Software beats humans at Commodities trading

didn't know you were based in europe, thus your cost concerns make sense. There are other options - like hooking up to one of IB's european offices (zug, switzerland), or using a VPN. Perhaps IB will allow internet CTCI in the future but for now you'll have to go elsewhere for an internet ctci connection.
 
It is foolish to believe you will never be bested by a machine. Soon machines will trade circles around you. You won't stand a chance against an brain of semiconductors that crunches numbers a zillion times faster than your inefficient grey blob. Electrons move down copper far far faster than oozing neurotransmitters. While you sit there drawing your trendlines on a CSCO or INTC chart, a machine will have scanned, analyzed, and executed dozens of trades!

Oh pleeease...

LOL.

tradeRX
 
tradeRX - one thing I'm curious about is how a computer program will interpret NEWS. Will the program simply react to increased/unexpected trading volume or will it be a little proactive and make a move first ? If a computer program is able to interpret good/bad NEWS combined with a phenomenal number crunching ability - YIKES !! How will folks trade in that kind of environment ?
 
wirelessbull,

The short answer is...

Neural Networks & Genetic Algorithms.

Interpretation of news is very subjective and fraught with error. However...

Imagine sensing EVERY aspect of the markets in real time down to the finest nuance...

THEN REACTING IN A SPLIT SECOND to the subtle symmetry that emerges!

Evolution did not "build" the human brain for stock trading. But imagine a machine built for just this purpose...with the light speed processing and depth for assessment of HUGE numbers of variables!

WOW! That is beautiful!

How can anyone think they could compete?

tradeRX
 
By the time that super computer is out, machines would have replaced humans in 99% of the jobs out there. Not to mention if there is a goose that lays the golden egg used in mass there will be people who can find patterns and exploit it.

Any moron can put in 50K and watch it gets traded into 1 million without doing nothing? Won't happen in this life time . . .
 
Hitman,

Perhaps.

However, in this lifetime....

We have come a long way since 1975. What wonders will we have in our hands in another 30 years? 50?

Do this. Take a few minutes and reflect on your very simplistic market methodology. You literally take into account only a few of the variables that affect your potential trade before hitting the buy button. Then, after the trade is in place, you never "see" most of what is impacting your result. It slips by unnoticed and unappreciated.

The computer has made us better traders. With more processing power, we get even better!

tradeRX
 
I don't see how computers made us better traders, other than it presented the market in a better way, we still made all the decisions just like the day we made hand drawn graphs.

You forgot that in trading simpler is better, additional variables doesn't neccessarily produce better results, and this has been confirmed in many many backtests.

Market is also constantly changing, and machines must be updated, last time I checked we haven't had any true artificial intelligence that can learn by itself yet, ultimately, traders program the machines, and the machines are as good as the trading systems behind them.

Last but not least, as I said, by the time the machine can outperform traders en mass, most of the professions would have been long gone. I personally just can't see it happen any time soon, unless we are doing a "Matrix" . . .
 
Originally posted by Hitman
You forgot that in trading simpler is better, additional variables doesn't neccessarily produce better result
More...
i second that. our experience shows that each introduced variable comes at a cost. throwing a large number of variables at a program just makes it "see ghosts": more and more random patterns will be interpreted as "real".

last time I checked we haven't had any true artificial intelligence that can learn by itself yet
More...
well, that depends on your definition of "learning". while i can say that neural nets and genetic algorithms are nothing more than function approximators with a flashy name, what they do is usually considered "learning". of course, you problably meant that they do rely on humans to hand-pick the (type of) training data for them -- that's very true.

also, re news: it is already possible to do an automated keyword scan in news texts, and have a program to analyze the correlation between various keyword combinations and stock price. however, i think the problem is that once the news is available as written text, a lot of it is priced in already.

having said all the above, i'm also a believer that _in_general_ machines are quite far from overthrowing humans, because the simple fact that they lack the ability to _comprehend_. keywords to search for more info on this: penrose, hameroff.

- jaan
 
TradeRX
Read page 107 of "The New Market Wizards" until you understand that computing power is irrelevant to trading success. If you do not, go take a logic class.
The whole thread Robot vs. Human is a f%&@#$* non-sense.
 
You must log in or register to reply here.
Back
Top