Strong May Results for Mechanical/Trend Following Systems

[akhoury]

I wanted to throw something out there and get some feedback from the community members that have a point of view. Over the recent past (maybe 12-36 months) a lot of the Trend Following systems have had a fairly difficult time. But, I am seeing evidence that things may be changing a bit. Take a look at this data from Jez Liberty’s website ( http://www.automated-trading-system.com/ ):



From my point of view, this is important news. Coupled with anecdotal information gleaned from other Trend Following/Mechanical Trading sources, mechanical systems are again starting to outperform. For those souls, like me, that feel that some of the central banking intervention has suppressed natural trends, but also believe that underlying trends eventually can only be suppressed for so long, perhaps there is a turn at hand. While one month of data can mean nothing (or represent important signal) take a look at this Year to Date chart from the same site:



From a purely technical read of the performance curve, it seems to me that a base has been built and we have a fresh break-out. Importantly, a broad group of shorter term mechanical systems have done well in reacting to the rapid reversal that we seem to see quite often in the post Lehman type markets.

For those interested in the individual system performance here they are:



For the most part, all of these systems ship standard with Trading Blox and all of the rules of the systems can be found for free in the Trading Blox documentation and more information can be found at Jez’s site referenced above. This indicates to me that solid, simple, robust systems can still drastically outperform and that identifying the structures and timeframes that are profitable in the current environment is doable, profitable, and identifiable.

I am curious to get some opinions as to what others feel works best in the current environment, and if more recent data deserves more weight within the confines of interpreting back-tested results.

I would like to thank Jez for letting me quote liberally from his work and want to mention that people who would like to discuss this can contact me at amin@tradingblox.com and folks that would like to work with Jez can find his contact information in the preferred vendor section of the System Development portion of our website.

Thanks ack

[Johnedoe]

May was a very good month for me as well.
I use an uncluttered 2 MA crossover system that is a bit different than what you will find in TB.

[rhc]

I am curious to get some opinions as to what others feel works best in the current environment, and if more recent data deserves more weight within the confines of interpreting back-tested results.

Amin,

If we look at Jez’s table of Individual system performance we see that it’s the shorter term systems that really shone (These are the first 4 in Jez’s table)
The longer term systems were as dull as dishwater (These are the last 4 in Jez’s table)

The month of May was excellent for the short termers and not so excellent for the long termers

This is also illustrated in a quick comparison of the same system but with parameters adjusted to produce short, medium & long term trades (see image below)

So far, 2012 has proved a very good year for shorter term systems.
Perhaps we should change to a shorter term timeframe. Results have been excellent

[rhc]

. . . . . . but wait.

How did these short term systems fare prior to 2012 and how did the longer term systems fare prior to 2012 ??

If we refer to the image below which shows the exact same systems but with parameters adjusted to produce short, medium & long term trades.
It would appear that since year 2000 the Long termers have performed admirably whilst the Short termers have left much to be desired.

- Can we draw conclusions from this?

- Is long term trading the way to go over the long run, or are we entering a new era where short term systems are about to outperform.??

- If as Amin suggests; “more recent data deserves more weight within the confines of interpreting back-tested resultsâ€

[LeviF]

Here's what my 2012 looks like. One more lousy day and I am pulling the plug - wish me luck tomorrow.

[rhc]

Levi,
What is your average trade hold time ?

[Turtle40]

Levi,

Your monthly returns are interesting. Do you have a high risk/trade? Or are you trading a very correlated portfolio?

I designed my system to be more conservative, I don't have any losing months (yet) greater than 20%, conversely I don't have any winning months greater than 24%.

I see you are in the area of 40/60% losing/winning months. Is this creating very large drawdowns?

Anyway, I wish you luck as the last few months appear pretty tough!

[LeviF]

Levi,
What is your average trade hold time ?

According to my trade log its 280 days, which includes early exits for profit taking.

[LeviF]

Levi,

Your monthly returns are interesting. Do you have a high risk/trade? Or are you trading a very correlated portfolio?

I designed my system to be more conservative, I don't have any losing months (yet) greater than 20%, conversely I don't have any winning months greater than 24%.

I see you are in the area of 40/60% losing/winning months. Is this creating very large drawdowns?

Anyway, I wish you luck as the last few months appear pretty tough!

Yes, Yes, Yes, Thanks.

[akhoury]

rather than discuss any particular person's returns, i was hoping to focus on the shorter term nature of the systems featuring outperformance, what it may mean for our business, what it may suggest about adjustments we can make, and if it is perhaps signal over noise...these topics, if well examined can have a benefit to the entire community even if the evidence leads to discarding the theory...

in particular i was wondering if anybody has applied the theory shorter term systems to the post Lehman period in any significant way?

one final word on the nature of this discussion...the power of this forum is the user group and their deep information base about mechanical trading...by asking these questions (and attempting to answer them in public) it is my hope that the collective experience resident here can become an edge for this user group...

[Chelonia]

We allocate our equity over short, medium, and long term systems. Just another diversifier. No massive up months, but also no major drawdowns. Has given us a nice middle ground in terms of risk/return.

[Chelonia]

In light of your question I want to repost an earlier article I wrote.

At the brink of Monetary Failure
Debts keep on multiplying exponentially by interest. It´s like a bucket of water with a big hole in it which is perpetually expanding, so long as we can borrow enough obfuscated money (water) back into circulation, to keep the bucket sufficiently full, that surviving industry can pay through the hole, and borrow further.

The water is our circulation; and so it is the principal of our purported debts to a banking system which merely publishes further representations of our promissory obligations to each other. But under a purported banking system then, "water" isn't just water (principal) ― it's an unwarranted (and falsified) obligation to perpetually pay principal and interest out of the hole, to the purported banking system.

Here is how the hole gets so big that we inevitably fail under terminal sums of falsified debt:

If we circulated our promissory obligations without a purported banking system, the obligation would only be to pay and to retire principal from circulation ― and all our actual debts (which are actually only to each other) could be fulfilled.

But because we allow a purported banking system to obfuscate these obligations into falsified obligations to pay principal and interest out the hole (from a bucket filled with only some of the remaining principal), and because we have to keep the bucket full to remain able to do so (because we have to sustain sufficient surviving industry), the principal which gushes out the hole must return to the bucket by borrowing the principal back again.

Any fact we have to borrow so much principal back into the bucket as equal, new debt, therefore makes it impossible to pay down any prior sum of falsified debt.

Yet on the other hand, because the interest which must return to the bucket is borrowed back as new debt; therefore the sum of debt perpetually increases by so much as periodic interest on an ever greater sum of debt.

This is how and why the hole perpetually gets bigger: ever greater sums of principal and interest are falling through the hole, to return to the bucket in such a way as increases a related sum of debt ― even an inherently escalating rate of ever greater sum of periodic interest on an ever greater sum of falsified debt.

At the same time, the capacity of surviving industry is finite: it can only service so much debt.

Thus when the hole (soon) gets so big that more water is exiting than industry can further afford to return to the bucket by (compulsory) further borrowing, the level of water in the bucket begins to fall ― and less industry can be sustained by the remaining water... ever more of which water yet, is solely dedicated to servicing debt... as opposed to sustaining industry.

Even less surviving industry (and even less possible surviving industry) therefore are required to maintain the falling level of the bucket.

Thus in each subsequent period, the level in the bucket falls itself at an escalating rate, until very soon, nothing remains to sustain industry; and thus in a failure which is inherently terminal, no new industry either can afford to survive the falsified sum of debt which has been wrongly multiplied into the unwarranted possession of a purported banking system.

It is a vicious circle. This worked for a long time because there was always some industry (internet/housing) able to loan a lot of new money (at interest) to maintain/re-inflate the vital money circulation. But there will always be a breaking point whereas a critical mass will have lost its credit worthiness and are not able to borrow further. Starts to shows its terminal face on the individual level, then peripheral banks, then governments...

So, what we are witnessing now is a denial of these terminal ramifications by the very perpetrators, who only publish further representations of our promissory obligations; who deny every real creditor interest; who thus do not even have any actual, legitimate property at stake which might (falsely) be argued to justify interest; and whose imposition of interest therefore is inevitably terminal.

QE therefore is itself a proof of this inevitable failure, for the obfuscation of the purported banking system has only forced the purported banking system to pour the water back into the bucket which is impossible for the un-assenting subject to pour back into the bucket.

No advantage or possibility of saving the system yet exists, because all QE accomplishes is to artificially extend the lifespan of a terminal system some while ― only so long as it is possible for failed and failing industry to absorb new water by production which has been made impossible by the old water.

Being at that critical terminal point now the changes from deflationary to sustaining the level or inflation (or the expectation thereof) are causing ever more rapid changes in market direction (increased frequency).

[Toosday]

This has turned into a good discussion, thanks to all who have shared something. My research basically backs up the conclusions about short and long term systems. IE May was good for short term systems but brutal for long term ones. Conversely the past decade was a struggle for short term systems while long term ones produced decent returns.

To add something, the one thing I find as a major problem, is not diversifying among time frames but if finding robust systems that can produce returns short selling. The risk adjusted return for short (sell) systems, long term or short term from what I have found over the past decade, are roughly 1/10th of long (buy) ones. Obviously I would never expect to get this ratio to 1-to-1 but I do believe that getting to about a third would make me feel much better about my systems.

If anyone can confirm this, criticize this or offer some advice, I would appreciate it.

[Jez Liberty]

Thanks Amin for the mention and bringing the discussion here.

From my (limited) experience of running the report every month, I have not seen a trend in shorter term systems out-performing the longer-term ones post-Lehman (ie sometimes it is the other way around - I have not run any "quantified" analysis of this on the report, just my observation).

I think it is a "challenge" (that I havent cracked) to "predict" ahead of time what timeframe will perform best and position your trading accordingly. I think rhc makes a good point in showing that shorter-term systems had worst performance in 2009-10-11 (which is quite similar to what I have been seeing on the state of TF index). How would you know to switch back to shorter-term in 2012?
Apart from the proverbial crystal ball, trying to identify market environments/trading regimes with "meta indicators" (inflation, volatility, etc.) sounds like an appealing way of doing it (with the potential pitfall of the regime identification backward-looking). I have not researched this much at all but I believe this is an interesting topic for this discussion (and for system research).

Meanwhile, I'm with Chelonia: I think at a basic level including several timeframes and passively trading them all adds a layer of diversification (if you can afford to stretch yourself over several timeframes): different "baskets for your eggs" with the potential for lower correlation/smoother equity curve.
That was the initial main reasoning to include different timeframes in the index.

[wguan]

I've read a passage from css analytics yesterday late at night about momentum lookback periods and it goes over the same idea of whether to take 12 months or shorter time frames. From what I've read, its a matter of trade off if considering only whether to trade the longer time frame systems or shorter time frame systems. Trading the long term systems, you simply miss the short term reversals while trading the shorter term systems you will increase the whipsaws and trades where commission and short term losses will eat away long term equity.

One solution that was mentioned was kinda like what Jez Liberty said, incorporating both time frames. Another one I found interesting was to apply a variable weighing, which varied the allocation to short term systems under certain conditions which deemed favourable for short term systems. And increase the allocation to long term systems under favourable conditions.

Just my thoughts,

[rhc]

We allocate our equity over short, medium, and long term systems.

I think at a basic level including several timeframes and passively trading them all adds a layer of diversification (if you can afford to stretch yourself over several timeframes

One solution that was mentioned was kinda like what Jez Liberty said, incorporating both time frames.

Yes, this is all good advice, however it is mostly applicable to those with deep pockets who are able to afford this type of diversification.
As Jez says above, “if you can afford to stretch yourself over several timeframesâ€

[Moto moto]

For those that cannot do this, the question still remains open.
Do you trade a shorter term system based on the last year’s results or do you trade a Long term system based on the last decade’s results ?

My solution for what it is worth.....I trade short term highly leveraged system in what I deem moving/trending for that time frame - this is mechanical and rules based on technical price action (not so much indicators) (I am trying to automate more of it) but involves a fair bit of discretion as to when to turn it on ... which instrument etc. Not diversified (works really only in highly liquid FX, bonds (to test), equity markets and not true trend following but in a similar vein - looking to get onto trends, ride them - i spread my money between two variations of the same theme - one takes profits, the others let it ride, but the short term nature means the trend may be small. (automation currently a WIP)

For longer term stuff I give my money to others.....it was more economical that way.

[sluggo]

... all good advice, however it is mostly applicable to those with deep pockets who are able to afford this type of diversification ...

... For those that cannot do this, the question still remains ...

Here is one way to simultaneously trade numerous different timeframes without having deep pockets: viewtopic.php?t=8606

[rhc]

Nice one!
Thanks sluggo

[TK]

On 1 October 2010, I started trading a tiny forex account. After lots of back- and walk-forward testing, I decided to set up a small account and go after absolute returns of 50%+ annual returns with occasional 80%+ drawdowns. I figured out it would be psychologically easier for me to start with a small amount of money and accept ridiculous volatility of the equity curve than trying to contain drawdowns with a bigger account.
Without divulging my system rules, I trade a variation of a trend-following breakout system. My optimum parameter values for entries and exits on 30 Sept 2010 were 11.0 and 0.4. If I had traded this system without changing these entry and exit parameter values, here’s what my results would look like now:



I would have lost 90% of my equity and I would be at -92% DD.

But I believe that markets change and I follow Bob Pardo’s walk-forward technique. After lots of testing and thinking, I decided back then that my optimum in-sample test window is 5 years and my optimum out-of-sample trading period is 1 month. That means that every month I re-optimize my system based on the best my-figure-of-merit results for the past 5 years. Trading that way, I have achieved these real-life trading performance (my current entry and exit parameter values are 7.0 and 0.1, respectively – the walk-forward testing has shortened my timeframe considerably over time) :



Please note: these results are far from great. I have lost close to half of my starting equity. Still, I am pretty confident I am going to crawl out of the current DD and win big going forward. I believe that the post-2008 period is extraordinary in terms of market manipulation by central banks and politicians so I am not that concerned that drawdown figures for many trend-following systems are much worse than whatever backtests indicate. I believe that you have to adapt to the "current" market environment and the walk-forward technique as described by Pardo is as close as you can get to what real-life trading looks like. Please note that while I am still in a deep DD, in 2011 and 2012 my walk-forward results have been better than theoretical "don’t change your system ever" backtest results. Also, I am in a -57%DD now vs. the -92%DD theoretical backtest DD.

Other than walk-forward vs. backtest type of trading, there are two other factors that can help your trading performance that I consider crucial. These are: (1) betting strategy – while it is extremely difficult to improve upon the standard fixed-fractional betting strategy, it can be done provided you’re a private trader and don’t trade OPM – ironically, you have to have guts to do the reverse of what Eckhardt taught the Turtles; (2) open risk management – I have found that scaling out at a pre-defined total portfolio heat thresholds can prevent you from going bust at times of extreme market volatility.

And yes, I have had a great May and YTD results (but I would be negative for the year, had I stuck to my original parameters).