Mechanical System Traders - How Long Did You Take To Get Your First System?

[Remiraz]

Hi, anyone mech system traders wanna share stories of their startups?
I'm a fledging newcomer trying to absorb all the information overflow i have gathered attempting to trade mechanically these 3 mths. Thankfully my tuition fee was small (US$500).

How long did it take for you to get ur first system up?
Countless years of number crunching or just a couple of months? Any sob stories?

 

[Denny]

Mechanical Systems

Call me a cynic but I am rather dubious about mechanical systems, how can they work?
Because if they did then the designer would be a billionaire within months and everyone would be using it which in turn would negate the system.
I agree that the turtle traders seemed to make a go of it I doubt if it would work now, which is why they published the system for all to see. Besides with the draw downs they sometimes suffered would have made it a failure for 99.9% of the population.
My experience (for what its worth) is that mechanical systems often work perfectly in the past but not so well in realtime.
There are literally 100s of mechanical systems published but not any of them can work over any length of time and through every concievable circumstance.
Trading / Investing is an art rather than a science, if it was a science then pure knowledge would be king and everyone could make money by just studying the science.
IMHO

Denny

 

[TWI]

Market is dead today.
Here is my story.
I started looking at Mechanical systems in mid 90's using Tradestation 4.0. I knew nothing to very little in those days and the system testing was very crude. It was necessary to write your own code to get an equity curve that appeared like an indicator under the chart. I remember thinking the optimiser was just fantastic and showing other traders "amazing" systems "making" huge amounts that I had optimised with MA's and indicators. I learnt a few expensive lessons, fortunately for me, with the companies money. As a result I ditched the whole mechanical trading system thing and just continued to trade manually as I always had. I was only trading commodity futures in those days and had no real idea about anything other than fundamental analysis with "Pring on technical analysis" my only source of technical knowledge.
One day somebody showed me an article written in "Futures Magazine" which talked about the correct way to use optimisation, it was really good so I decided to have another go. Wasn't until 1999 that I had a system that actually made some money in the real market. It traded grains and softs and was a simple trend follower. By that time I had a decent equity curve so I could see how volatile historic returns were and used a spreadsheet to analyse the optimisation. Problem was that despite the fact that it did make money there were some horrendous drawdowns along the way and I was asked to explain what the hell I was doing several times. The stress got too much and I ended up only trading the system on my own account and not for the business anymore. It nearly wiped out my account a couple of times and had one large margin call but eventually actually made about a 100% return on my capital in the first 18 months. The draw incidentally was just under 100% along the way so it was not that good. Things did get better and by 2001 I had a system (built together with a colleague who was a good programmer) that traded 23 commodity markets with a nice smooth curve and what looked like consistent 25-35% returns pa. Unfortunately this failed to set the Hedge Fund industry alight, as they didn't seem to keen on the commodity markets and particularly when I mentioned it would be fully capitalised at 40-50mln so after being unable to get more than $4.5million funding for it I decided to do something different and switched to trading financials for myself.
I do not trade that system anymore as I do very little with the commodity markets and I have better systems to follow in bonds and FX but, in answer to your question, 6-years was my struggle. I learnt a lot and continue to learn more all the time.
I run systems on FX but I still trade a lot of my capital manually because the method I use to trade interest rates I have not, to date, been able to code, I have only got as far as a nasty looking spreadsheet.

 

[JTrader]

Hi twalker

I get the impression that you are referring to EOD mechanical trading strategies in the above post?

What about intraday mechanical trading systems (MTS) - have you found profitable intraday MTS's. easier to design and use than EOD ones?

As an experienced tradestation easy language user, How long would it normally take you to come up with a decent and profitable basic intraday MTS - either scalping or momentum trading - that you could then optimise for, and apply to different markets?


Many thanks

jtrader.

 

[TWI]

I largely rely on intra-day MTS's now. The older ones were indeed EOD.
I find that optimisation is an easy term to use but very hard to get right. I just try to find stability of profit in the variables to the point where I personally feel happy about the values being used. That part is more about not losing than winning.
Even when all is done I do not have 1 system that works across all markets. Best I have is one system that works across 4 currency pairs. Even one of those pairs is not pretty but together they look good. Also have systems that work on 2 pairs effectively but not on any other market and you have seen systems posted on this board or others that work on 1 pair.
The EOD system for commodities I mentioned worked across 23 markets but was only really viable when trading all 23 at once to obtain portfolio effect of smoothing returns. Sometimes systems which show quite volatile profit returns on a given market can look much better when applied to several markets at same time. Often if you get a great smooth equity curve on one market the system will fall over when applied to a portfolio as it is likely to have been badly optimised to the original market. Rationale for taking signals and diversification is, I have found, by far the best way to approach mechanical systems trading.
As to time scales, I have no idea really, I just take an idea, code it , test it , see where problems are and try to improve it. It is all about personal perception, some people could knock something up in an afternoon and then be happy to put money down. I do not use a system today that I have run with real money for more than 1 year, so, I am not completely sure that they will not fall over next month or next week or keep working for me. I just feel comfortable with the rationale used and the stability to date. I made some reasonable returns so far and that is great but that is all I know. At some point you just have to pull trigger and get stuck in, you will never be sure it is just probability.

 

[knorrie]

Hi twalker

I notice you said you're not currently trading anything you've been using for more than a year? How do you decide when to stop trading a particular system? Do you trade any systems like JonnyT's SpotOn and variants where you can expect significant drawdowns and/or a long series of losing trades, and if so how do you determine when they are no longer working or just in extended drawdown?

BTW I've had this discussion elsewhere and am aware of some solutions like an equity MA curve or Monte Carlo techniques, but I'm interested to hear what you as someone with a track record of using mechanical systems over time actually does.

Thanks in advance,
KenN

 

[TWI]

I do not know how to answer your question as it is not something I have a corporate, political, buzz word, **** covering type answer for. Sorry.
I consider that when there is a drawdown which is substantially greater than anything I have seen before it is time to increase my stake. This is the element of faith as the rules are simple and logical and with diversification of systems and products the scenario where things totally fall apart is less likely. It doesn't mean they will not. Certainly the systems I use today are very different from the ones I looked at years ago. There has been a slow but continuous evolution of ideas. I have found this and other forums have been a great source of new ideas.
Would be interested to hear what came out of other discussions you had, it may save my rummaging in dustbins if things go bad.
Actually I always thought that If it all goes very horribly wrong I will have to sell them on a website showing my track record from when they worked well

 

[Roberto]

I agree that the turtle traders seemed to make a go of it I doubt if it would work now, which is why they published the system for all to see.

Sorry to be argumentative, but actually no: not at all. Rather the opposite, in fact. Their reasons for publishing the system for all to see are well explained on their website www.originalturtles.org . IMHO the reasons why their system wouldn't work for an _individual_ trader are (i) that few people have the discipline and money-management skills and capital to make a long-term go of it, and (ii) it requires trading very many different markets at the same time to try to even things out (as discussed in other threads here), otherwise the volatility of the results would be prohibitive. In my opinion there's little doubt that it would still work just as well now, but it would need too many years and too much capital to make it worthwhile for a "sole trader". In short, it's not the system that wouldn't work: it's the traders.

 

[Tuffty]

Denny, Who said the development and trading of mechanical trading systems is a science? I believe it is just as artistic as any other type of trading. The artistry is just in a different place.

Re the Turtles; I think I've seen a hedge fund that trades the classic turtles program still. From what I remember the returns had come down to risk free returns - which is what one would expect over time.

twalker, I liked your story. I'm sure it won't come to flogging the code on the web.

 

[knorrie]

twalker

Interesting about the 'increase my stake' idea. But you did say that you are no longer trading any system you have followed for more than a year, so I was just interested to know what made you stop trading the old ones.

Regarding the other discussion I had, here is a summary of the key points, taken from some posts by mmillar (no longer posting here it seems) on TacticalTrader. The reference to 'Jake' is Jake Bernstein, who I collared about this at the end of the recent TT Symposium - he mentioned using a moving average on the equity curve but didn't follow up with details on exactly what or how this worked:

1. Use a moving average or trend-line of the equity curve as Jake described.

2. Work out the system drawdown specifications before you start trading it and get out when the maximum theoretical drawdown is reached.

3. Rotate systems as they come and go. i.e. you live trade 5 systems and paper trade another 20+. Then as some of your paper trade systems get better equity curves than your 5 live systems you swap them around. So you are only every trading the top 5 systems.

To understand the real drawdown characteristics you need to take your backtesting results and stick them into a Monte Carlo simulator.

KenN

 

[Denny]

Mechanical Systems

Roberto,
I have also read the article and I know the “stated” reasons why the turtle system was published. However whether it would work today is still a matter of conjecture, the probable fact is that it would never have worked for an average trader. Not necessarily due to lack of discipline but for the drawdowns and capital required. The basic system was not new at all really as it was only a breakout system which is a much used basis for systems. They however refined it and the exits were they key as to any trading.
It has often been suggested that a good trader could make money by selecting stocks at random and with clever money management and exit strategies still make money and I am sure this is true, but would not want to test this theory myself.
My contention is still that mechanical systems cannot work and if it does it is only for a short period of time in a particular type of market which can often be put down to luck. If such a system did work why are there 100s of these systems published you would only need one system the one that worked !!!!!!!

Denny

 

[chump]

"I consider that when there is a drawdown which is substantially greater than anything I have seen before it is time to increase my stake"...sounds like a contrarian to me....I have the same outlook on property holdings...if they breach my drawdown to the extent I am feeling pain then give my numbers I know that the majority are going to be crucified so I would wish to increase my position still within the framwork of numbers that allows me to stay afloat...afterall what is the point of taking an opportunity just to have to join the ranks of all the other walking wounded.....funnily enough I would consider this to be the most low risk ,high probability play that I could get...now is that contrarian or what ?

 

[alpha_monkey]

Roberto,
I have also read the article and I know the “stated” reasons why the turtle system was published. However whether it would work today is still a matter of conjecture, the probable fact is that it would never have worked for an average trader. Not necessarily due to lack of discipline but for the drawdowns and capital required. The basic system was not new at all really as it was only a breakout system which is a much used basis for systems. They however refined it and the exits were they key as to any trading.
It has often been suggested that a good trader could make money by selecting stocks at random and with clever money management and exit strategies still make money and I am sure this is true, but would not want to test this theory myself.
My contention is still that mechanical systems cannot work and if it does it is only for a short period of time in a particular type of market which can often be put down to luck. If such a system did work why are there 100s of these systems published you would only need one system the one that worked !!!!!!!

Denny

Trend following systems, as far as i can tell, have never stopped working.

Trend identification, proper bet sizing, diversification of markets traded and 100% discipline in trading methodology are arguably the key components for success.

 

[chump]

Capitalisation is a major factor Alpha...all of your points are well taken ,but capitalisation will make some of them unattainable for many people...

 

[TWI]

I was just interested to know what made you stop trading the old ones.

It wasn't that I stopped trading the old ones so much as they were slowly superseded by what I considered improvements. I did leave systems completely alone for a couple of years as I explained earlier but came back to them starting pretty much where I had left off. What I now follow is actually considerably simpler than what I originally had.
Just price related ranges and breakouts, no rocket science. I often look at how the original systems would have performed vs. what I now have, I keep a copy of all the code, sometimes they do better sometimes much worse but one thing that has improved is the volatility of the returns.
One can never be sure how long this will continue. If I was, I would give up my other trading and concentrate solely on putting equity into mechanical systems but I really am not sure, I have had nasty drawdowns in the past and they definitely could happen again.

 

[alpha_monkey]

Capitalisation is a major factor Alpha...all of your points are well taken ,but capitalisation will make some of them unattainable for many people...

Agree completely.

My suggestions weren't meant to be an exhaustive list, rather, thoughts purely on developing a trend following system. (A given that this was poorly articulated.)

 

[jmreeve]

Hello TW,

Were the big drawdowns you had in the past the result of using long term trend following systems?

Have come to the conclusion that most portfolios constructed solely with LTTF systems will have this
problem and that it is important to add some counter-trend systems to smooth returns.

 

[TWI]

jmr - That was the cause yes. This is why I tend towards intra-day systems now and look for ranges before breakouts.
I do not think it possible to avoid the draws with the long term trend followers unless you have a very diversified portfolio. I found that the system which looked at 23 commodities was much better overall due to the diversification but still a 20% draw could not be considered extreme.

 

[jmreeve]

I do not think it possible to avoid the draws with the long term trend followers unless you have a very diversified portfolio. I found that the system which looked at 23 commodities was much better overall due to the diversification but still a 20% draw could not be considered extreme.

Did you look at the return volatility vs the number of commodities for this portfolio?
I have seen some work that suggests once you get past five or six very different security types
that the improvements decline rapidly.

I think diversifying by having trading systems trading in different time frames e.g. some LTTF, some swing and some day trading systems as well as using strategy diversification helps smooth out the returns. However, it does start to get difficult to trade with so many different systems.

 

[Tuffty]

Denny said trading / Investing is an art rather than a science and if mechanical systems worked then the designer would be a billionaire within months. Following this logical argument through then wouldn't the most artistic person be a billionaire within months too!

I think it doesn't matter whether you believe in artistry or mechanical systems to decide on entry and exit points. What does matter is if you think the markets are efficient or not. It is the inefficiencies that can make you money whether it be through a mechanical system or discretionary trading.

Any sort of trading is a game of exploiting the inefficiencies before they get arbitraged away. Maybe some inefficiencies will never get arbitraged away. This may be due to new generations having to learn how human nature influences the markets, or the market stucuture (i.e. rules, regulations etc) do not change.