Thursday, April 21, 2016

Trend Following: Part One

What is it?

"The core idea of Trend Following is extremely simple...Wait for momentum to build in one direction and get on the bandwagon.  Expect to lose about two thirds of the time and so make sure your winners can pay for your losers and leave enough over to cover the rent." (Clenow)

In other words, buy after prices start moving up, sell after they start moving down:

The rules used are simple, for example:
  • moving average crossovers above, 
  • channel breakouts
  • "buy when the price has spent 5 days above the moving average", etc.  
You may also have an overriding rule to decide "market direction" first, and only take the trades in that direction.  For example:
  • based on a 200 day MA.  
  • For the stock market - based on macro economic data.  
  • For a currency - based on its strength compared to all other key currencies.

Key Components of Trend Following Systems

Back Testing

Rules are back tested against old data to be chosen for the system.  You must know computer programming.  The process is harder than it sounds - there are pitfalls like over-fitting. The book Systematic Trading by Robert Carver covers these.

After testing, you start the system in real life and hope it keeps working.


60-70% of trades fail.  These should be made up for by rare winners.  You must systematically take every trade, you can't start second guessing which ones will win.  

Individual markets often have long periods, sometimes years long, where they are trendless, like soy from 2009 to 2011.  Even after bleeding for years, you need to take every trade.  After eight, nine, ten, eleven, twelve trades in a row lose, you still have to take the next one.  Hard to do, unless you enjoy pain.

Position Sizing by Volatility

Unlike investors, Trend Followers measure risk by price volatility.  Specifically, volatility from the recent past is assumed to be close to volatility in the near future.  Potential loss is measured based on this volatility and the stop loss level given by your rule.  This should be limited to between 1% and 1.7% of your capital per trade.

Potential return is also measured by volatility (again with the potential profit given by your rule). More volatile instruments require a smaller position, less volatile ones require a larger one (or maybe leverage).

You shouldn't trade instruments where past volatility is artificially constrained e.g.:
  • Pegged currencies, e.g.: trading CHF when the Swiss government removed the peg --> you get blown up.  Maybe your brokerage too.
  • Going long on bonds in an low interest rate environment: Upside is limited as interest rates can't fall much further.  Downside will explode once rates move up.


Trend followers typically trade 30-40 futures markets at the same time, spanning commodities, currencies and bonds.  The large number of markets is required because most markets are usually trendless. Adding new markets decreases your risk and increases your returns at no cost.

The problem with exchange traded futures is that one contract size is large.  Clenow suggests you need a minimum of USD 1m to start trading safely.

Yeah...if I had a million, would I spend my time implementing a Trend Following System?

May not work on stocks

Trend following may not work on stocks, as they are more volatile than futures.  Stock markets are highly correlated, and become more so when the SHTF, like in 2008.  So a trend following system in different stock markets is not diversified.  Also, vicious short squeezes will kill a simple automated system.  A systematic trend following system dealing with stocks will probably be long-only.

What Next?

The main benefit of a futures-based Trend Following system is to provide diversification, in particular, they tend to do well when stock markets do badly.

If I was to stick to equities-only, to avoid the high capital requirements for exchange-traded futures, a trend following system may be useful as a counterpoint to value investing. When the market is high - like now - and buying and holding is too risky, a growth/momentum strategy with a cut-loss could gain exposure to the market with limited risk.  So it gives me a way to trade when I'm no good at discretionary trading.

Next, I'll look into whether I could build my own system, or buy into a fund.


Books and Websites

  1. Trend Following by Michael Covel: The book that popularised it.  Well written and entertaining, but lacking information.  Read if you're bored.
  2. Following the Trend by Andreas Clenow.  Describes what its like to implement and follow a system through his experience as a fund manager.  Well written, clear and concise, it is non-technical and an easy read.  An excellent first book for building a system - also see his writings online (1) (2).  Anyone buying into a TF fund should also read this, to discover what TF funds do, how they work, and what you're in for.
  3. Systematic Trading by Robert Carver.  More technical that the previous book, it goes into more detail on why Trend Following works, and common pitfalls in building a system.   I usually skimmed each chapter several times, first to determine what he is trying to do, then how he is doing it.  Its still well written, clear and concise, with the occasional dash of humour.  I think Part One of the book ("Theory") will be useful to any investor/trader.
  4. Au.Tra.Sys blog: Still digging through all the stuff here; useful for anyone building a system or buying into a fund.