Discretionary Trading vs Automated Trading

[TheBramble 10]

why is it that the majority of successful intraday traders are discretionary traders?

Becuase they are skilled in reading the markets and not necessarly skilled in codifying what it is they're doing when they analyze the markets. Most would not even know what they are doing on a conscious level.

 

To break down and model what somebody else does is tough enough. To do it for yourself on yourself, while you're doing it is almost impossible.

 

There seem to be two distinct camps: Those than can trade and those that can write great code that doesn't trade well. The two cannot communicate effectively.

[Horus 10]

Anyone here trade fully automated? Or have you tried to trade fully automated in the past?

 

What kind of issue did you face?

 

Any day trading system traders here? What kind of commission do you pay? How many trades do you make a day? Whats the slippage on each trade?

[syswizard 10]

In reply to syswizard: "emotional and psychological aspects of market behavior" are presumably reflected in the charts and tape or how else would a discretionary trader see them? So if they're reflected in the charts and the tape then they could be automated. And yes it is possible to analyse and interpret news events. In fact it's extremely simple to code an Excel macro that enters an order on the basis of an economic data release (released real time by Bloomberg and Reuters). Reuters, Bloomberg and Dow Jones also use XML tags that enable automated systems to react instantaneously to news events. Why do you think the market reaction to news events is instantaneous? Because automated systems are reacting within milliseconds - much faster than any discretionary trader could react.

Being FAST doesn't correlate to being CORRECT.

re: "...are presumably reflected in the charts and tape"...

sorry, I totally disagree.....the reflection doesn't appear till AFTER THE FACT.

Same situation is seen in professional poker....the instincts and intuition of the best players is simply amazing.....and not easily "programmed". They are acting upon the NEXT CARD, not what is currently "showing" ie the "charts".

re: "excel macro to enter orders"....oh yeah, any DUFUS can now easily enter automated orders thanks to IB's work in simplifying the ever-so-awful FIX interface. And with this, I've seen so many traders crash-and-burn especially when trading the financial index futures.

[brownsfan019 17]

Brownsfan: you could be as flexible in coding an automated system as you are in implementing a discretionary system. What are you looking for in interpreting your candles? For price to swiftly reverse on high volume? That can be coded into an automated system.

 

Notouch - I am not looking to code that at all. I don't believe I posted that the way I trade candles is to look for high volume on a price reversal anywhere on this site. So, perhaps what you posted can be automated, but that's not how I trade at all.

[Don4 10]

This is a debate that could last forever,they both have good and bad points, but in the end, its what works best for you !!!

[notouch 10]

Notouch - I am not looking to code that at all. I don't believe I posted that the way I trade candles is to look for high volume on a price reversal anywhere on this site. So, perhaps what you posted can be automated, but that's not how I trade at all.

 

It was just a hypothetical question. :p

 

The point is that to be a successful trader you have to follow rules and any rule-based system can be automated. Your rules can be as strict or flexible as you like.

[Soultrader 12]

It was just a hypothetical question. :p

 

The point is that to be a successful trader you have to follow rules and any rule-based system can be automated. Your rules can be as strict or flexible as you like.

 

Well thats the tricky part. Discretionary traders may not have rules for each trade they take. Alot of discretionary trading is based on "gut" feel just like a poker player will be able to read the opponents hand through player psychology and probabilities. Do not mistaken this with impulsive trading. Discretionary trading involves good judgement and intuition. As TheBramble pointed out "Most would not even know what they are doing on a conscious level." This is why discretionary trading can not be automated 100%.

 

Ive pointed this Linda Raschke quote before somewhere else:

 

"I've known hundreds of professional traders throughout my career. I don't want to disappoint you, but I know of only two who were able to make a steady living for themselves with a mechanical system.â€Â

[notouch 10]

I'm not sure trading on the basis of a gut feeling is going to give you an edge. I think sticking to your rules will give you the edge. I am open to the idea that the human mind is like a neural network that can be programmed to spot complex patterns but I doubt there are many people who could successfully trade that way.

 

I'm sure Linda Raschke was thinking of canned systems like MA crossover systems when she spoke of mechanical systems. I doubt she was thinking of advanced black box systems run by investment banks. I'll try to find the article I read somewhere saying that there'll be 90% fewer traders in London within the next few decades - replaced by programmers.

[Nick1984 10]

 

Ive pointed this Linda Raschke quote before somewhere else:

 

"I've known hundreds of professional traders throughout my career. I don't want to disappoint you, but I know of only two who were able to make a steady living for themselves with a mechanical system.â€Â

 

I guess that might be true for traders who use a mechanical system that is static and doesnt develop. If someone invested sufficient time into programming so that they could modify the programs they use to suit changing market conditions this might not hold true in the long term. But then again to be that good at programming you really need to spend a long time learning it. The opportunity cost is learning the markets urself as a discretionary trader!

[januson 10]

I'm not trading automated but I'm studying a lot about it. The seach for the holy Grail.... well dreams are free :P

First of all I think I need to trade by myself and if that succeed (ie. not losing money) then and only then I should be able to develop a semi-automated trading system.

 

However I stumpled across this website: Collective2 - The Trading System Authority which compares a lot of different automated trading systems. I've found this one: Collective2: woowy 2007/3 especially interesting

[brownsfan019 17]

Collective2 is an interesting website. I ran a couple test systems there and it's a great exercise if you really want to see how you perform in real time. The only issue I had was that if you need to execute orders quickly, there was a serious lag at times b/c it is a web based order entry system.

 

I would suggest taking a look if you want to test some ideas out and/or make a few extra bucks from people subscribing.

[TheBramble 10]

Look, unless you're a Quant doing it for salary, and don't need to show consistently profitable results in all market conditions - forget it. There's little point.

 

Read "Trading Systems"; Joe Krutsinger and find out how those who are supposed to be guru system developers for trading systems do it. There are hundreds of systems they use and have used. What does this tell you? There isn't one system they've automated or codified which does the job. OK?

 

Most of the systems they use are based on MAs and BOs. Really simple, basic stuff. Which work real well. Manually. You don't need a system. If you're too lazy to do your research and run your analyses yourself then you deserve all the success you're unlikely to get. Trading is not a get rich quick and stay rich endeavour. You get what you put in.

 

If you want a plug-n-play approach to trading, use a mutual fund.

 

If you think you can program a better solution to your current trading than your current trading - you can't. Trust me. If you were than good you'd be inventing the next Microsoft.

 

You're either a Trader or a Programmer.

 

If you're a Trader, take the time and make the effort to Trade!!! And do it perfectly.

 

If you're a Programer, take the time and make the effort to Program!!! And do it perfectly.

 

Both endeavours are equally gifted and equally appreciated. But confusing the two leads to not much at all really.

[januson 10]

I work as a systemdeveloper for a living and I don't think it's possible to program a piece of SW that can do better than an experienced trader. However I've a strong believe that a program can do scan through a list of stocks and filter out the most potential ones. For instance, give me all stocks which is accumulating along with a growing momentum or scan for stocks with ma(50)day is crossing ma(200)day from above or.... but to create the Holy Grail huh... imho impossible, however I do think that a nice implemented trading system can perform alot better than the average trader.

[TheBramble 10]

For instance, give me all stocks which is accumulating...

Well actually, you can do this even with the most basic of scanners.

 

What constitutes accumulation? What factors lead YOU to belive a stock is being accumulated? And it's not OK to simply say 'a stock which doesn't appear on anyone else's scanner' as that is a very circular argument.

 

What tells you a stock is being accumulated as opposed to simply being ignored? How do you know?

 

I'll give you a cluse - it's CONTEXT.

 

I'll see if I can figure a way to put this in non-trading terms as it sometimes helps to get a fix from non-trading bearings when we discuss the deeper apsects of what goes on in reality.

[Horus 10]

Look, unless you're a Quant doing it for salary, and don't need to show consistently profitable results in all market conditions - forget it. There's little point.

 

Read "Trading Systems"; Joe Krutsinger and find out how those who are supposed to be guru system developers for trading systems do it. There are hundreds of systems they use and have used. What does this tell you? There isn't one system they've automated or codified which does the job. OK?

 

Most of the systems they use are based on MAs and BOs. Really simple, basic stuff. Which work real well. Manually. You don't need a system. If you're too lazy to do your research and run your analyses yourself then you deserve all the success you're unlikely to get. Trading is not a get rich quick and stay rich endeavour. You get what you put in.

 

If you want a plug-n-play approach to trading, use a mutual fund.

 

If you think you can program a better solution to your current trading than your current trading - you can't. Trust me. If you were than good you'd be inventing the next Microsoft.

 

You're either a Trader or a Programmer.

 

If you're a Trader, take the time and make the effort to Trade!!! And do it perfectly.

 

If you're a Programer, take the time and make the effort to Program!!! And do it perfectly.

 

 

I have to disagree with you there. I come from the exact opposite side of everything you are assuming above.

 

I do trade fully automated models. I do code professionally for myself. So for those of you out there that dream of automated trading, it is possible, BUT it comes at a great price.

 

Joe Krutsinger and the gurus that sell stuff, don't trade for a living, and never made a dime trading. So I agree with you on that, they are not examples to follow or even consider. All their work is useless to me. There is so much more valuable knowledge that is priceless published for free in academic circles, I dont think you need to spend a dime on courses etc.

 

Possibly the best public model EVER is Mark Brown's OddBall system which was published a few years back. Some of the largest hedge fund and bank traders are keying off of OddBall or some variation of OddBall. I know this for a fact and I have seen thousands of contracts pure into the pit and globex when Oddball fires of its signals.

 

This is proof that it is possible to create your own systems and trading models, however having siad that I did hear that Mark Brown used cray computers to create and find these models.

 

I personally still use a variation of OddDball to this day after about 5 years of it being published. Track record is solid and anyone with some basic prog knowledge can get it running!

[torero 10]

Hi Horus, very encouraging and true that systems do work. It's a matter of skills, creativity and lots of sweats that you'll eventually get a decent system that earns something.

 

Back to Oddball, wouldn't this system become obsolete because everyone is trading at the same time and eventually cancel themselves out? Eventually the market will absorb and change to make the system becoming useless. I'm speaking of too many using the same system, not systems in general as useless.

[januson 10]

Hi torero... exactly my thoughts also. I've found this: NeoTicker Blog » Oddball System - An Update it's clear that the system performed very bad after 2002.

Could that be the evidence that it has been over used?

 

Horus -> You mentioned you're using a slightly improved variation of OddBall, do you mind to share - not the exact algorithm, but just some bits?

[januson 10]

What constitutes accumulation? What factors lead YOU to belive a stock is being accumulated? And it's not OK to simply say 'a stock which doesn't appear on anyone else's scanner' as that is a very circular argument.

 

If the stock accumulates or distributes you can tell it by looking at A/D - Accumulation/Distribution Technical Indicators, Technical Analysis

 

So back to my point, the scanner could then look for long entries by filtering what stocks that have a growth rate greater then 3% taken from bottom reversal, furthermore it could use some sort of a regressionline between two earlier top or bottoms to enhance the probability....

 

All this is only pure guess work, but it's something that I'm going to test as time goes by

[torero 10]

Many many system failed after the 2002 because they were best during the tech bubble but after that volatility changed. Systems need to be improved or optimized from time to time (when is the million $ question). No system will last forever, they will work at one time or another, like discretionary traders, even they go into drawdowns from time to time, even the best and experienced happen to them.

 

The fact that so many systems are trading the same way eventually they get diluted, new systems will come and take advantage of them and make them obsolete. It's like everything, empires will come and go.

[Horus 10]

The Neoticker implementation of OddBall is incorrect. They missed the timing issue which is one of the key factors in OddBall. I had no problem in 2002, the returns were less but the system made money at the end of the year. The core concept behind oddball is unbreakable because its absolute market truth.

 

Its funny from one stand point discretionary traders always say the reason their methods like retracement, Fibonacci and pivots work is due to self fulfilling prophecy, if enough people follow it, a support level will hold!

 

Then on the other hand, when discretionary traders talk about system, they tell you, it failed because too many people are following it! The fact is no one knows and if the core concept is solid, its very hard to break a system. SO the reason systems and methodologies fail is not because "too many" followers, rather its core concept is only a fad. Its a patterns observed which has nothing to do with the core dynamics of markets.

 

As for variations of Oddball, just do your own research and it will be fruitful. The core concept is unique and solid. There are hundreds of variations possible some good, some great! It does take a lot of sweat to make something you are comfortable with trading. The keys to any method is the creator first then the concept. I can give a model and I GUARANTEE you out of 100 people only 2 can match my results in REAL LIFE. Turtles are system traders, at least start that way, now run empires in money management.

[torero 10]

I found this page on this method, is this as true to the original concept you mentioned?

 

Trader.Online.pl® - MetaStock™ Zone - Oddball S&P System by Mark Brown

[januson 10]

Just wanted to share: Progress Apama Algorithmic Trading Platform for Buy-side and Sell-side Trading: Apama: Progress Software Corporation

 

IMHO the future is getting darker and darker :s

[januson 10]

Another evidence that automated trading is getting bigger and bigger: Traders Online: The stock dealers and institutional traders complete interactive news and information service of Traders Magazine

 

I'm tired and amazed... it's 01:23 midnight here in Denmark wonder if I'll be able to sleep... there's so many things to read

[TheBramble 10]

I have to disagree with you there. I come from the exact opposite side of everything you are assuming above.

 

I do trade fully automated models. I do code professionally for myself. So for those of you out there that dream of automated trading, it is possible, BUT it comes at a great price.

 

Joe Krutsinger and the gurus that sell stuff, don't trade for a living, and never made a dime trading. So I agree with you on that, they are not examples to follow or even consider. All their work is useless to me. There is so much more valuable knowledge that is priceless published for free in academic circles, I dont think you need to spend a dime on courses etc.

 

Possibly the best public model EVER is Mark Brown's OddBall system which was published a few years back. Some of the largest hedge fund and bank traders are keying off of OddBall or some variation of OddBall. I know this for a fact and I have seen thousands of contracts pure into the pit and globex when Oddball fires of its signals.

 

This is proof that it is possible to create your own systems and trading models, however having siad that I did hear that Mark Brown used cray computers to create and find these models.

 

I personally still use a variation of OddDball to this day after about 5 years of it being published. Track record is solid and anyone with some basic prog knowledge can get it running!

Then I stand corrected. You are a trader and a system developer that makes consistent profits (more than 5 years) using an automated trading system they have devised and written themselves, but you are the only one, maybe just the first of that breed I have never before encountered.

 

I came from a technical programming background (many years ago) and initially thought there was a direct link between sexy code and great trading. The thing is, even when you finally get to consistent trading results, you realize (sorry, I realized) that the process I used was (a) too simple to require automating and (b) too complex to automate because of all the little discretionary aspects of it which I hadn't realized existed before I tried to codify what it is precisely that I do.

 

The effort to produce and mantain any system for me was WAY too expensive in time and effort and it also diminished that aspect of trading that I enjoyed the most - actually trading!

 

I am pleased to have found someone though who seems to have successfully combined the two disciplines and I can update my world model accordingly.

 

I'll check out Oddball as you give it a recommendaton.

[TheBramble 10]

If the stock accumulates or distributes you can tell it by looking at A/D - Accumulation/Distribution  Technical Indicators, Technical Analysis

 

So back to my point, the scanner could then look for long entries by filtering what stocks that have a growth rate greater then 3% taken from bottom reversal, furthermore it could use some sort of a regressionline between two earlier top or bottoms to enhance the probability....

 

All this is only pure guess work, but it's something that I'm going to test as time goes by

januson, throwing a textbook definition at the topic doesn't really help. The A/D you relate is flawed, it's validity is based upon either a trending instrument or on a divergence with price. Neither work.

 

I was talking about the deeper market forces, strong hands, big money actions to quiety acquire large blocks of stocks without impacting th eprice and the same divesting themselves of these stocks to their advantage.

 

Accumulation of this nature does not appear as big volume (the major weighting factor of the A/D). The divestment occurs generally in the public participation part of the bull run (the blowoff) and required somewhat less finesse to operate or identify.

 

For our purposes, looking at this from a more professional manner rather than the textbook manner that most newbies, including myself when I was going through the learning curve, will require slightly different thinking.

 

My initial question was: What do we need to know about a stock's price and volume profile to assess whether it is simply being ignored or is being quiety accumulated.

 

Your suggestion of 'growth rate greater than 3%' is possibly going to find those stocks which are in or starting a trend. Possibly. And that's fine, but I didn't think we were discussing the process at that level.

 

I'll bow out if I've misunderstood what it was I thought you were attempting to discuss.