Title: Trading Systems - A new approach to system development and portfolio optimisation
Authors: Urban Jaekle & Emilio Tomasini
Published: Reprinted 2011 (first published 2009)
Trading Systems by Jaekle and Tomasini is a well presented demonstration of how one can go about testing and improving a trading system. It's more about how to test and improve an existing concept, rather than coming up with ideas or entry/exits combiations.
The authors take a very balanced view with their discussions, which adds a lot of value. The authors often discussed possible weaknesses and alternate methods to their approach which was a refreshing change.
I wouldn't recommend this book for someone just starting out as there is a fair amount of knowledge assumed (e.g. familiarity with drawdowns, max adverse and favourable excursion). But if you've had some experience with developing your own system, then Trading Systems would be a worthwhile addition to your collection.
Cost: $42 online ($95 RRP) pretty cheap as far as trading books go
Size: 197 pages (not including appendices) concise but the topics are well covered
Format: Softcover, large page, large font seems typical for publisher Harriman House
1. A Practical Guide to Trading System Development and Evaluation
2. Trading System Development and Evaluation of a Real Case
3. Systematic Portfolio Trading
Part 1 The introductory bit.
Part 1 is primarily a lead in for Part 2. The topics include the obligatory definition of a trading system, importance of data integrity, some typical system testing metrics, and optimisation. But, although a lot of the content in this part wont be new to most, the authors didn't mess about with common terms and instead used that time to better explain some of the more subtle details of trading system evaluation.
The discussion on optimisation and robustness actually including useful ways to measure it and not just being the typical warning spiel about over-fitting.
Part 2 A well put together system development case study
Part 2 is where the real value is in the book. This part follows the development of a simple moving average cross over system, running through many of the concepts that were introduced in Part 1. The process followed through this development was fairly standard (define stops, tweak exits, test over various parameters etc), however the detailed explanation of the impact caused by the changes at each stage of the process was possibly the best Ive seen so far.
Part 2 also contains a solid explanation of application, benefits, and potential pitfalls of Monte Carlo testing along with some really good detail on walk-forward optimization.
The clear explanations of the impact each change to the system has on its performance.
The quality of the data and graphical presentations.
The plots showing stability of system performance against parameter values (e.g stop size).
The idea of testing system performance at different times (e.g. in the morning vs. afternoon).
The crystal clear hypothetical example explaining how model complexity can lead to over-optimisation.
Part 3 A (very) brief touch on combining systems into a portfolio
The main thrust of this part Part 3 was how to best combine several strategies into a single portfolio. There is some discussion about correlation and how it can be of benefit, as well as some basic discussion on topics such as trading the equity curve of a system, but none are treated with the same rigor as the concepts in Part 2.
A (very brief) description of a new concept using walk forward testing to determine which systems to allocate funds to.
For me, Part 2 was where the real value was and was worth the (fairly) low cost of the book on its own. The clear presentation of the methods and tools made it easy to understand how they could be applied to any system not just the day traded GBP currency example used.
After the quality material in Part 2, Part 3 was a bit of a disappointment. The authors do say up front that the topic is too large to cover in detail, but I always find this comment to be a bit of a cop-out.
I didnt like the format the publishers use. It puts me in mind of the formatting you would find in a Student thesis or white paper rather than a professionally printed book. This certainly isn't a deal breaker, just gives the book a less professional feel.
Overall I thought Trading Systems by Jaeckle and Tomasini was a winner. It deepened my understanding and appreciation of concepts I thought I already knew, and also added some new test methods and data representation to my system development repertoire.
It's a thumbs up from me!
P.S. - If your still sitting on the fence about this one, then surely the author Urban Jaekle should be the clincher because his name is so damn cool.