Top critical review
Not for everyone/Unpractical
25 October 2017
I am a quant trader/strategist who has been working in high frequency market making strategies for the past ten years. Despite its attractive title and presentation this book is not made for everybody. You will be interested in this book if you are a quant working in a Market Making firm, a hedge fund or an asset management firm (for these last two, only if you are an execution quant implementing algos to minimize market impact for large trades).
All the book is based on a single mathematical framework which translates an objective function (representing an expected wealth constrained by variance, inventory, opportunity cost ...) to a sophisticated differential equations, DPEs. This boils down to continuous reinforcement learning whose environment is defined by stochastic processes. Despite the effort made by the authors to make us believe that these techniques are used everywhere, they are in reality not. The dynamic programming equations (DPE)/Stochastic control framework is quiet new and as far as I know only very few market making firms are using it. This framework is only useful if your company is the main market maker on a market (and backtesting becomes less of an option).
The concepts describe are not practical, the book provides the DPEs and analytical solutions for some specific cases but do not focus on the most important things, how do you solve DPEs and how do you implement them and scale them. In reality you never deal with three stocks etc... Do not get me wrong the framework is interested but the problem is how to truly implement it. There is no answer to that.
I read some comments that seems very out of lines compare to what the book is really about, I will be cautious about them, they are probably fake. I put one star for the attempt the authors to make us believe the book is about "Algorithmic and high frequency trading" in general, this is not the case. The book should be renamed "Stochastic control applied to finance".