The last book I read on market microstructure theory was deadly dull. So, I've got for the other end of that particular spectrum. Subtitled 'How to build your own algorithmic trading business', it really does try to tell you how to do that - as a retail trader, how to make money using quantitative analysis - statistical arbitrage etc.
The great thing about the book is its wonderful enthusiasm. It's all 'If I can do this, so can you!'. He skims lightly over the fact that he's got a PhD in physics from a well-respected university and got experience at several major banks and hedge funds before setting up on his own. I think he gets away with it because everything he discusses is actually very straightforward. How to analyse time series with Matlab to get a viable strategy, how to set up a brokerage account, how to automate your trading, a few common strategies and pitfalls, etc. He makes particular emphasis on Sharpe ratio and Kelly criterion, without oversimplifying (too much).
The competition is continually getting harder, but apparently it really was possible to make consistent money with simple algorithms, trading from home as a retail customer, in the mid-2000s! It's a great book for saying 'No, that pretty much is it! It's not magic!'.
The downsides? It's pretty pricey for a short (160 pages), not particularly dense book that won't really tell you anything new if you have much experience (although it does present it very pleasantly). It does feel a bit like it might be a little 'the book of the blog' for epchan.blogspot.com, but even in that case it's much more cleaned-up than such books tend to be. In summary, if you don't know much about the area, or aren't so fussed about value for money, it's rather good fun.