Ed Seykota’s Donchian Breakout Trading System

I’ve been researching and following Ed Seykota’s work for a while now (several years off and on actually). I first came across his name while reading Jack Schwager’s “Market Wizards”. From his interview you can tell that he is a unique thinker and a smart guy – but that is a common trait of all of the wizards that were interviewed. One of the things that stands out about Ed which raised my curiosity was that in their interviews, several other market wizards had referenced him as a source of insight and one that is worth to learn from. That in itself is impressive when you consider the caliber of those interviewed.

On his web site, the trading tribe, there is information that is interesting for traders and trading system developers alike.

One of the very first automated trading systems I developed is taken from Ed’s site. Ed publised the system as an exercise in developing a trading system and refers to it as “A Simple Trading System – Support and Resistance”.

ed_seykota_simple_SR

I initially developed it as out of curiosity wondering if such a simple mechanical trading system actually makes money over a long period of time. Its provided here for educational purposes mostly. The intent is to see the type of thinking that goes into developing automated trading systems. Notice the usage of layers and the usage of states. This practice is quite common in trading in general although you might not be aware of it on a conscious level.

The previous link has a full description of the system but I’ll summarize the main characteristics and my own notes below.

  • The system is a long term trend following system. The exercise provided by Ed uses gold futures contracts on a daily chart.
  • The system is layered – the first layer provides what I call the context. In this case its the overall trend direction. The second layer will help us time entries and exits.
  • The only technical indicators utilized are Donchian channels. A long term channel is used for determining the trend direction – I call this the slow channel. A shorter term channel is used for triggering entries and exits – I call this the fast channel.
  • Channel lengths are configureable – part of the exercise is to figure out the appropriate length for the market one is trading.
  • The system can generally be in one of three states: long, short or flat. It initially starts out flat.
  • If price breaks the upper long term channel the system goes into a long state. When in a long state, the system only takes trades on the long side.
  • If price breaks the lower long term channel, the system goes into a short state. When in a short state, the system only takes trades on the short side.
  • The fast channel is used for timing trades. The channel is composed of 2 lines, an upper lineĀ  and a lower line. When in a long state, when price breaks the fast channel to the upside, we enter long. We use the fast channles lower line as a trailing stop. The opposite for short trades.
  • Like a lot of long term trend following systems, there is no profit taking. You hold a position until the market takes you out by hitting a trailing stop.

Currently, we have an implementation available for Sierra Chart. Ninja Trader is on the road map for future release.

Example of how the system changes state from long to short

see below how the system changes state from short, to long and then to short again. When it breaks the slow channel (magenta), depending if its the upper channel or the lower channel, it changes state to either long or short.

ed_seykota_simple_SR3

Example of 3 Trades

See a sequence of 3 trades below. Once in a long state, we look to the fast channel for entries. See how we enter long at point 1 when price breaks the upper fast channel. At that point, the lower blue line is our stop. It trails and we finally get stopped out at point 2. We are then out of the market till we reach point 3 where we re-enter only to the short side. The stop trails and we exit at point 4 where we get stopped out.

ed_seykota_simple_SR2