Scan barcode
Reviews
The Perfect Bet: How Science and Maths are Taking the Luck Out of Gambling by Adam Kucharski
beeeeg's review against another edition
4.0
An interesting look at how technology is attacking gambling, using statistical models and some AI concepts.
kevinm56's review against another edition
4.0
I really wish I had paid more attention in math class.
Good read.
Good read.
generalheff's review against another edition
2.0
Perhaps I’ve simply read too many of these kinds of books recently. You tend to realise this when you start to recognise what must be obscure stories. Like how some students at MIT beat the Massachusetts’ state lottery, or how Kurt Gödel wanted to point out in his US citizenship interview how logical flaws in the constitution could open the way to a dictatorship.
I suppose I shouldn’t hold Adam Kucharski responsible for this, but I can’t help but feel that my awareness of all this repetition of my recent pop-maths reading betokens something more problematic.
The Perfect Bet – Taking the Luck out of Gambling seriously needs a through line, a thread to hold it all together. As it stands, we get a series of interesting, adequately written chapters coverings various happenings in the worlds of gambling and prediction, but with no clear reason why we are reading all these bits and pieces together.
Some chapters hold to the ‘taking on the house’ line on the book’s back cover, with stories like the MIT students who realised that a certain type of lottery resulting in (on average) a net pay-out (a very unusual position for any lottery to be in). Or clever betting strategies in blackjack that can make you (almost) break-even; throw a little card counting in and you can move nicely into the black (no pun intended). Then there are the chapters that discuss predictive models – covering topics like sports betting. The back of the book veers jarringly again, this time into discussion of automated poker machines and how to make computers unbeatable at various games.
Even this makes it sound more coherent than it is, though. The discussion of card games and probabilities brings in some history of probability discussion, making this feel very much like a popular maths history book. We get ‘statistics for dummies’ with descriptions of regression and overfitting in the chapters on prediction. From here, we stray into the theory of games and onto artificial intelligence and Turing’s test. It’s all incredibly bitty and utterly lacks any flow.
While much of the book is fairly well written and the science seems sound enough, there are some very poor sections. The layman description of regression was so bad that I (someone with a pretty good grounding in statistics) genuinely didn’t realise that the author was describing this very familiar technique. Admittedly, non-technical descriptions are not easy so perhaps I’m being over-harsh. Where I’m not being over-harsh is in calling out the awful use of the ‘scientific’ principle of Occam’s razor: scientists avoid putting too many parameters in models because they overfit, not for any questionable philosophical (note: not scientific) principle of least complexity.
This book went a bit awry. Kucharski did not, I believe, have a clear vision of what he wanted the book to cover, which is a pity. There are truly interesting sections of this book, but it almost wants to be a series of newspaper articles or blog posts – detailing this or that fascinating mathematical discovery or computational marvel – rather than a single work to be read straight through.
I suppose I shouldn’t hold Adam Kucharski responsible for this, but I can’t help but feel that my awareness of all this repetition of my recent pop-maths reading betokens something more problematic.
The Perfect Bet – Taking the Luck out of Gambling seriously needs a through line, a thread to hold it all together. As it stands, we get a series of interesting, adequately written chapters coverings various happenings in the worlds of gambling and prediction, but with no clear reason why we are reading all these bits and pieces together.
Some chapters hold to the ‘taking on the house’ line on the book’s back cover, with stories like the MIT students who realised that a certain type of lottery resulting in (on average) a net pay-out (a very unusual position for any lottery to be in). Or clever betting strategies in blackjack that can make you (almost) break-even; throw a little card counting in and you can move nicely into the black (no pun intended). Then there are the chapters that discuss predictive models – covering topics like sports betting. The back of the book veers jarringly again, this time into discussion of automated poker machines and how to make computers unbeatable at various games.
Even this makes it sound more coherent than it is, though. The discussion of card games and probabilities brings in some history of probability discussion, making this feel very much like a popular maths history book. We get ‘statistics for dummies’ with descriptions of regression and overfitting in the chapters on prediction. From here, we stray into the theory of games and onto artificial intelligence and Turing’s test. It’s all incredibly bitty and utterly lacks any flow.
While much of the book is fairly well written and the science seems sound enough, there are some very poor sections. The layman description of regression was so bad that I (someone with a pretty good grounding in statistics) genuinely didn’t realise that the author was describing this very familiar technique. Admittedly, non-technical descriptions are not easy so perhaps I’m being over-harsh. Where I’m not being over-harsh is in calling out the awful use of the ‘scientific’ principle of Occam’s razor: scientists avoid putting too many parameters in models because they overfit, not for any questionable philosophical (note: not scientific) principle of least complexity.
This book went a bit awry. Kucharski did not, I believe, have a clear vision of what he wanted the book to cover, which is a pity. There are truly interesting sections of this book, but it almost wants to be a series of newspaper articles or blog posts – detailing this or that fascinating mathematical discovery or computational marvel – rather than a single work to be read straight through.
chinney's review against another edition
challenging
informative
slow-paced
4.5
Great insights on betting. "Average speed of horse is most important factor in determining winning horses". But way better at beginning than end. The AI and machine stuff he doesn't get, and it's clear.
lukewhitestone's review against another edition
4.0
This book reminded me a lot of The Signal and The Noise by Nate Silver. Both books are a whirlwind tour of topics steeped in uncertainty, prediction, and luck. However, while TSATN is more about modeling and creating predictions, The Perfect Bet focuses on games (of chance) and how it is possible to change the parameters of the games themselves. Many games and topics are discussed, the most interesting of which are: Roulette and the Eudaemons, Chaos Theory, Stability of Ecosystems (incl. stock markets), Arbitrage/Sports Betting, Modeling, Poker, Blackjack, and oddly enough, Gödel's Incompleteness Theorem.
When at it's best, the book sheds interesting light on how theories can come in and completely change the way we look at certain "random" processes. For example, the discussion of of how larger and more complex ecosystems reduce stability is an interesting one.
My principal frustration with this book is that, while it is heavy on examples, it is light on takeaways. Often the authors skips through applications of a theory so fast there is no real lesson to be learned. I think the book would have benefitted from a longer form discussion of the theories (and perhaps another 100 pages).
Overall I enjoyed it and would recommend it. 3.5 stars.
When at it's best, the book sheds interesting light on how theories can come in and completely change the way we look at certain "random" processes. For example, the discussion of of how larger and more complex ecosystems reduce stability is an interesting one.
My principal frustration with this book is that, while it is heavy on examples, it is light on takeaways. Often the authors skips through applications of a theory so fast there is no real lesson to be learned. I think the book would have benefitted from a longer form discussion of the theories (and perhaps another 100 pages).
Overall I enjoyed it and would recommend it. 3.5 stars.
More...