Quote

---

Originally posted by: Stefanone
I just want stress how amazing is London's book. Never seen anything like this before.

---

Quote

---

Originally posted by: giorgio
the book is a joke; nevertheless not one of the good ones
it's one thing the writer (some poor graduate student that got
hired for that??) has no clue on the math (Dirac delta funcs are
carbon copy of Wilmott's "student intro" to start with) but his/her
C++ are at best of failing grade at any community college in
the country...
i will defintely recommend waiting 1-2 weeks; the bool will be
further discounted ($96 to $60 in a month) to $15 or so...

cheers,
giorgio

---

Quote

---

Originally posted by: spursfan
i'm glad it isn't only me that is disappointed (though i did pay $60)

my guess is that the book is a combination of his old msc lecture notes (all the tedious derivations and borrowed material) plus hopefully his own poor quality c++ code

also time (say 1998) seems to have stood still for nearly all the equity stuff whereas there is some more modern bond stuff (G2++, LMM) but even for the latter some of the implementations such as the HW or BK trees are pathetically inefficient - and the added software doesn't seem to cope with more than 100 steps in a binomial tree

---

Quote

---

Originally posted by: ametrano
> his coding style appeared to be based on that of the quantlib guys.
I would hope so. Unfortunately this is not the case.

> What I am getting from the reviews is that the code is not
> very good. I know that Quantlib has a version of G2++, HW and BK.
> Did he write his own versions?
This is the typical plot: the author sketches his own C++ functions, then for a "more robust implementation" he shows some QuantLib code and classes. He goes so far as to print more than a dozen consecutive pages of QuantLib codes with just few lines of comment. No introduction, no design, nothing to help the reader understand the rationale behind the coding.

Besides that I must add that the QuantLib version he uses is quite old, he manages to show code which was deprecated (then removed)., and even untested buggy code as the QuantLib G2 code which is labeled as untested and buggy, and I see no reason to publish it without a notice.

Last but not least the author try to cover too much material, and much of it is clearly out of his knowledge and comprehension. I haven't read the whole book, but its coverage of Brownian Bridge and Sobol low discrepancy sequences is a good example of what I mean. The author copies some code from Jäckel's book, completely misses Jäckel point about Sobol initialization numbers and his conclusions are a naif confirmation that he actually hasn't understood the peculiarities of low discrepancy sequences and how they have to be used.

The last semester has seen at least 4 books about C++ and quantitative finance (Joshi, Duffy, Levy, and London): London is the worst of the batch. QuantLib has its own limits and problems, but please do not associate it to London's book.

ciao -- Nando

---

Quote

---

Originally posted by: Cuchulainn
Yomi,
I am a bit confused. Your review is somewhat different from an early post.

This is perhaps one of the best books written in the field of quantitative finance, especially for implementing models especially using C++. It is equivalent to Numerical Recipes in C but for derivative models. It is one of the most comprehensive and details books I've seen at roughly 900 pages. It shows not only the mathematical derivations for all the major fixed income and equity models, but provides all the actual code (there is enough code to run an entire quant engine especially on the CD) and implementations for many complex models like HJM and BGM. It is the most comprehensive and detailed book I've seen in prereleased versions and will be invaluable and a must reference for any quant.

---