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

State of Trend Following in August

September 5th, 2011 · 5 Comments · the State of Trend Following

State of TF
August – What a volatile month! Instead of the more usual quiet Summer market lull, markets got really nervous.

And this also shows in the results of the State of Trend Following – albeit with a positive outcome, completely opposite to “traditional investments”.

Let’s cut to the chase with this month’s results:
August return: +13.46%
YTD return: -6.71%

Despite a strong month, the report is still in the red for the year. Below are the more details results:

Detailed Results

There is some very strong volatility from several systems – I will probably run a check on leverage levels at the end of the year to readjust them if needed (although the goal with this index is not to derive an absolute figure but rather get a feel for the general direction and relative volatility of a diversified trend following strategy).

And in tabular format, showing the normalized returns for each strategy and the composite index, as well as the YTD figures:

System August Return YTD Return
BBO-20 25.15% -33.99%
Donchian-20 31.75% -20.54%
MA-10-20 12.38% -11.23%
TMA-10-20-50 7.47% -17.97%
BBO-50 7.31% -14.92%
Donchian-50 9.51% -21.72%
MA-20-50 6.14% -16.53%
TMA-20-50-200 13.16% 2.5%
BBO-200 3.11% 6.59%
Donchian-200 4.91% -2.84%
MA-50-200 25.62% 23.73%
TMA-50-200-800 15.06% 26.41%
COMPOSITE 13.46% -6.71%

Composite Index for 2011

Below is the performance of the average of all system/timeframe combinations used in the report for the year 2011:

Still under-water, but a nice bounce from the lows experienced in July.

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 trade friction (slippage or commission) was applied. No return on margin is added to the system performance

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/position sizes are set at 2x, 3x and 5x ATR respectively.
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/position sizes are set at 2x, 3x and 5x ATR respectively.
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/position sizes are set at 2x, 3x and 5x ATR respectively.

Portfolio Instruments

Covering over 50 instruments across Equities, Interest Rates, Currencies, Agriculturals, Metals and Energies, from around the world, the portfolio contains the following futures (CSI Symbols): AD, BP, C, CC, CD, CFC, CL2, CT, CU, EBL, EBM, EBS, ED, EOX, ESM, FC, FEI, FFI, GC, HG, ICL, IND, JK2, JP2, JP6, JR2, JRB, JTI, JY, KC, KPO, KTB, LC, LGO, LH, MFX, MP, NG2, RA, RS, S, SB, SF, SI, STW, SXE, TRY, US, W, YM, YTC .
Click here for a tabular view with description and exchange information.

Result Normalization

The system performances are adjusted for volatility to normalize the results. See why and how here.

Related Posts with Thumbnails


5 Comments so far ↓

  • Monday links: Euro stress | Abnormal Returns

    […] How did trend following systems do in August?  (Au.Tra.Sy Blog) […]

  • Michael Harris

    Hi Jez,

    Thanks for the report. However, I still do not see the need for result normalization. Let us assume that Bob trades the MA-50-200 system for example and Alice trades the BBO-200 system. We intend to rank then according to YTD return on a 100K account. The normalization you are using is based on calculating a factor by dividing the system drawdown by 50% and then use it to multiply the YTD return.

    However, the result is not the actual return of these people. Actually, we would like to rank their performance based on the actual results of the different systems they use and not attempt to normalize them. I would do the normalization, for example, if I was to use all those systems together to trade and its purpose would be for allocation, not for comparing the individual system results. What do you think?

    Looking forward to your reply.

  • Jez Liberty


    Maybe this is just confusion on the goal of this report.
    The way I intend for this report to be used/useful is to give an overview of Trend Following’s relative performance month after month (i.e. it is more intended to be an “index” than a system comparison). The main reason why I use a dozen of system/timeframe combination is just to cover several trading styles and horizons (and get some diversification). But the idea is not to “rank” systems between each other. Hence I completely agree with your comment of “I would do the normalization, for example, if I was to use all those systems together to trade and its purpose would be for allocation, not for comparing the individual system results.“.

    Because I am mostly interested in the end figure for the whole index/portfolio of systems, I want to give each system a fairly equal weight and the normalization I apply is just a very simple one – because I do not feel there is a need for a more complicated approach for this index.

    In any case, any traders wanting to trade any of these 12 systems independently or in combination will have to decide of the leverage to apply to the system(s), which will directly impact the Max DD figure – so there is not ONE result for the MA 50-200 system but a theoretical infinity, depending on which leverage is applied to it. If I was ranking systems directly as part of the report, I would most likely publish performance figures that are _less_ dependent on leverage (MAR, Sharpe, etc.)

    Hope that answers your questions…

  • Michael Harris

    Hi Jez,

    I get your point. Thanks for the reply. However, I’m still troubled by the method. I believe that when using a mathematical model it must also justify some boundary conditions, otherwise it is maybe ill-conceived. Thus, let us assume we have two systems in the index, as follows:

    System A: Return = 0%, Drawdown 50%
    System B: Return = 100%, Drawdown 1%

    The coefficients according to your method are:

    For system A: 50/50 = 1
    For system B: 50/1 = 50

    The normalized returns are:

    System A: 0%
    System B: 5,000%

    If I am correct on the method, what is the meaning of this normalization? For one, it appears that systems that return 0% will not be penalized regardless of their drawdown. That is not fair. Then, systems with exceptionally good return and very small drawdown will unrealistically skew the index return.

    It appears that this normalization method works well in the vicinity of small perturbations around 50% drawdown and returns away from 0. But returns close to 0% will skew the results a lot.

    All these provided that I understood you method, of course.


  • Jez Liberty

    Michael, how would you suggest to adjust the position size/leverage for each system?
    Picking a random value does not seem to make much sense to me and it feels that the normalization done here is something that most traders would apply one way or the other when trading multiple components together (unless you go into something more sophisiticated like optimizations, ie LSPM, or MVO).

    Your example is fairly extreme (and I wish I could find something like system B!) but in essence the normalization you applied is “correct” and any trader would overweigh system B when trading both together, and so does the index. In any case there are no extreme discrepancies in the results of the systems in the State of TF and I feel the method works reasonably well for what I intended it for (making sure that higher leveraged/volatile systems were not overweighed in the index calculation, and vice-versa with lower leveraged/volatile systems).

    Also note that this is a simply a “calculation shortcut” (which I find acceptable as this is not back-testing but simply index calculation). I could have gone back to each system in the report and adjust its position size to obtain a MaxDD of 50% in the back-test. This would have done pre-normalization instead of post-normalization. (precision just to be sure: I do not recalculate the normalization factor every month – I ran a back-test for 20 years for each system and calculated a “normalization” factor so that MaxDD was equal to 50% – and every month I apply the same factor to each system).

Leave a Comment