I just finished reading the book Calculated Bets by Steven Skiena. When I saw the book mentioned on Hacker News I immediately downloaded it to my Kindle and started reading. I built a pokerbot before, so I figured I’d enjoy the book.
I did enjoy the book, but I’m not sure that I’m the right audience. I’m also not sure that this should be a book. The writing style makes me feel it should be a collection of blog posts.
Even though the book is short, it contains a substantial number of detours that should have been edited out. For example, he spent several pages talking about 1) why programmers dislike Microsoft and, 2) what the Y2K crisis was all about.
Furthermore, I found the author’s professorial tone at times endearing and at times annoying. I appreciated his descriptions of writing a program to predict football games when he was a teenager, and then convincing his local paper to publish his quotes. I also enjoyed his recounting of his graduate student years at the University of Illinois. The book also concludes nicely, with Skiena listing several fun projects that the reader might want to dig into.
On the other hand, for most of the book, the author depicts himself as an outdated Computer Science professor who doesn’t know how to code. For example, he had to recruit grad students to write programs to fetch and parse simple web pages from the jai alai websites. The description of that project phase included an awkward description of the Perl programming language. I found it awkward because he didn’t convey his own love for coding Perl. Instead, he seemed awe struck that his student could hammer out Perl code that parsed a web page.
Also, the book was ultimately too amateurish on the gambling topics. Most glaringly, he never touched on risk of ruin. He deposited $250 with the off track bettor and started making daily bets, as dictated by the program, of around $100 per day. At one point he nearly went bust, but luckily bounced up. I would have liked to see him talk about expected variance, including some notion of what his limits were for deciding that a losing streak was more than a bad turn of luck.
Finally, in terms of cold cash, the ultimate results aren’t impressive. After all of the time spent, both by him and his cheap (free) student labor, I think he made a couple thousand bucks in a year, on an investment of $250. Then he got spooked about legal issues and also wasn’t able to maintain the code after his students left. So he shut the whole thing down. In my opinion, if you’re going to go work so hard to create a winning program, you should have the drive to squeeze some serious money out of it.
Having said all that, I enjoyed the book. I’m happy that Skiena wrote it, and I appreciate his passion for systematically gambling on jai alai. Hopefully he’ll write up a summary (without detours) of his efforts on a blog and also provide all of his data and code. Then maybe some other hackers will dive in and push his system further down the track.
