After having read many things about it, I read, The Blank Swan - The End Of Probability, Elie Ayache, author.
Yes, the thought behind the book is deeply challenging (Dan Tudball, Editor of Wilmott magazine). Ayache himself: I tend to only write the things that will be as difficult to read as they were to write. IMO, there is also an "economic aspect" of writing and reading and one might also think of the ink-factor one brings to reading. But this does not say anything about the quality which cannot be measured by operational semantics (like computational knowledge), only about my Blank Swans (as drawn in the picture here)
This is a book that I place to the most-influancial I read, re-read and re-re-read and change them, because I bring new context into my reading . Like, Dawkins, River Out Of Eden; Levy, Artificial Life; Axelrod, The Revolution Of Cooperation , Wolfram, NKS; Brown, The Poker Face of Wall Street .....
(I know that books do not have similar scientifc-weights, but I find they have in common counter-intuitive thoughts).
We, at UnRisk often have discused model-risk, relating to the traps derived from model-complexity-that-is-insane-to-calibration.
In his book Ayache (when calling "derivatives", "contingent claims") give deep insight into the "materialization" of a contingency when it gets a price on a market place. A contingent claim is not a contingent claim, if it is not dynamically replicable. If prices are functions of contexts (that we try to replicate by models), models need to be re-calibratetd to prices for explanation (not the other way around for pricing predictions). Probability is not a theory for predictive modeling but explanation and its incarnation in a model (like BS) is a guide (or game rule)?
When it comes to options, I let speak Elie Ayache (EA) (from the interview with Dan Tudball (DT): DT: .. dynamic hedging and the use of the valuation tool is merely a means to inert yourself into the market, recalibrate against the market and then repeat the procedure ..? EA: ... It is only because the dynamic trader has to be hedging continuously his position in the option and therefore is following the option continuously; it is only because of this that he is basically entitled to compute something we call implied volatility - and implied volatility is the simplest instance of recalibration ... recalibrating the BS model to the option market price ....
Traders do not price because of models, (the market does price) but use models to get insight into the context that led to the price (like risk spectra)?
For continuous calibration and recalibration you need to solve inverse problems. Continuously. Thisis not without danger: Calibration problems - An inverse problems view.
If the market prices, all market risks are model risk then? And will a market with future be able to avoid them?