Nerds on Wall Street: Math, Machines and Wired Markets Review

Nerds on Wall Street: Math, Machines and Wired Markets
Average Reviews:

(More customer reviews)
With all due respect to the previous Amazon reviewers, it's hard to believe they both (a) read this book and (b) have any familiarity with Wall Street technology. The book is a collection of articles written for technology magazines from the mid-80s to the mid-90s. Even within an article entire paragraphs are repeated, and the same idea in more or less the same words can often be found a dozen times or more in the book. This is interspersed with apparently random cut-and-pastes from the Internet and lots of tiny black-and-white pictures which the author tells you are only meaningful with color and animation. You get the feeling the author cleaned out his desk, and decided to make some money from the stuff he didn't want anymore.
There is some useful information in here, and the author does know a lot about automated equity trading before the advances of the late 90s. The trouble is it's not presented in coherent sequence and the technical level is too uneven. For example, it is asserted five separate times that garbage collection is a problem for LISP, without any background material. Anyone who knows what garbage collection means in this context, or has worked with LISP, already knows this and will get annoyed at even the second repetition. Anyone without that background will find the repeated explanations meaningless. There is nowhere near enough technical information for nerds who want to understand Wall Street (or the Wall Street of 20 years ago) or Wall Streeters who want to understand nerds, but there is far too much unexplained jargon for non-technical readers.
Another complaint is the author makes significant errors when he steps beyond his expertise, which is often. For example, he claims if you have 1,000 statistical results significant at the 5% level, 50 of them will be false. The correct statement is if you test 1,000 rules with no predictive value, you expect 50 of them to show significance at the 5% level. The number of your significant results that are false depends whether you start with rules that are mostly useful, or mostly random. This is the key insight to the concept of data mining, the author's misunderstanding makes his chapter on the subject misleading.
Another error is the claim that futures markets were developed to allow farmers to lock in prices. This is false historically (no farmers were involved in the creation of futures markets, farmers have never been big participants and have often tried to have them shut down, when farmers do transact it is much more often to double up their bets by buying the crop they grow than it is to hedge) and anyone who believes it misunderstands the economic function of futures. That's dangerous if you also have a computer that can send trades to financial exchanges. Professionally, the author stuck to equities so it didn't matter to him, but it could matter to his readers if they rely on his account.
There is one up-to-date section at the end, which the author admits was tacked on to make the book more relevant, even though he knows nothing about the topic. His angry rant about the current financial crisis appears to be constructed from reading the first paragraphs of other people's rants. He relies almost exclusively on quotes from politicians, senior regulators and bank CEOs, who all agree it was the nerds' fault. He condemns "complex and opaque" techniques in strong language and great lengths. This from a guy who built black-box trading systems. While it's true there can be a long path between a mortgage dollar a borrower sends in (or, more to the point, doesn't send in) and the end investor, and there can be matches from phantom securities along the way, all of this is done by clear rules which are disclosed. You don't really know what a black box program will do until you turn it on, and its workings are never made public. I'm not defending synthetic CDO-squareds, I'm just pointing out opinions on complexity should come from people who know the field. A non-programmer might look at 1,000 lines of computer code and say it is hopelessly complex and opaque, when a programmer finds it a clear and elegant solution. When disaster strikes, everyone will agree it was the computer's fault.
Then he's "mad as hell" at the irresponsibility of Wall Streeters. Again, without arguing the point, this is a guy who loves the Cold War doctrine of mutually assured destruction, and worked on military projects involving weapons of mass destruction for, in his own words, "the guys in the five-sided nuthouse." The worst financial idea in history does not compare in irresponsibility to supporting the capability to destroy all life on earth, at the direction of people you believe to be insane. In my opinion, the system the author supported and still supports had something like a 10% chance of killing me and everyone else (and still might do it), with absolutely no moral or other human justification. And it was done by people, like the author, who were avenging no personal tragedy, were not hungry or trapped or desperate, who had no great spiritual rationale; just irresponsible nerds with toys.
Finally, the coverage is entirely based on projects the author happened to work on and write about at the time, so a few areas are overcovered and many other areas are ignored. With a good editor to remove the redundancies and sections the author is not qualified to discuss, to order the material and to insist on background explanations, links and transitions, this might be a pretty good account. Until that happens, I suggest you avoid this book.

Click Here to see more reviews about: Nerds on Wall Street: Math, Machines and Wired Markets



Buy NowGet 34% OFF

Click here for more information about Nerds on Wall Street: Math, Machines and Wired Markets

0 comments:

Post a Comment