Page 1 of 1

Automated trading V. Live Human trading

Posted: Wed Nov 15, 2006 12:03 am
by 4x=0
I've read that automated trading is inferior to direct human trading because the market is psychological in nature; Programmed computers cannot adapt to the complexity of human reaction, and thus fail if not constantly modified. Naturally a human trader has the best chance at reacting favorably by quickly interpreting what s/he sees. If automated trading is profitable, then there IS a holy-grail no?

On the other hand, coding your strategy rules into a computer seems like you would only be eliminating the most poisonous factor in trading: emotion.

If you have any experience with automated trading and human based trading please post with what you have learned.

:smt032 :-({|= :smt022 :smt069 :smt071 :smt063 :smt066 :smt065 :smt024

Posted: Mon Nov 20, 2006 4:42 pm
by 4x=0
Well so far it appears that Live-Human-Trading has 60% of the vote.

Good to know. Good to know..

Although, the most intellegent traders that utilize robot-trading will probably not vote or participate in this thread.

:smt036

Posted: Mon Nov 20, 2006 10:05 pm
by michal.kreslik
The best blend is an automated, constantly maintained/monitored trading of the human-devised set of rules.

A human should not be allowed to place a single trade based on "hunches". A human might develop a set of rules based on hunches, though. Then, if and only if this set of rules stands the backtest, it should be recast into the human-independent system.

Michal

Posted: Wed Nov 22, 2006 6:53 pm
by 4x=0
hmm. Sounds to me that you are just adding another layer between human decision making and trade execution. Typing away at all this code to tell your computer to do what you can already do. It's like playing a guitar, but with one hand programming on a computer to tell a robot how to finger the chords. If human-independent trading does offer some benefit, surely it would be something revolutionary for traders. However the benefit or benefits of robot-trading have not become clear to me.

I would also like to know why hunches should not be a basis for trading, yet hunches are a viable basis for the strategy that is to be programmed into a robot. If these hunches, which are dirived from reading indicators (even if only price action and bar patterns) have worked for a trader in the past, what is wrong with a human trading based on their signals, without the robot. How does coding the hunch rules into a computer benefit the trader. I will do my best to form three reasons that human-trading is as good as or better than robot-trading, and I would expect the proponent of robo-trading to do the same.

Pro Human Trading Rational:

1. The market is represented by numbers, yet psychological in nature, and thus constant modification of computer code to keep current with the human traders' irrationality is unnecessary, considering you are a human already.

2. Humans are still far more complex than robots. Some successfull strategies are incredebly simple, yet extremely difficult, if not impossible, to convert into a language a robot does understand. Just as a robot cannot task something so simple such as take drink from a glass of water, it cannot trade many of the simple(complex for a robot) intracacies that a human does with ease. Example; see: TRO pinbars.

3. Without human trading, robot trading is not possible.

:smt024 :smt032 :smt063

Posted: Wed Nov 22, 2006 8:41 pm
by michal.kreslik
4x,

I'm focusing on trading and developing my systems, so the "discretionary vs. objective trading" flamewars are no longer in the center of my interest :)

"Objective" means repeatable based on the same input data.

Objective trading works for me and that's a sufficient reason for me to employ it. Indeed, I gain nothing by urging anyone to use this or that particular style of trading. Everyone should decide on his or her particular style of trading by himself.

Just a brief reaction to your post:

>It's like playing a guitar, but with one hand programming on a computer to
> tell a robot how to finger the chords.

Not at all! It's beneficial for the guitar player to be emotional. But for a trader, it's disastrous.

> I would also like to know why hunches should not be a basis for trading,
> yet hunches are a viable basis for the strategy that is to be programmed
> into a robot.

It's clear why: at the beginning of every new discovery or invention, there is a human with a bit of luck, a bit of hunches and lots of work. The nuclear power plant is not built upon hunches - there's a lot of work, testing, calculating and so on before one is engineered and built, even if it's clear that at the very beginning, there was only an idea - a hunch, if you please - that the atomic nucleus might be the source of energy. It's the same with trading systems.


Michal

Posted: Mon Dec 11, 2006 1:39 am
by 4x=0
Mr. Kreslik,

By abandoning your persuit of the emotionless trader robot, you are opening yourself to the posibilities of taking the emotion out of your trading without comprimising the fluid perception of ones mind.

The physicists and scientists that develop nuclear reactors are doing so based on the rules of physics. Trading is not based on physics but based on human reactions, which are based on an ever changing environment. Therefore a system directly perceiving the market still and always will prevail.

4x=0

Posted: Fri Dec 15, 2006 12:03 am
by Han
If you lack confidence in the probability of your developing a profitable "black box" your probabilities are probably right. With such apprehensions the impossibility that you would succeed were you to start down the long and arduous trail of developing an autotrading system is even more probable. I'm fool enough to try, believing there are things computers see statistically that traders can't. I also believe the trend toward computer assisted trading is relentlessly accelerating market activity. In such a cloud of activity an IFR vs. VFR trader's pilot license may be the difference between eventually crashing or not in the vertigo of program-driven trading, which CME says accounts for something over half of total volume. ](*,)

Posted: Fri Dec 15, 2006 3:51 am
by 4x=0
The probability of my developing any black-box system is 0% to be sure.

Your comment about a 'relentlessly accelerating' trend towards computer assisted trading is interesting, however questionable. Please provide a link to CMEs data.

With millions of independent human-independent trading systems executing trades based on information generated from millions of other similar systems, the robot is no better off than the superior human trader, concerning the vertigo.

Because of selfishness and greed, two human qualities, a purely robotic market would be disastrous.

Posted: Mon Dec 18, 2006 5:16 pm
by eudamonia
There are advantages and disadvantages to both methodologies. Just pick the method that you are most comfortable with and go with it.

I'm comfortable with both black box (no human input) and grey box (minor amount of human input).

Find what works best for you.

Edward