Systematic Trading research and development, with a flavour of Trend Following
Au.Tra.Sy blog – Automated trading System header image 2

The State of Trend Following report – Draft V0.2

April 7th, 2010 · 12 Comments · the State of Trend Following, Trend Following

State of TF

Another month brings you an updated version of the State of Trend Following draft report (V0.2) – in which we check a collection of standard Trend Following systems over a portfolio of diverse futures. The idea being to monitor how Trend Following is doing as a whole, as an alternative to tracking the Trend Following Wizards performance.

Report Changes

As mentioned last month, this is very much Work In Progress. Not many improvements to this version of the report compared to last month’s first draft.
The main addition was the inclusion of uber-classic Donchian system, which leaves us with the following system combination:

  • Bollinger Band Breakout
  • Donchian System
  • Moving Average Crossover system with ATR-based stops
  • Triple Exponential Moving Average

Each system being tested over a short, medium and long timeframe. Details of the system rules, parameters used and instruments traded can be found at the end of the post.

March Result

A composite index of all strategies tested has also been added to the report. The overview that this gives us is a positive month for March 2010: +2.2%:

Trend Following systems performance (01-Mar-2010 to 31-Mar-2010, base 100)

Trend Following systems performance (01-Mar-2010 to 31-Mar-2010, base 100)

The number of lines on this chart makes it slightly hard to read, so please find below the end value of each system:

System Final Value
BBO-20 100.30
Donchian-20 98.63
MA-10-20 104.44
TMA-10-20-50 104.94
BBO-50 105.52
Donchian-50 104.28
MA-20-50 105.25
TMA-20-50-200 103.59
BBO-200 100.76

The most profitable systems for March seem to be the medium-term ones.

The year through the Composite Index

To give some background, here is how the composite index looks so far for 2010 (no rebalancing in the index):


It seems like the downward trend has bottomed around the end of February and to turn around and pick up a bit of steam. It will be interesting to see how this compares with the Trend Following Wizards performance

Appendix: System Details

System Rules and Parameters

All the systems were tested with the same simple position sizing rules of 1% per new trade. No other Money/Risk Management rules were used. No slippage was considered and a $15 RT commission applied.

The system rules are detailed on the Trading Blox online documentation.
The MA Crossover system was used with moving average pairs of 10-20, 20-50 and 50-200 days. The stops are set at 5 x ATR.
The Bollinger Band system is the classic use of the Bollinger Bands with entries taking place at Breakouts. The parameters used were 20, 50 and 200 days with 2 standard deviations.
The Triple moving Average system was used with moving average triplets of 10-20-50, 20-50-200 and 50-200-800 days. The stops are set at 5 x ATR.
The Donchian System is a simple version (with no Trade Direction filter) with channel lengths of 20, 50 and 200 days for entries (and 10, 25, 100 for exit). The stops are set at 5 x ATR.

Portfolio Instruments

The instruments included in the simulation are a semi-diversified list of futures, contained in the All Futures Trading Blox set:

CSI Num. Futures Contract
AD Australian Dollar-CME(Floor+Electronic Combined)
BP British Pound-CME(Floor+Electronic Combined)
C Corn-CBT (Floor+Electronic Combined)
CD Canadian Dollar-CME(Floor+Electronic Combined)
CL Crude Oil-Light-NYMEX(Floor+Electronic Combined)
CT Cotton #2-NYCE
EC Euro(Floor+Electronic Combined)-CME
ED Eurodollar-3 Mth-CME-Globex(Floor+Electronic Combined)
EM LIBOR(1Mth)-CME(Floor+Electronic Combined)
FC Cattle-Feeder-CME(Floor+Electronic Combined)
GC Gold-COMEX(Floor+Electronic Combined)
HG CopperHG-COMEX(Floor+Electronic Combined)
HO Heating Oil #2-NYMEX(Floor+Electronic Combined)
HU Gasoline-Reformulated Blendstock(Combined)-NYMEX
JY Japanese Yen-CME(Floor+Electronic Combined)
KC Coffee-CSCE
LC Cattle-Live(Floor+Electronic Combined)-CME
LH Hogs-Lean(Floor+Electronic Combined)-CME
MP Mexican Peso(Floor+Electronic Combined)-CME
NG Natural Gas-Henry Hub-NYMEX(Floor+Electronic Combined)
S Soybeans (Floor+Electronic Combined)-CBT
SB Sugar #11-NYCE(Floor+Electronic Combined)
SF Swiss Franc-CME-(Floor+Electronic Combined)
SI Silver-COMEX(Floor+Electronic Combined)
TY T-Note-U.S. 10 Yr w/Prj A-CBT(Floor+Electronic Combined)
US T-Bond-U.S.-CBT (Floor+Electronic Combined)
W Wheat-CBT (Floor+Electronic Combined)
Related Posts with Thumbnails


12 Comments so far ↓

  • Pumpernickel

    You have chosen an approach which minimizes your personal effort: (1) use prewritten “sample” systems as-delivered from a software vendor; (2) use preselected “sample” data from the same software vendor, to construct a portfolio of USA-only futures markets; (3) use system parameter values obtained via rectal extraction, which appear, to you, as though they ought to represent short-term, medium-term, and long-term trading.

    What is convenient and minimizes effort, might not be representative of the universe of professional traders you seek to benchmark. It seems unlikely that CTAs and fund managers would compete for investor dollars using out-of-the-box vendor supplied trading systems.

    Your predecessor R. Spurgin constructed a trend following benchmark using completely different systems that are not presupplied by any software vendor; he wrote them himself, and disclosed their rules in his study: ctatraderindex.pdf He also chose the parameter values of his “short”, “medium”, and “long” term systems, to maximize correlation between the returns of his system-suite, and the managers he wished to benchmark. See his exhibit 3. That’s pretty good science, and an admirable example for you to follow.

  • Jez

    Hi Pumpernickel,
    Thanks for the link to that paper. I have been collecting a few papers dealing CTA benchmarks as I came across them but never saw this paper, which sounds interesting.

    Re: the “minimum effort” point, I can only say: “yessir, guilty on all charges” and not hiding from it ;-)

    Unfortunately time is only a finite resource and with other commitments and projects on my plate, I cant dedicate as much time as I’d like to everything. I suspect it will take me a few months to get to the finished product (which I expect to be largely different from this current draft/0.2 version).

    Also, I still think that a large portion of Trend Following performance can be explained/replicated with “fairly simple” strategies. If you have not had the chance, please consider reading the Beta of Managed Futures paper that I linked to in my recent Betafication of Alpha / Commoditization of Trend Following? post. It shows how a combination of simple un-optimized Donchian systems correlate very well with CTA returns. It makes a good point about using simple systems.

    As explained at the time of publishing the first report, I could keep this project in the back-burner until it has been perfected but decided to get the ball rolling by publishing a very rough draft; and the strategy seems to be working since I am receiving valuable comments like yours (+ some people already find the report useful in its draft form).

    Thanks for the other points/remarks also. I am definitely making a note of them and planning to take them on-board at some point… when time allows.


  • athfutures

    Just a side comment to pumpernickel’s…. I think the concepts in your blog(s) are a great mechanism to get the ball rolling for further individual R&D. Even though your concepts are not always fully developed, I find them useful to kickoff my own research pathways as I’m sure they do for other readers.

    Well done and keep em coming.

  • Jez

    Thanks for the kind words athfutures.
    This is one of the main ways I intend the blog to work for my own individual R&D: a balanced symbiosis where the research feeds the blog, and the blog feeds the research (through forcing me to look into different topics more in detail as well as getting good feedback and interaction with readers for example).

    Even better to also hear readers say that the blog helps with their own research!

  • Anarchus

    Off topic but thinking about your past discussions of trend following, optionality, serial correlation, fat tails and MPT.

    When real world markets violate the math assumptions underlying MPT, it isn’t just that the tails are fat or that there’s serial correlation. The multiplicative anomaly is that serial correlation spikes when you’re in the fat tails, so that extreme returns are quite non-randomly generated.

    I’ve got a quote from Myron Scholes in my files somewhere that has him observing that option buyers generally lose money steadily and predictably except for the brief periods where they make lots of money. If it’s true that trend following has a lot of the statistical properties of buying options, then I suspect it’d be true that for naively-designed trend following systems the P&L would follow a similar path.

    What this suggests to me is to seek out unique markets at any point in time where serial correlation is above average and rising (I’d guess that the volatility of the underlying would also be above average and rising) and trade those and leave the uninteresting markets alone.

    Just a thought.

  • Anarchus

    Ah, forgot a footnote – I think you know this, but leptokurtic is an imprecise definition of fat tails because there are disproportionate datapoints clustering around the mean as well as in the fat tails. While it’s the fat tails that produce the high positive or negative results in buying options (and maybe in trend following strategies), leptokurtosis is a slightly mixed blessing due to the higher peak.

    Antics with semantics, maybe.

  • Jez

    Anarchus – thanks for sharing your ideas and thoughts. That’s great!

    This idea of measuring serial correlation at extreme price levels is something I want to investigate further and actually a mini-project I started working on (started thinking about it here: ), but somehow left in the back-burner (maybe a good idea for a post next week if I find the time to complete the coding at the weekend).

    Ultimately, the idea I want to investigate is: is it possible/useful to track and measure these price data characteristics to use as a filter/portfolio selection tool to identify markets with more/less potential?

  • RiskCog

    Jez and Anarchus I like this line of thinking “let’s develop an indicator to tell us when the MPT based portfolios are going to start losing to us” ;>

    MPT stated simply is “let’s assume market returns can be adequately modeled as random processes and follow the math where that leads”. The previous discussion of the Leverage Space concept started the same way “assume … random processes…”.

    If markets are not adequately modeled as random processes then it is reasonable to stop doing that. I personally don’t use statistics at all in my trading systems or portfolio optimization.

    Back on topic: would be useful to see the historical equity curves for the example trend following systems. This at least wouldn’t require a bunch more system development work on your part ;-)

    Also it would be interesting to see a performance ranking of all notable fully disclosed systems. Has anyone seen something like that?

  • Jez

    Hi RiskCog,
    I did put the historical equity curves in last month’s first draft. I did not want for the report to be a huge confusing collection of charts and did not include them this time. I think if this report becomes a regular monthly feature, I might regularly (ie every 3-6 months) put the “long-term picture” charts.

    Haven’t come across a “performance ranking of all notable fully disclosed systems”…

  • Caroline

    Surprised you don’t know about Futures Truth, which is a magazine that publishes performance ranking of all notable systems, vendor-sold and free. They run the tests and publish the tables monthly. Their website lists 156 systems (A-L) plus 120 systems (M-Z) currently ranked. The URL is: . They’ve been doing this for more than ten years.

  • Sniperov

    Tried accessing the link that Pumpernickel posted in the first comment, but it is no longer accessible – any chance you have a copy handy?

  • Jez Liberty

    I’m afraid I do not have a copy saved. A quick google did not seem to yield that specific pdf.. But I found this page, which might be of interest:

Leave a Comment