This is a very good work by the authors. As a developer of trading systems with a lot of failures and some successes, I'm increasingly being aware of more and more pitfalls, still this book brought me some very interesting aproaches, which I'm not sure they are widely used.
Good points are the idea of looking for a stable area of parameters when optimizing, checking different timescales for that too.
The authors points to the development of some simple strategies that are time tested and that aren't very susceptible to deterioration on Out Of Sample tests. Also, they warn against the other side, when there is too much curve-fitting.
Walk-forward chapter could use a bit more info, like some minimum criteria for setting your timescale of WFO chunks, like number of trades or such.
On Part III, previous discussed trading systems are applied to a Portfolio level management. Authors advocate using equity line crossovers instead of a drawdown rule to drop strategies, but also shows other money/risk management systems.
Well written book for those algorithmic traders out there.
Estimated reading time: 5h30m.