Discretionary, Systematic, Mechanical: which are you?

[tech/a]

Re: Discretionary, Systematic, Mechanical which are you?

MIT.

Like you I've been developing systems for many years.

I'm interested in your comment that your method generated more signals than you could buy------.

This is true of all long term systems.As most holds in those methods I use are over 300 days and a full portfolio is 10-12 positions----there are 10's if not 100's of signals that cannot possibly be taken.

Without knowing how you develope your systems--- I cant comment on your systems developement methodology----

Point I'm making is that a well developed system will make consistant profit REGARDLESS which stock signalled makes up the portfolio.
Certaintly eyeballing is a good method for selection and one I as well have used.But if it out performs the initial method by a great deal the this should be a FILTER within the method.

This in essence could control your universe of stocks more so than your actual portolio.
Sure in the end if you had enough capital to place a meaningful trade on EVERY one triggered within the FILTERED universe then you could have an optimum method.
(I know that running EVERY stock picked in a test on my systems that the returns over 8 yrs are like the GNP of a small country!----But in practice I'd need starting capital of around $3 million.)

My current veiw is that trading by any means (Discretionary,systematically,mechanically) can be greatly enhanced by selection of stock universe--------and further enhanced by trading the composite charts of those universes (to act as a FULL portfolio entry exit mechanism).


Finally you could do better yourself IF you employed the principals that made your system profitable in the first place,you then as I have mentioned above are not restricted by those variables you have selected in your system which are rigid.
You will more than likely increase your % wins but decrease your $/trade win.
I'd be interested to hear if this was/is the case.

 

[mit]

Re: Discretionary, Systematic, Mechanical which are you?

tech,

I couldn't agree more. I did quite well, much better than just buying the index, but probably below average of all potential backtest outcomes for my system (A version of Alan Hull's system) . I think that this was because of my selection methodology.

If say I had 3 candidates for purchase and enough money for a single position. I selected basically on fundamentals such as low PEs/debt and the analysts calls (In commsec they have that screen that produces a score based on all analysts recommendations).

The opposite seems to be more true*. My observation is that the fully-valued and overvalued stocks just keep going and in many cases the company's business caught up with the valuation. If it didn't, well that's what stop losses are for.

I mean 18 months ago who would have bought Aristrocat. I had a buy signal three times before I bought it and it became one of my largest profits.

MIT

*There is a study done in America that seems to back this up.

 

[tech/a]

Re: Discretionary, Systematic, Mechanical which are you?

MIT

An interesting observation certaintly I agree,remember the likes of Davnet.

Still we are both preaching to the converted it seems.

 

[Hanrahan]

Re: Discretionary, Systematic, Mechanical which are you?

Hi guys, I'm new to this forum so we haven't met and I'll admit to posting on this thread before reading it all or visiting Tech's web site. But I'm very interested in the subject.

I am currently playing around with a modest bank. Have been doing so for over 5yrs and with the exception of one bad trade in ERG which slowed me down for a year until I got back in the black, I am doing quite well.

I'm in the process of selling an investment prop, I'm a beneficiary of a will and both my lady and I will be retireing in a year or so and will have some super to invest. Although I'm more interested in "investing" it would be nice to trade too.

Currently I'm doing top-down study on metals, energy and PM's to find safe sectors but then I use a scattergun approach using chat-board tips mainly to buy individual stocks. I'm been in and out of many with no harm done but I have found a few gems (BSG which I read about on Egoli) which I hold long term and trade the swings.

I am conscious of beginner's luck and have drawn down the bank to test if I can repeat my success and so far so good.

I'll study the thread and ask some Q's later.

 

[aarbee]

Re: Discretionary, Systematic, Mechanical which are you?

Hello guys and gals,

This is my first post. So far I have been lurking and trying to absorb the wealth of ideas on this forum. I have been educating myself in stock trading and have just about completed my first trend following long system to trade ASX FPOs. I have been paper trading for 3 months and intend to go live within a month.

mit and tech/a,

The subj of choosing one out of a few qualifying shares has been on my mind for sometime. Experienced mechanical traders seem to opine that the pick should be random but logic seems to point towards using discretion based on fundamentals etc. I understand from your posts that the random pick with no pre-concieved ideas about the stock's performance and without involvement of emotion is the way to go. Am I correct in my understanding? My intention is to be a totally mechanical trader till I have a firm understanding on trading matters.

Cheers!!

 

[kaveman]

Re: Discretionary, Systematic, Mechanical which are you?

Second time tried to read this through, got most way before found was skimming
having no idea who makes money at trading and how or how long they have been doing cannot comment.
but I have done the search for the pot of gold mechanical system, and yety to find one that turns profitable in real time when appied to shorter term trading. I believe from memory tec's system was medium term, (holds mfoe months)

One thing I can comment on. Mechanical systems are only discretionary trading with defined rules. You create a mechanical system from your ideas and obsevations, and decide that such and such setups could be traded. To me this is the discretion, mechanical just means you applied sets of rules and guidelines to define what you trade.

 

[Milk Man]

Re: Discretionary, Systematic, Mechanical which are you?

Second time tried to read this through, got most way before found was skimming
having no idea who makes money at trading and how or how long they ahve been doing cannot comment.
but I have done the search for the pot of gold mechanical system, and yety to find one that turns profitable in real time when appied to shorter term trading. I believe from memory tec's system was medium term, (holds mfoe months)

One thing I can comment on. Mechanical systems are only discretionary trading with defined rules. You create a mechanical system from your ideas and obsevations, and decide that such and such setups could be traded. To me this is the discretion, mechanical just means you applied sets of rules and guidelines to define what you trade.

And that these ideas can be thoroughly backtested- therein lies the key. You can plan to succeed- failing to plan is planning to fail.

 

[tech/a]

Re: Discretionary, Systematic, Mechanical which are you?

Hmm great discussion.

I tend to agree with Graham.I know Grahams Amibroker abilities go before him and I'm sure he is a capable developer of trading methods.

There are a few who I know who are accomplished systems developers (Albeit we are not of the calibre of Pring or Kauffman) and can design a profitable system.

One thing I have found in common with them ALL is that no one has reported being able to design a profitable short term systems trading method for trading a portfolio of stocks.(Distinctly different to FUTURES where there are many profitable short term methods)

This is my 3rd go at it!

I have however come to a few (Unproven) conclusions from the HRS of designing and backtesting failures.

These are my observations and my hypothesis so far--.

(1) We should take a leaf out of the Futures systems developers and simply design a method to suit 1 or at (Perhaps) maximim 2 stocks---blue chip OR Small cap--doesnt matter.

(2) Short term methods should use PRICE ACTION ONLY (Unless you have 15 min tick data) rather than Oscillators---as Oscillators in EOD form are simply FAR TO SLOW.

(3) It is entirely possible that short term methods may not be quantifiable in terms of Systems tests as the best trading methodology could well mean the incorporation of some discretionary element/s


Interested in all comments.

 

[mit]

Re: Discretionary, Systematic, Mechanical which are you?

Hello guys and gals,

This is my first post. So far I have been lurking and trying to absorb the wealth of ideas on this forum. I have been educating myself in stock trading and have just about completed my first trend following long system to trade ASX FPOs. I have been paper trading for 3 months and intend to go live within a month.

mit and tech/a,

The subj of choosing one out of a few qualifying shares has been on my mind for sometime. Experienced mechanical traders seem to opine that the pick should be random but logic seems to point towards using discretion based on fundamentals etc. I understand from your posts that the random pick with no pre-concieved ideas about the stock's performance and without involvement of emotion is the way to go. Am I correct in my understanding? My intention is to be a totally mechanical trader till I have a firm understanding on trading matters.

Cheers!!

I still have angst when I have more shares than money when running my portfolio system and try to pick the best. I just no longer look at FA but see which has the best chart.

MIT

 

[mit]

Re: Discretionary, Systematic, Mechanical which are you?

Kaveman,

I don't know if you are talking just about shares but I have a short term (1 day hold) Forex mechanical system which backtested okay (Took 9 months and playing with heaps of systems to find it).

I agree with Tech that you wont find a short term system that you can use across the asx200. Shares seem to have their own personality. I used to swing trade the shares of the company I worked for with about 80% success as it seemed to follow overbought and oversold like a text book.

After I finish the long term system. I'm going to have a look at some short term systems as well. More because I can't believe this bull market is going to last forever.

MIT

 

[tech/a]

Re: Discretionary, Systematic, Mechanical which are you?

Hello guys and gals,

mit and tech/a,

The subj of choosing one out of a few qualifying shares has been on my mind for sometime. Experienced mechanical traders seem to opine that the pick should be random but logic seems to point towards using discretion based on fundamentals etc. I understand from your posts that the random pick with no pre-concieved ideas about the stock's performance and without involvement of emotion is the way to go. Am I correct in my understanding? My intention is to be a totally mechanical trader till I have a firm understanding on trading matters.

Cheers!!

Sorry didnt see this until MIT answered.

There is NOTHING Random about my selection of Universe or selection of stock to trade.
Wether this selection is better than Random can be argued infinitum as my trading systems are tested on Random entry as well,and they all perform well even when picking a random entry.
Trading mechanically (unless you can log trades with a broker) for some can be as difficult to execute as one who trades in a discretionary manner.

Much easier to do if you design your own system AND you trust the integrety of your testing.The benifit of a mechanised syetm is that you have a blue print of whats likely to happen.Youll also have parameters in which to work.

Techtrader had its largest drawdown in the 3 yrs its been traded just reciently and even that was well within the tested Peak to Valley Drawdown results.It certaintly helps when your trading it and in this case dropped 30K profit in a few weeks and you have an idea what the maximums could be.
If they fall outside the maximums then Id close the method down as something is wrong with your testing.

On the point of not enough capital to take ALL trades---that shouldnt be a problem with any system as testing-----in particular Monte Carlo testing---
will show that (Or should show that) any trade is as good as the next so trade selection theoretically doesnt matter.
BUT like MIT I to select trades on a discretionary eyeball basis further filtering prospects.

Techtrader returned in testing betwen 17-28% P/A unleveraged---now I and a few others have returned around 38-48% P/A unleveraged over the last 3 yrs-----as the maximum profit of the method was only 28% in Monte Carlo testing then the eyeball filter must have a positive influence.

What is that filter?
(1) Must be in an OBVIOUS uptrend OR have clearly BROKEN a longer term downtrend.
(2) Must not have ranged over a period of years IE between $1.00 and $1.40
Unless ofcourse the breakout is from this range.

Summary.
Its shouldnt matter which one you purchase but they should have a buy signal either Technical OR Fundamental.
Selection of what stocks or watchlist you trade from (Known as your Universe of stocks) and what constitutes a buy is an important aspect but not THE most important.

Does that help?

 

[wayneL]

Re: Discretionary, Systematic, Mechanical which are you?

Hmm great discussion.

These are my observations and my hypothesis so far--.

(1) We should take a leaf out of the Futures systems developers and simply design a method to suit 1 or at (Perhaps) maximim 2 stocks---blue chip OR Small cap--doesnt matter.

(2) Short term methods should use PRICE ACTION ONLY (Unless you have 15 min tick data) rather than Oscillators---as Oscillators in EOD form are simply FAR TO SLOW.

(3) It is entirely possible that short term methods may not be quantifiable in terms of Systems tests as the best trading methodology could well mean the incorporation of some discretionary element/s

Tech,

I almost totally concur with you here....my comments on your three points:

1/My total universe of stocks for short term swing trading is 20 stocks...that is out of about 10,000 total on the US exchanges. More than your 2 or 3, but still severely restricted.

2/ I wouldn't bother without intraday data...I look for patterns on the daily, but trade off 15 min charts. So to agree with you, price action only, on the daily charts and a simple MA + price action on the 15 min. (taking note of sup., res., & trendlines at both timeframes)....er, well, I admit to having a 20 & 50 MA plotted on the daily...and....ummm...well, CCI/20 as well. But these are a secondary reference and make the chart prettier. LOL

3/I don't think it IS quantifiable mechanically...even on intraday data...hence my discretionary (but rule based) approach. What causes problems with testing is the volatility in the opening hour. I often have my stops violated in the opening hour...especially in the first 20 mins.

Whether I execute the stop is completely discretionary at the open. Only after the first hour has passed will my stop become absolute.

This is something you cannot code.

My historical expectancy equation has been running at around 1.5 for quite some time now so it appears to be working well this way for me.

Cheers

 

[kaveman]

Re: Discretionary, Systematic, Mechanical which are you?

Yes my experience is mostly on shares, although I would never try short term trading asx200, unless using leverage (warrants/options/futures,CFD etc). Oh and by short term zi don't include daytrading as that is totally different scenario picking a single days result over a series of days results which is what short term trading is all about.

With mechanical system it should not matter which trade you choose if you get multiople buy signals, and can be toss of a coin. All signals to buy are based on the same setup and as such all have the same probability of being profitable. Assuming your system is profitable overall.

The problem I see in backtesting is often that the group of stocks you would select today would not necessarily be the same stocks in trading 3 years ago. Example your basket of stocks could be asx200, but if you backtest over 5 years you would need to adjust the stocks being backtested every time a change occurs in the index listing. You cannot test a stock over last 5 years if it was not in the index 2 years ago.
Cannot recall exaclt how often the index stocks are checked, but say your backtest would need to include all stocks (including deaduns) ove rthat period of the backtest, and have a date condition for when each particular stock can be included in the backtest.
Whe I was into mechanical system searching I used a basic price/liquidity to determine the stock basket, and all stocks in the test were included from an exploration over the backtest peiod. I then included that liquidity condition in the buy conditions for the backtest.

 

[Milk Man]

Re: Discretionary, Systematic, Mechanical which are you?

Kave-

The index is one flaw I just realised the other day and now test just on liquidity- It must be possible to do it acurately but i'm too lazy

I agree with you that a system doesn't have to be mechanical to work- not that i'd know but it just seems logical to me. I see it all the time- farmers that don't plan or set rules to go by fall by the wayside. I would imagine a true discretionary system would take a great deal of mental fortitude to be able to pull off. Not saying youre wrong at all but some kind of systematic approach would surely be advantageous wouldn't it?

 

[kaveman]

Re: Discretionary, Systematic, Mechanical which are you?

Even discretionary traders need a set of rules to govern how they trade.

It must be possible to do it acurately but i'm too l

Loaky, what exactly would you want to do accurately

 

[Milk Man]

Re: Discretionary, Systematic, Mechanical which are you?

use the XJO in backtests

 

[kaveman]

Re: Discretionary, Systematic, Mechanical which are you?

you can do, just remember XJO only has 5 years history
List of suspects in XJO, I beleive I saw a post somewhere recently about historical consituents.
You an get the list of changes on the S&P site, and probably others. Just a matter of setting up the dates for which each ticker applies.

eg say a stock was a member before 1/1/04, and another from that date took its place
something like this

UseTicker = iif( name()=="XYZ", datenum()<1040101, iif( name()=="WXY", datenum()>=1040101, 1 );

Buy = YourBuyConditions and UseTicker;

this statement can be done easily in excel

 

[aarbee]

Re: Discretionary, Systematic, Mechanical which are you?

mit and tech/a,

Thanks for your comments. The Tradesim results of my trend following med term system show a 200% profit in last 10 years with a max peak to valley drawdown of 6.1%. Have been papertrading for a few months in a totally mechanical way and intend to keep it mechanical when I go live in a couple of weeks.

Cheers!!

 

[tech/a]

Re: Discretionary, Systematic, Mechanical which are you?

mit and tech/a,

Thanks for your comments. The Tradesim results of my trend following med term system show a 200% profit in last 10 years with a max peak to valley drawdown of 6.1%. Have been papertrading for a few months in a totally mechanical way and intend to keep it mechanical when I go live in a couple of weeks.

Cheers!!

ararbee.

Your results are 400% better than any system I have seen developed for portfolio trading over the 4 yrs Ive been involved in systems testing.48% being the best I have ever seen verified.
T/T returned around $150% from live trading but only in the 20%'s on testing.
So 200% not impossible.

The highest returns I have seen use peak and trough formulas---do you?

 

[GreatPig]

Re: Discretionary, Systematic, Mechanical which are you?

But is that 200% over the whole 10 years, or 200% per annum over the 10 years?

GP