Au.Tra.Sy blog - Automated trading System Au.Tra.Sy blog - Automated trading System
A practical Guide to ETF Trading Systems

A-Practical-Guide-to-ETF-Trading-SystemsIt’s been four months since the last book review. I wanted to make book reviews a more frequent feature of the blog. Problem is I often start few books at once but struggle to finish them.

The book for today’s review is A Practical Guide to ETF Trading Systems by Anthony Garner, who is a Trend Follower and fund manager (and user of Trading Blox). The book is aimed at investors wanting to investigate rule-based trading and explains, with numerous examples, the path to designing, building and testing mechanical trading systems, with an emphasis on how this can be applied to the ETF instruments.

The book is a quick read (170 pages) with lots of charts to illustrate the testing.

Introductory information

The author starts by presenting systematic/mechanical trading by the way of a very simple system example and the track records of systematic funds such as Renaissance and many of the Trend Following Wizards.

The next chapter deals with Data. Stressing the importance of good quality data, the author lists a few sources to get historical data to be used as a proxy for the (fairly recent) ETFs. An example of investing with/without dividends makes a compelling case for trying to source that data as well (over slightly more than one century, including and reinvesting dividends nearly double the annual return – and with compounding, that means multiplying your end balance by a factor of over 150!). Finally, Garner touches on commodities and the need to construct indices including less obvious factors such as roll return.

A quick chapter on Backtesting software follows on (Garner uses Trading Blox throughout the book) leads to the final chapter of Part 1 on Strategy and System design.

Aspects such as optimisation and curve-fitting are discussed as well as the need for realistic testing assumptions, in simulating slippage and transaction costs – which can also have a dramatic impact on the simulation results.

Part 2: The meat of the book

The second and final part introduces several investing strategies including systematic (or rule-based) trading. As highlighted by the author, asset allocations is a one of the most important source of return. The good news is that most major asset classes (and more) are available in the form of an ETF. Various asset allocations are tested using the different systems throughout this part.


As a base for comparison, various statistics (CAGR, drawdown, etc.) are calculated for different Buy and Hold portfolios (Equities-only, Equities with Bonds and Commodities, Commodities-only, etc.). One conclusion is that a diversified portfolio helps generate smoother and higher returns. Another point is how regular rebalancing is important to keep an appropriate level of diversification in the portfolio.

Garner uses the ill-named risk-adjusted return metric to compare strategies. Let’s just call this a semantic error (it is widly used throughout the industry), the astute reader should be able to mentally translate this to volatility-adjusted return. (sorry… a bug bear of mine: Variance != Risk)

Finally, an interesting side note specific to commodities and their equivalent ETCs for long-term holdings. The author demonstrates that because of negative roll yields (for markets in contango) and additional ETC sponsor fees eroding much of the returns, these instruments are not an ideal option for long-term holding.

BBBO system

The first rule-based system that Garner introduces is a Bollinger Band BreakOut. He describes the full system using rules, not only for entry and exit, but also for risk and position sizing, rebalancing, money management, etc. (as any good system should). The nice extra for Trading Blox users is that the system is made available on the forums over there.

The author runs comparisons with the various asset allocation benchmarks discussed in the prior chapter, which fairly conclusively demonstrate the superiority of this Trend Following system over any Buy and Hope Buy and Hold strategies. Playing with the system parameters also shows some form of robustness, with the performance staying fairly constant.

One point is made about trading short. It is recommended not to.
This is mainly in the light of bad performance of short-only strategies tested in the book. But the problem with this reasoning is that most backtests go back only to 1982! I believe the author might have fallen prey to a biased view of the markets, making macro predictions on the continuity of market conditions (suggesting that despite very good performance in 2007-2008 short-positioned systems will “go back to negative performance when stock markets recover”).

I have several problems with this:

  • Who knows if markets will recover any time soon? Without getting into macro-economics discussions in much detail here: since the 80’s, the world markets have been subjected to the greatest inflationary expansion in the last 100 years (ever?), helping support (create?) the major macro bull run (bubble?) in most asset classes, globally. I do not want to be making any prediction about where we are heading next, but who knows if the next 25 years will not rhyme with 1931-1955 (ie and/or give us a long “Japan-style” deflationary decline spiral).
  • The assumption has not been tested on earlier markets (mostly because of lack of data prior to 1982).

As a side note to the above: it would be interesting to measure and test the impact of a macro filter on a Trend Following system (ie something in the vein of: “favour long trades in the system when the macro indicators indicate a period of expansion and short trades during declines”). Maybe exploring a mechanized version of Schumpeter Business Cycles or Kondratiev Waves as a very long-term filter would be a worthy approach.

Increasing Returns

This is an interesting chapter addressing one of the shortfalls of ETFs versus futures as systematic trading instruments: leverage (or rather lack of). The new altered Money Management algorithm dictates to the system to concentrate the equity on the available signals, rather than separating the equity equally by instrument (and sitting in Cash where there are no signals). This is an interesting concept that deserves further investigation.

The performance improvement that this Money Management rule change generates is quite interesting. The conclusion is that it is possible to increase returns without increasing leverage.

Momentum System

The last chapter gives the same comparison treatment to a momentum system (buy strongest performing markets), which seems to over-perform the BBBO system (and obviously Buy and Hold as well).

Again, this system code is available on the Trading Blox forums.

After Word

A fairly short and pleasant read from a seasoned and real-life Trend Follower and Trading System Developer. There are a few ideas to take away, which is always appreciable. A bit on the expensive side (but I somehow managed to get it at nearly -50% on amazon…).

Related Posts with Thumbnails

arrow10 Responses

  1. 61 mos, 3 wks ago

    Thanks for the review. Out of curiosity what were some of the highest back-tested CAGRs and corresponding max DDs for the example systems?

    My experience developing ETF trading systems mirrors a few highlights from the book that you mentioned. Leverage is not as readily plentiful, so you have to look for a system with not just a good win-loss ratio etc but the system also needs to deliver acceptable return with the leverage available. Said another way you need to design up to Kelly instead of levering up to Kelly.

    The point about focusing your money on the signals that are available matches my research. As does the benefits of momentum type systems which are built around focusing.

    Since the author warns against playing the short side, I would be willing to bet a nickel that he doesn’t spend much time on volume indicators in the book ;-)

  2. 61 mos, 3 wks ago

    The BBBO system applied to a diversified portfolio yielded 14.12% CAGR with a MaxDD of 24% and the enhanced version (ie concentrating positions on available signals: 19.18% with MaxDD of 37.4%.
    You are right about the author not using volume indicators – I take it you found some good use of them?

  3. 61 mos, 3 wks ago

    Thanks for the additional stats.

    Yes, I do find volume to be useful and trade-able. The issue many trend followers have is that their models can’t handle the sharp price reversal that come at the end of a good short opportunity. You can see this reversal coming if you keep an eye on volume because trading volume falls off at the end of a move. Tim Ord has a book about this…

  4. kevin
    60 mos, 3 wks ago

    @ riskcog

    What about the comment on Tim Ord’s book that it’s merely a sales pitch for his Ord Volume, a newsletter that you have to pay for on his website?

    Is the book interesting enough without subscribing to his newsletter?

  5. 60 mos, 3 wks ago

    Hi Kevin, I found the book useful. I don’t subscribe to his newsletter so I can’t comment on its value. I also don’t happen to use “Ord Volume” exactly the way he lays out in the book. What I did instead is just take the general principle that volume will start to fall in aggregate as you get to the end of a move because fewer and fewer marginal buyers are driving the price.

  6. 55 mos, 2 wks ago

    You made a comment about the author not backtesting as far as 1982. Which I thought interesting for a few reasons, namely that I should be backtesting my trading system farther!

    Anyways, this book is about trading systems for ETFs which have only been around since 1993. (according to WSJ)

    So its OK to only backtest ETF trading systems to 1993 right?

  7. 55 mos, 2 wks ago

    In a strictly manner, you can only back-test ETFs (or anything else) from the date they start trading (ie 1993 for ETFs). But I don’t really consider ETFs as brand new investments given that most of them simply attempt to reproduce an underlying index, which has usually been around for much longer (stock market index ETFs, commodity ETFs, even real-estate ETFs). These indexes should probably be used as proxy for the ETF back-testing for longer back-test periods (this is what the bookauthor does with index data going back to 1982).

    Of course this does not work with some more exotic ETFs (which are usually much more recent), which do not track an index for example. In that case back-testing is quite limited.

  8. 54 mos, 2 wks ago

    It seems to me that today’s leveraged ETFs (2x and 3x) are just begging to be used in short trades. The daily resetting and the decay involved with those types of ETFs make it hard for them not to lose value over the longer term.

  9. 43 mos, 2 wks ago

    @ Max: You are right to say that some of these leveraged ETF’s have daily reseting and time decay. A method for combating that would be to utilize a timing methodology that gets you out after a certain number of days if the ETF did not make a new low or a new high. Say, for example, if you are using a long ETF and it did not make a new low for 12 day consecutive days, you could close that position.

  10. 43 mos, 2 wks ago

    Oops and it did not make a new HIGH for 12 days consecutively, then you could close the position on that basis.

Leave A Comment