Discretionary vs. mechanical trading

[tech/a]

There are other places he mentions that we don`t trade the marhet, we trade our belief about the market.This is the paragraph word for word on page 19.

"We typically trade our beliefs about the market, and once we`ve made up our minds about those beliefs, we`re not likely to change them. And when we play the markets, we assume that we are considering all of the available information. Instead, our beliefs, through selective perception, may have eliminated the most useful information."

He's certainly not advocating this to be a correct way to trade even think.
Nor is he even coming close to saying that if it "fits" then go ahead and trade it.
He clearly makes the case that most peoples pre determined biases are flawed.
No arguement from this black duck.

 

[motorway]

So the Broad distinctions imo should be between

Subjective or Objective
Arbitrary or Systematic
Gross or Fine
Inflexible or Flexible

More of the right side ( as appropriate )
is BETTER -More of the left side is undesirable ---
In the context of what works what makes profit

mechanical <---> discretionary
can be meaningless semantics
Both can have negative or positive connotations

more to do with available tools ( eg code can/can not)
and application
than good or bad

motorway

 

[Temjin]

Really interesting comments and discussion going on guys.

Flexible rules/psychological make up can and mostly do add up to disaster.
There is an implication that you need to trade more often in a discretionary manner to profit---Why?
While backtesting may well be impossible the reason and structure of a good discretionary traders method wont/shouldnt change---if they dont know what structure will (long term) trade after trade after trade return a profit then they arent trading they are gambling/mostly reacting to emotion.

While some automate most physically place the trade/s. Bias should be removed---the blueprint will clearly show whether the results found during the testing process are being reflected in live trading.
If they arent then there is only one question which needs to be asked.

Has the market changed significantly from the market/s used in the testing of the mechanical system being traded.Generally the system will clearly show this(as I found out when 3 of mine were fooled by randomness) by being exited out of all or most positions and finding less than average prospects.

It's certainly true that a lot of mechanical traders out there don't fully automate at all. However, how do you determine if the market has changed significantly? Or your system is no longer capturing the "signals" as well as it used to be? Do you determine it discretionary? Or you have already pre-determined when it would be considered no longer working?

Explanations below.

Disagree strongly with this.Mechanical systems are designed to be proven wrong due to change in market conditions as explained above---not open or due to designers ability to read markets and use a discretionary skill!
That in Blue is one solution there are others.

I guess I have not explained this with enough details.

What I mean about the "system" to generate or eliminate profitable systems is purely mechanical, at least from my own perspective. For example, setting up the structure for back testing, pre-determine your objective for system selection, your ruin level based on your risk profile (how much drawdown to consider eliminating or turning off system), confidence level via monta carlo simulations, etc, etc.

All of these factors need to be pre-determined first. Discretionary skill from a mechanical trader only comes from selecting which "strategies" to test on and then using the "system" to determine if it has a tradeable edge. (and when it has lost it)

I would certainly not use my intuition at this stage to say whether a previously working system has failed or not. If it has lost its edge, then analyzing the results would certainly reveal it.

I dont know that they are (Advantages and disadvantages).
I think the discretionary trader like it or not needs the mechanical blueprint of his structure to KNOW if what he is doing is actually skewing his results in his favour or if in fact he has a disjointed string of trades which have no structure toward being profitable---he's gambling.

It's up to you on how you choose to label their differences. I see it as pure advantages and disadvantages, others may think as positive or negatives, or others may see it as strengths or weaknesses. Doesn't matter, since the differences are clear enough, at least to me anyway.

I agree with you that discretionary trader needs a blueprint to determine if his trading style actually has an edge or not. He/she may not be able to backtest it, but like I said before, the results can be analysed to reveal more information. Mistakes that caused for the dismay performance may become more clear.

I havent seen where Van Tharp has actually said that but to me its a cop out. Fits you more means? Sure you can be warm and fuzzy about patterns but trade 100 of them or any other discretionary signal in a disjointed jumbled fashion it will and has become very clear to many that a bias or feeling toward a particular trading idea placed into a loose strategy---becomes a pretty un comfortable "fit".

Perhaps I may have mistaken his point?

But here is where he has a habit of repeatedly it in his newsletters every so often, and usually right from the start of his article. (knowing Wysiwyg has posted a similar one)

I always say that people do not trade the markets; they trade their beliefs about the markets. In that same way I'd like to point out that these updates reflect my beliefs. If my beliefs and your beliefs are not the same, then you may not find them useful. I find the market update information useful for my trading, so I do the work each month and I'm happy to share that information with my readers.
However, if your beliefs are not similar to mine, then this information may not be useful to you. Thus, if you are inclined to do some sort of intellectual exercise to prove one of my beliefs wrong, simply remember that everyone can usually find lots of evidence to support their beliefs and refute others. Just simply know that I admit that these are my beliefs and that your beliefs might be different.

I would interpret this way. I believe mechanical trading or whatever trading strategy is more "useful"/"applicable"/"suitable" to me, based on whatever logical reasonings and evidences I have. Others may not agree with it, so therefore, would not find it useful.

bingk6] Nice Post Temjin.

I also add that mechanical traders tend to perform a lot of backtesting of their strategies, which can give them some sort of instant feedback as to whether the stats that they have received are robust enough to commence trading a system.

Discretionary traders tend to paper trade, which on its own is not a bad thing, but it does take time, sometimes a lot of time. If the paper trading stats are not good, well back to the drawing board and then potentially more time spent in paper trading.

Mechanical systems are quicker off the mark. One can evaluate and choose from a number of alternatives very quickly.

Thanks.

I agree on the ability to perform a lot more backtestings and getting instant feedback. This is why people like Howard Bandy tend to favor mechanical system. I like his book a lot and it has given me alot of new insights on backtesting.

 

[tech/a]

Really interesting comments and discussion going on guys.






It's certainly true that a lot of mechanical traders out there don't fully automate at all. However, how do you determine if the market has changed significantly? Or your system is no longer capturing the "signals" as well as it used to be? Do you determine it discretionary? Or you have already pre-determined when it would be considered no longer working?

The system devistes from the known tested results---in the case of the GFC-- the 3 I used did so significantly.The resultant effect on the systems is now included in the testing and the blueprints have altered to include the GFC trading. Didnt kill them either.

I guess I have not explained this with enough details.

What I mean about the "system" to generate or eliminate profitable systems is purely mechanical, at least from my own perspective. For example, setting up the structure for back testing, pre-determine your objective for system selection, your ruin level based on your risk profile (how much drawdown to consider eliminating or turning off system), confidence level via monta carlo simulations, etc, etc.

All of these factors need to be pre-determined first. Discretionary skill from a mechanical trader only comes from selecting which "strategies" to test on and then using the "system" to determine if it has a tradeable edge. (and when it has lost it)

I would certainly not use my intuition at this stage to say whether a previously working system has failed or not. If it has lost its edge, then analyzing the results would certainly reveal it.

A wish list.
Again I dont know that you nee to pre determine it.
A developed system will return all those stats you are looking for. You will accept or reject the stats before trading it.
Then once outside those tested stats youd stop trading it---well thats what most would do.

It's up to you on how you choose to label their differences. I see it as pure advantages and disadvantages, others may think as positive or negatives, or others may see it as strengths or weaknesses. Doesn't matter, since the differences are clear enough, at least to me anyway.

I agree with you that discretionary trader needs a blueprint to determine if his trading style actually has an edge or not. He/she may not be able to backtest it, but like I said before, the results can be analysed to reveal more information. Mistakes that caused for the dismay performance may become more clear.

Sure there are clear differences I can use analysis I cant code and skew results with its use.I can also move stops add trailing stops add to positions multi position size--add subtract trades---endless.
BUT the bottom line for me is always.
AM I SKEWING the results of this trade in my favor.
Am I increasing win rate,Increasing Reward to Risk (Or decreasing Risk to reward) or both during the trade.

Perhaps I may have mistaken his point?

But here is where he has a habit of repeatedly it in his newsletters every so often, and usually right from the start of his article. (knowing Wysiwyg has posted a similar one)
[/SIZE][/FONT][/I]

I would interpret this way. I believe mechanical trading or whatever trading strategy is more "useful"/"applicable"/"suitable" to me, based on whatever logical reasonings and evidences I have. Others may not agree with it, so therefore, would not find it useful.

Which after reading Van Tharp a few times I would add
At your peril.

Thanks.

I agree on the ability to perform a lot more backtestings and getting instant feedback. This is why people like Howard Bandy tend to favor mechanical system. I like his book a lot and it has given me alot of new insights on backtesting.

More than anything testing and developing systems has taught me what I need to keep in mind when trading in a discretionary manner.
What works/how I can adjust all of those things mentioned above to increase my overall return on risk.
I dont need HARD stats I have the ability to define in a discretionary trading method ---- what to use and how to apply it to put discretionary trading squarely on the ++ side from the knowledge both positive and more so negative (What doesnt work) gained through hrs of testing and developing.

So I would advise mechanical first discretionary second.
99.9% of the time its the other way around and most fail.
I certainly did before I spent the time and effort!

 

[Temjin]

A wish list.
Again I dont know that you nee to pre determine it.
A developed system will return all those stats you are looking for. You will accept or reject the stats before trading it.
Then once outside those tested stats youd stop trading it---well thats what most would do.

Then you have already pre determine it. "the stats you were looking for" And that if recent results deviate from those tested stat, you would have specific actions for it. Either re-evaluate/optimise the strategy and getting back the stats you are looking for, or like you said, just stop trading it. Permanently or temporary.

Which after reading Van Tharp a few times I would add
At your peril.

That goes to same to anyone who are trading with a specific system that they have confidence in. (with statistical evidences to back that up of course!) It does not matter if it is discretionary or systematically.

The bottom line is to find an edge, know where it is coming from, and maintain and trade it for as long as possible and leverage the compounding effect.

More than anything testing and developing systems has taught me what I need to keep in mind when trading in a discretionary manner.
What works/how I can adjust all of those things mentioned above to increase my overall return on risk.
I dont need HARD stats I have the ability to define in a discretionary trading method ---- what to use and how to apply it to put discretionary trading squarely on the ++ side from the knowledge both positive and more so negative (What doesnt work) gained through hrs of testing and developing.

So I would advise mechanical first discretionary second.
99.9% of the time its the other way around and most fail.
I certainly did before I spent the time and effort!

So what you are suggesting is, one may trade mechanically first, but would eventually develop their own intuitions over time to determine what factors are responsible for maintaining and improving his edge without going through hard statistics all the time? So the trader now acts in a more discretionary manner when developing and testing systems. Whereas, newbies have to rely on pure statistics to gain the confidence first.

If so, then I agree and it certainly would come over time with experiences.

Of course, that is assuming if that trader's account balance would last that long or understood why he/she had failed.

Thanks for the comments.

 

[tech/a]

T

I think we are in the same book---if not the actual page.

 

[Temjin]

Phew! Thanks, great to hear I got it this time.

 

[Wysiwyg]

Both can have negative or positive connotations

more to do with available tools ( eg code can/can not)
and application
than good or bad

motorway

E.G. ..... Code can when market conditions are right but code can't see, hear, smell, taste or feel conditions so brain helps out.

 

[Wysiwyg]

Has anyone successfully created a rigid, long only mechanical stock trading system that they use with real money? By created I mean coded, backtested and functioning in real time with real money. By coded I mean ...

1) buy conditions
2) sell conditions
3) risk/position size

I am not having any luck with creating a successful long only mechanical stock trading system. I have success stock trading using eyes and ears.

 

[Gringotts Bank]

wysiwyg, yes I've a coupe of systems, but they're not spectacularly high in % returns.

Try this: optimize for 'smoothness' and 'low draw downs' ahead of '% return'. If you get a very smooth curve over many years, but a % return of say 5%, simply add leverage. As much as you want... or as much as you dare! Leverage it up to 500% if you want. AB will do that for you.

 

[Wysiwyg]

wysiwyg, yes I've a coupe of systems, but they're not spectacularly high in % returns.

Try this: optimize for 'smoothness' and 'low draw downs' ahead of '% return'. If you get a very smooth curve over many years, but a % return of say 5%, simply add leverage. As much as you want... or as much as you dare! Leverage it up to 500% if you want. AB will do that for you.

Thanx for replying. I hope some more reveal their experience in creating a mechanical stock trading system.

Yes I have some systems that return mediocre but not anywhere near as effective as (my?) discretionary decision making. The damn codes can't read or see and have a terrible time when conditions aren't favourable such as the strong bull run from 2003 to mid 2007 and the bounce from Mar. 2009. Obviously these are retrospective events and if another bull run started now then results would be good again.

I have tried optimising but change the period and results are different again.

I will tell it straight. Mechanical systems (just mine?) don't know when to close a trade and closing a trade is timing and an acquired skill. Something to become better at for mine.

 

[Gringotts Bank]

wys, the last month or so has been very hard on my 3 systems. Luckily I don't system trade. I just use the systems to get a general idea of where an edge is likely to be found.

Look at the market right now, see what it's doing, then if you want to, put it into code. eg. low volumes in spec shares, weak movements with lots of gaps, no stocks putting on 40+%, etc. All this can be coded easily into your own custom indices. The mood of the whole market can be decided by looking at stocks under $1.

Closing trades - I haven't found better than the parabolic stop and reverse.

 

[colion]

Has anyone successfully created a rigid, long only mechanical stock trading system that they use with real money? By created I mean coded, backtested and functioning in real time with real money. By coded I mean ...

1) buy conditions
2) sell conditions
3) risk/position size

I am not having any luck with creating a successful long only mechanical stock trading system. I have success stock trading using eyes and ears.

A very simple and fairly effective system is:

Buy = C > Fixed Stop
Sell = C < Fixed Stop

The Fixed Stop % will depend on the market but 5% is a good reasonable starting point.

Paper trade it and see how you do.

 

[tech/a]

Has anyone successfully created a rigid, long only mechanical stock trading system that they use with real money? By created I mean coded, backtested and functioning in real time with real money. By coded I mean ...

1) buy conditions
2) sell conditions
3) risk/position size

I am not having any luck with creating a successful long only mechanical stock trading system. I have success stock trading using eyes and ears.

Yes

Techtrader was traded live for 7 yrs and still used by many traders.
Its record is still on the net
It will also have a place in Radge's new book.

What many find difficult is they expect a long only longterm method to work well in ALL market conditions. Simply it wont--just as a short only system wont work well in a longterm trending market.

Why am I not trading T/T since 2007
(1) I was certain that there was going to be a large and devastating fall in the market and the funds generated by T/T were enough that I did not wish to risk them.
The system basically shut itself off a few months later (All positions stopped) at a closed equity figure close to my liquidation of funds.
(2) Discretionary trading is (for me) easier in this turbulence.

By the way though T/T did make new equity highs after the GFC.
Even surprised me!

 

[Wysiwyg]

A very simple and fairly effective system is:
Buy = C > Fixed Stop
Sell = C < Fixed Stop
The Fixed Stop % will depend on the market but 5% is a good reasonable starting point.

Thanks but not really chasing a strategy as such because I have tried so many that are mediocre that I know it is market conditions one needs to trade using any simple code i.e. MACD/Signal cross. As tech/A mentions it is obviously the market conditions.

Trading three strategies; one for uptrend market, one for ranging and one for downtrend market. Trick could be foreseeing the trend direction and trading it with the defined strategy. Yes, so simple.

 

[colion]

The neuronet guys use this type of scheme. Simple? I hope you find that to be the case.

 

[Boggo]

Has anyone successfully created a rigid, long only mechanical stock trading system that they use with real money? By created I mean coded, backtested and functioning in real time with real money. By coded I mean ...

1) buy conditions
2) sell conditions
3) risk/position size

I am not having any luck with creating a successful long only mechanical stock trading system. I have success stock trading using eyes and ears.

Some of these videos such as the one in the link below demonstrate a short term hit and run approach that does work even in the current market.

The video is using 3 min YM charts which are more continuous than the daily ASX stock charts which can have erratic gaps between one days close and the next days open etc.

Regardless of that it highlights and demonstrates methods that can be used in almost any markets with the emphasis being on controlling risk and knowing when to take profits.
(I am not plugging the software here but it does make it easier to apply the method)

Worth a look anyway I think for the concept.
http://www.youtube.com/watch?v=mb_gT_tO9FY&feature=related

 

[Wysiwyg]

Regardless of that it highlights and demonstrates methods that can be used in almost any markets with the emphasis being on controlling risk and knowing when to takeprofits.

That being an individual thing which is as variable as the number of market participants and their strategies.

 

[Boggo]

That being an individual thing which is as variable as the number of market participants and their strategies.

Or the reason why only 5% to 8% make consistent profits and the rest either give up, go to the casino or race track, or become 'investors' with a large bottom drawer.

 

[motorway]

E.G. ..... Code can when market conditions are right but code can't see, hear, smell, taste or feel conditions so brain helps out.

"We need a face-saving way to let go of our pre-conceived notions and the best way to do that is to understand how we make mistakes in our thinking. Thinking is really built for survival, not for truth; that’s how we made it as a species.

Our thinking is programmed to be very efficient and to draw premature conclusions so we don’t spend too much time deliberating about ‘whether to run from the sabre-toothed tiger or not’. And that’s wired into us.

The natural mistake of thinking is something called heuristics. This says that we like to tell comforting stories to ourselves to explain observed phenomena in our environment."
Doug McGuff

Objective , Reproducible , Systematic & Flexible as long as we avoid the natural mistake of thinking which is a significant factor connected imo to Boggo's post

==>

the reason why only 5% to 8% make consistent profits and the rest either give up, go to the casino or race track, or become 'investors' with a large bottom drawer.

You would not have anything in the Bottom Draw if you did not Think too much.

But instead just ran from the "sabre-toothed tiger"

Motorway