machine learning to interpret day trading charts and claiming to make money doing it. I simply said that your experiment proves nothing. Do you still think it does?

 

[algofy]

do your accusations ever end? I don't recall ever saying that I use machine learning to interpret day trading charts and claiming to make money doing it. I simply said that your experiment proves nothing. Do you still think it does?

Trust me, I know his type, you won't ever win anything with him as he always has to be right and technically superior.

 

[Tim Smith]

What is a holy grail when it comes to trading?

There isn't. That's the point ?

Why do you think all the algo funds employ rooms full of PhDs ? Its not because "they can", its because they need to constantly adapt and refine their algorithms in order to retain the tiny miniscule little edge in the market.

If you're playing with machine learning and algo-trading, then you're going to be fighting those PhDs. So if you think you can "write and forget" some magic algorithm you're much mistaken.

 

[galvinlee888]

Ahhhhh it all makes sense now the trouble you have had with me and the superiority and name calling. You're an acedemic random walker type. You believe everything is random therefore no edge even exsists for me to exploit.

There is edge in market.

The point is you can't find edge in charting, you have to do better than that. I already give you a lot of hints in this regards in my previous posts. I will not go any more further to disclosure this.

 

[galvinlee888]

There isn't. That's the point ?

Why do you think all the algo funds employ rooms full of PhDs ? Its not because "they can", its because they need to constantly adapt and refine their algorithms in order to retain the tiny miniscule little edge in the market.

If you're playing with machine learning and algo-trading, then you're going to be fighting those PhDs. So if you think you can "write and forget" some magic algorithm you're much mistaken.

Not only a room full of PhDs, but also the technology and equipment that enable those algo

 

[Tim Smith]

Not only a room full of PhDs, but also the technology and equipment that enable those algo

And exceedingly expensive datafeeds delivered over equally expensive fibre connections.

 

[lovethetrade]

Ahhhhh it all makes sense now the trouble you have had with me and the superiority and name calling. You're an acedemic random walker type. You believe everything is random therefore no edge even exsists for me to exploit.

The market is not random, I know that for a fact.

There isn't. That's the point ?

If you're going to make claims that something doesn't exist then surely you need to know what that something is...

 

[algofy]

The market is not random, I know that for a fact.



If you're going to make claims that something doesn't exist then surely you need to know what that something is...

Agree, the random walkers are the worst.

 

[wintergasp]

Do you ever generate any random number and do this type of experiment before? The number will not go to zero in the short team if you really know what the random number is, and how to simulate the market operations

Yes this is basic quant work, I have interns who do this better than coffee.

I'm not saying that random numbers will go to 0, but the strategy of randomly buying and selling on a **real** market data set will not output any positive performance that passes statistical significance test.

This is because real market data is not completely random. Try to randomly buy or sell on a random data and you should have a mean performance of 0.