Similar Systems / Different Luck in 2011

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Post Reply
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Similar Systems / Different Luck in 2011

Post by Chris67 »

I have been looking closely, at amongst other systems, the triple MA system and in particular its performance in 2011 so far.
Quite interesting how the results are dramatically different based on your choice of shorter term MA - for example a 10 day short m.a seems to be down / flatish in 11 but a 20-50 day M.A has yielded some excellent results almost regardless of the long and medium term ma.
No great scientific reasoning other than teh markets are obvioulsy choppy and a longer term short MA seems to be working best

Also its interesting how a triple MA that, perhaps tests out best for the last 20 years may not have worked so well for the last 2 or 3 years and vice versa. Again - psycholgy at play in that systems, I guess work best where the money isnt !!

Wondered how others got around these issues. I for one consider perhaps the best approach to trade a basket of triple's to cover all eventualities depending on what the market throws up in a given year / combined with other types of short/ medium and longer term TF systems.

Best
C
AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2071
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner »

I think you have said it. A recent Sluggo post made a very similar point by charting multiple results from multiple TMA systems.

I have been looking with disgruntlement recently as to how a TMA with 350/80/20 parameters has worked this year on "my" portfolio (combined with profit taking exits, wide ATR stops and a conservative risk management system). System metrics over the long term are very similar to the way I trade and yet my own methodology has me in a small net loss YTD while the TMA shows a nice profit..

Which brings me to a recent post by svquant: if you are convinced your system is a good one (and if you are right!), the very worst thing is to switch horses in a downturn.

No one has the right answer or the right system 100% of the time. There are plenty of multiple system CTAs out there in a downturn this year, so that is not a 100% guaranteed answer. Look at Altis for instance, or Transtrend.
sluggo
Roundtable Knight
Roundtable Knight
Posts: 2987
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo »

Beware, when you trade "N" number of systems simultaneously, you now have N times as many opportunities to become disillusioned or dismayed. Whenever you make a list of their recent performance, some system or other will be at the bottom of the list, and this can lead to frustration. Why oh why did you include that system anyway? Life would be so much nicer if you hadn't made that dumb decision, oh the misery, oh the recrimination, ...

Several times I've chatted with managers who just booked a double digit up-month, but nevertheless are morose and droopy because one of the trading systems is currently underperforming. This could happen to you.
AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2071
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner »

sluggo wrote:Beware, when you trade "N" number of systems simultaneously, you now have N times as many opportunities to become disillusioned or dismayed.
I have discovered over the past decade or so that the simpler my life is, the happier I am. I don't like clutter and chuck out old clothes and other stuff I don't use anymore. I don't like too many "things" in general. I was fed up with the hassle of having two houses (double the aggravation) and so abandoned Switzerland and based myself full time in London.

And so to systems. It is a truism that you should trade in a way that suits you. I have not seen too much evidence that great complexity is necessary to succeed in the markets. I have not seen vast evidence (from the 10 or so managers "our" little CTA FoF has positions in) that the complexity and apparent sophistication of many of them has actually aided them much this year.

As for our own trading (or rather that of "our" CTA single manager fund) we will stick to relative simplicity (and to our system!) in reasonable confidence that for the present at least what we have is robust enough to pull us through and reap future profit. And we will of course continue to research and cautiously implement improvements.

A simple solution for simple fellows!
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 »

AFG - Sounds like a darn good plan - thanks for your comments - you and Sluggo
I think as long as your systems are MOO for entries and exits then its not too bad to have a few on the go - I also have a few that are stop entries and exits and they are a gigantic pain - but its the old trade off between effort and results - sometimes however that correlation breaks down too
I feel like I should have made a tonne of money this year for some reason but havent - I look at other TF's and see some are up smalls and many are down - so Im sure Im doing nothing wrong - just one of them years - it will all looking so good today until the Swiss Franc !!! I wonder if that was Anthony selling out his final personal Swissy to get back home !!!!
Dam !!!

p.s. whats going on with AHL - Somethinmg looks sorely a miss with them
Aaron01
Roundtable Knight
Roundtable Knight
Posts: 113
Joined: Thu Jun 02, 2011 1:21 pm

Post by Aaron01 »

Chris67 wrote:I think as long as your systems are MOO for entries and exits then its not too bad to have a few on the go - I also have a few that are stop entries and exits and they are a gigantic pain
Can you please explain why do you feel stop entries create more "pain" than MOO?
Chris67
Roundtable Knight
Roundtable Knight
Posts: 1052
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 »

They do not create more pain - thet create more work
If you have a portfolio of 100 instruments and therefore you have a 200 orders every day feesibly ?? 100 possible stop entry longs and 100 stop entry shorts
With an MOO system you have no orders to watch other than if you get a MOO entry or MOO exit generated - so if that part of your order sheet is blank - day finished !
Aaron01
Roundtable Knight
Roundtable Knight
Posts: 113
Joined: Thu Jun 02, 2011 1:21 pm

Post by Aaron01 »

Chris67 wrote:They do not create more pain - thet create more work
If you have a portfolio of 100 instruments and therefore you have a 200 orders every day feesibly ?? 100 possible stop entry longs and 100 stop entry shorts
With an MOO system you have no orders to watch other than if you get a MOO entry or MOO exit generated - so if that part of your order sheet is blank - day finished !
This is what I thought, however I just wanted to check and make sure. Thanks :D
M20J
Senior Member
Senior Member
Posts: 39
Joined: Tue Aug 12, 2008 4:49 am

Post by M20J »

They do not create more pain - they create more work
Although to be fair, stop orders can on occasion be very successful at creating more pain too.
Attachments
3mEuroSwissFranc.jpg
3mEuroSwissFranc.jpg (103.64 KiB) Viewed 8780 times
ratio
Roundtable Knight
Roundtable Knight
Posts: 338
Joined: Sun Jan 15, 2006 11:07 pm
Location: Montreal, Canada

Post by ratio »

We do not use stop order either, exactly for the reason that Chris Mention.

Most of the day you worksheet is blank, no work to do.

Plus in our historical testing, it was demonstrated that in average we had no real advantage in using it. This is true as long as you use proper money and risk management in your position sizing.
SimJimons
Senior Member
Senior Member
Posts: 31
Joined: Mon Jun 20, 2011 9:59 am

Post by SimJimons »

I can't really understand why so many people use stops in their trendfollowing systems?! Stops don't provide an edge, they just change the distribution of returns. And in a world with transaction costs they not only change the distribution but actually end up eating into your capital. In addition, it's a hassle figuring out how to re-enter the trade.
AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2071
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner »

I think one must take great care in the use of the word "stops". And one must relate and distinguish a stop from a system exit. Take the built in TMA. It has both stops and system exit points. The stops are based on ATRs and in my view are essential for both position sizing, risk management and safety. The exit points define when the trend is likely to be over (in terms of the system and its parameters).

So, the question in my mind is not whether to use stops but whether those stops are placed in the market. Or CAN be placed in the market. Is a mechanism which says "exit at tomorrow's open if the price is below X at today's close" a stop?

Well, perhaps not and it is certainly not something one can enter in the market. But I would nonetheless consider myself to be using a stop by having such a mechanism. A question of semantics perhaps.
sluggo
Roundtable Knight
Roundtable Knight
Posts: 2987
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo »

Trading Blox provides a pair of systems, Donchian-using-Stops and Donchian-using-MOO. One of them enters and exits using stop orders exclusively, and the other enters and exits using Market-On-Open orders exclusively.

If someone wanted data rather than the unsubstantiated anecdotes and opinions of strangers, they might consider trying out this system pair at a half dozen different "speeds" (different durations-of-average-trade) to compare the performance of the two systems. You may wish to look at average performance (CAGR, Sharpe, Rsquared, K-Ratio, etc) AND you may also wish to look at the extremes too: 10 biggest losing trades, 10 biggest winning trades, 10 biggest losing months, 10 biggest winning months. The extremes are times when the all-too-human trader's emotions are inflamed the most. Perhaps if systems A and B have identical average performance (CAGR, Sharpe, Rsquared etc), but one of them has significantly smaller (10 biggest losers) than the other, you may prefer it for emotional reasons. (As opposed to analytical reasons). A and B have identical effects upon your long term profitability but significantly different effect upon your psyche. Does that matter, to you?

You may wish to carry this further and gather some data on Toby Crabel's Opening Range Breakout trading system idea, two different ways. Variation #1 enters and exits using Stop orders, and Variation #2 enters and exits using Market-On-Open orders. You may discover that quick systems like ORB, which hold trades an average of 5 days or less, have a very different sensitivity to order type than systems which hold trades an average of 40 days or more. In which case you may discover that the opinion "Stops are Superior" (or the inverse opinion) is neither absolutely true nor absolutely false; you may find that it is true at some speeds and false at others. In which case, why bother to even state it? "When you flip a coin, it will come up Heads! (sometimes)." Not terribly useful.

Be forewarned, it will take more effort to run these sorts of experiments and analyze their results, than to sit and read people's opinions. You may decide to allocate your finite research resources (time, programming budget) to other projects, whose potential payoff seems to be higher, in your opinion. Your call.
SimJimons
Senior Member
Senior Member
Posts: 31
Joined: Mon Jun 20, 2011 9:59 am

Post by SimJimons »

Take a simple moving average crossover system. In my view a stop-loss or a trailing-stop is redundant in such a system since the system already have a built-in "stop", i.e. when the MAs cross. Obviously it is always possible to come up with a system that performs somewhat better than the above "naked" system if we introduce some sort of stop (but it's actually more difficult than what at least I imagined). I believe, however, that this has more to do with luck than an actual edge. What we do know for a fact is that the distribution will change. If that is something we want, sure go ahead and implement a stop. But if the intention is to create a better system given some measure of goodness, I think it's easy to fool oneself.

I'm not too familiar with Trading Blox jargon, but when I'm talking about stops I don't mean "stop orders" but rather "stop losses". Sorry if that was unclear...
AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2071
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner »

SimJimons wrote:Take a simple moving average crossover system. In my view a stop-loss or a trailing-stop is redundant in such a system since the system already have a built-in "stop", i.e. when the MAs cross.
An additional "stop" (trailing or otherwise): (a) can make it considerably easier to it limit one's loss to an approximation of what one was contemplating when sizing the initial bet; and (b) especially in longer term systems, waiting for the crossover can cause some pretty painful and unexpected results (at least in my testing experience).

But of course we all have our views, do our testing and make our own choices.
SimJimons
Senior Member
Senior Member
Posts: 31
Joined: Mon Jun 20, 2011 9:59 am

Post by SimJimons »

"But of course we all have our views, do our testing and make our own choices."

And that is what makes it so fun :)
Post Reply