Hi everyoneQuote "Option hedging-pricing, and trading is neither philosophy nor mathematics." from "Why We Have Never Used the Black-Scholes-Merton Option Pricing formula" by E.G.Haug & N.N. Taleb.After recent crises, many traders and quants are somewhat confused and have wondered where we are going from here? And so I humbly ask for your professional views to be shared here in this thread.Thank you,TM

clearly they are not completely honest as each of them have written books that use or extol the BS model. Haug wrote "The complete guide" and NNT wrote "Dynamic Hedging".

Those strike me as grandstanding words aimed at selling books. Trading is ALL of mathematics, philosophy, tactics, instinct, psychology and more. You can survive without specializing in some of them but one way or another a trader uses all of them.

This reminds me a bit of the lengthy discussion we had with Elie Ayache about The Blank Swan: The End of Probability (thread) in the Book Forum.The Blank Swan (book)Some see a lot of depth in his writing, some say it is a load of PoMo French Abstraction going nowhere.It can be quite stimulating to think that way, if you have the background in PMFA and the inclination, but its good to get back to brass tacks too.Perhaps we could compile a few books or papers that resonate as being intellectually honest or somehow Truthful now, in spite of being written by creative mortal men. (And I do not mean to imply that those who are mentioned here are dishonest at all, merely that without the right background and philosophical perspectives, you can understand why some people become skeptical or confused or jealous of people who find the time and energy to write books...!).For a view of markets, An Engine, Not a Camera: How Financial Models Shape Markets, by Donald Mackenzie is excellent.For a long view of economic and financial history: The Origins of Wealth - Eric BeinhockerKnowledge and the Wealth of Nations: A Story of Economic Discovery - David WarshFrom a risk perspective:Red-Blooded Risk: The Secret History of Wall Street - Aaron BrownModels and Their Discontents:Models.Behaving.Badly - Emanuel DermanThis is all good for taking stock of what has happened and how we got where we are today. A number of other books have been published about quant finance, by way of physicists on Wall Street or Scott Patterson's novelistic-journalistic view, but with just a few exceptions, I have not found anything that really delves deep into the practical evolution and intellectual history of QF in a fair and impartial way.A book to write, perhaps!

My trading colleagues and quant students have raised questions as to where and who to look for advice and directions. Given crisis after crisis, we thought we had "tamed" the markets with our most advanced models, but we all admitted that we are entering a colder harder reality and feel that there is a need for a change in the way we deal with the problems - albeit modeling, pricing, hedging, CVAs, etc...all in a consistent maner. In the recent years, we have seen academics and practitioners (including Taleb and Haug) who have discussed these problems as if they are disparate issues, and those (including household names, eg Scholes and Merton) who discussed nothing as if there are no problems to discuss.Who do we look to for answers, advice and guidance?

Against Haug and Taleb.Or why philosophy is required after the maths, or maybe before.

QuoteOriginally posted by: daveangelclearly they are not completely honest as each of them have written books that use or extol the BS model. Haug wrote "The complete guide" and NNT wrote "Dynamic Hedging".And not to mention lattice methods and Monte Carlo? What about BDT?

I think we need to be clear about why this sort of stuff gets written, this is not a cookbook, the idea is to mink about make people think about what they're doing, rather than automatically doing what they've always done.

Up to 1987 when market participants played in conformance with the BS game rules everything went well - the headache began with the far out of the money options. It was a disruption: being in the comfort zone market participants tend to fall into the no-problem problem (it is enough to be good at not being bad), but after? With increasing model complexity you (most probably) run into the problem of uninformative data (IMO, a reason why the DSGE macro model has so many faces - and so many interpretations what they are "good" for. Some, even suggest that they should be used in financial markets - Beware?!). So, if the complexity drives you into the computational jungle or the dangerous data salt mines (copyright T4A), what to do? I tend to agree with numbersix' "riding the price waves". It is a change based on a philosophy (speculative reality). But, avoiding the discussion what is a BSM?, Taleb and Haug are right that "BSM" has lost its innocence. Do they ever have used BSM? Is this important?p.s. computational jungle and data salt mine danger? There are already ill-posed problems in the small: Black vs. Bachelier

To me, the use of BSM in the markets is somewhat like the use of FFT or DCT (discrete cosine transform) for JPEG images. DCT is a very useful mathematical representation of the structure of an image, but no one would claim that an image "IS" a literal collection of sinusoids. By the same token, is anyone claiming the market is exactly BSM in all it's statistic properties? Instead, the language of BSM provides a convenient way to talk about volatility "as if" the volatility structure was some amalgamation of BSM Gaussians of different volatilities. BSM or DCT offers a mathematical language for describing some physical thing and as with all languages, the atoms of the language and the atoms of the thing are not one and the same.The one huge discrepancy in the "BSM<->market" is like "DCT<->image" metaphor is that the use of BSM affects the markets in ways that DCT does not affect images. Market participants and especially the market makers do far more than "ride the price waves." "Riding" is too passive a word. In actuality. anyone who rides waves knows that the rider creates a wake so that the wave rider is a wave maker, too. The shape of the wave created by the wave rider is a function of the shape of their surfboard and how they steer it.Every act of buying and selling changes the price. The financial markets have no passive observers (unless ET is listening the CNBC). And every interpretation of price that affects whether someone buys or sells (or how much they buy and sell) ends up affecting the market. The act of hedging, especially dynamic hedging, cannot help but change the market. In this case, a BSM-shaped surfboard creates a BSM-shaped wake in the markets.Under most conditions at one's favorite beach (or bourse), the waves generated by the massive ocean dominate the puny wakes generated by the assembled surfers. But that's not always the case in the financial markets. It would seem that many a crisis (e.g., flash crash, 1987, 2008, LTCM, etc.) arose when the collective waves created by the surfers swamped the exogenous ocean of the economy. In such cases the model (even if used only as a language) matters and the result may be a BSM-shaped tsunami.

QuoteOriginally posted by: Traden4AlphaTo me, the use of BSM in the markets is somewhat like the use of FFT or DCT (discrete cosine transform) for JPEG images. DCT is a very useful mathematical representation of the structure of an image, but no one would claim that an image "IS" a literal collection of sinusoids. By the same token, is anyone claiming the market is exactly BSM in all it's statistic properties? Instead, the language of BSM provides a convenient way to talk about volatility "as if" the volatility structure was some amalgamation of BSM Gaussians of different volatilities. BSM or DCT offers a mathematical language for describing some physical thing and as with all languages, the atoms of the language and the atoms of the thing are not one and the same.The one huge discrepancy in the "BSM<->market" is like "DCT<->image" metaphor is that the use of BSM affects the markets in ways that DCT does not affect images. Market participants and especially the market makers do far more than "ride the price waves." "Riding" is too passive a word. In actuality. anyone who rides waves knows that the rider creates a wake so that the wave rider is a wave maker, too. The shape of the wave created by the wave rider is a function of the shape of their surfboard and how they steer it.Every act of buying and selling changes the price. The financial markets have no passive observers (unless ET is listening the CNBC). And every interpretation of price that affects whether someone buys or sells (or how much they buy and sell) ends up affecting the market. The act of hedging, especially dynamic hedging, cannot help but change the market. In this case, a BSM-shaped surfboard creates a BSM-shaped wake in the markets.Under most conditions at one's favorite beach (or bourse), the waves generated by the massive ocean dominate the puny wakes generated by the assembled surfers. But that's not always the case in the financial markets. It would seem that many a crisis (e.g., flash crash, 1987, 2008, LTCM, etc.) arose when the collective waves created by the surfers swamped the exogenous ocean of the economy. In such cases the model (even if used only as a language) matters and the result may be a BSM-shaped tsunami.Is riding the price waves passive? I am not sure.1. Yes, An image is not a movie. 2. Motion control (detect artefacts, anomalies, ...) of a movie of a real scene is an ambitious task for a computer program. Even if you have detailed context information. In the core you use models that take an image at a time and predict what may happen a lttle later and adapt your models trying to close the gap between reality and prediction..3. If you restrict movies to scenes about deformations of cubes under certain forces, it seems easy. One contextual (continuums mechanics) model can make the motion control? But wait, the cubes are of a material thats exact physical properties are unknown. You need to mine them from the deformations you observe (parameter identification). Your model instance improves during the movie.4. Yes, published prices change themselves. This influences the price waves as well as the "weather" and other conditions - but quantitatively unknown. Only the "wave" contains all information. To make a little next step, you need a model how. 5. Can you keep all parameters stochastic? IMO, no (computational complexity and uninformative data). And this is where the philosophy of speculative reality may help: If I do not know more properties of a real behavior than my model describes, then my model is reality. Tomorrow, I may know more.

QuoteOriginally posted by: exneratunriskQuoteOriginally posted by: Traden4AlphaTo me, the use of BSM in the markets is somewhat like the use of FFT or DCT (discrete cosine transform) for JPEG images. DCT is a very useful mathematical representation of the structure of an image, but no one would claim that an image "IS" a literal collection of sinusoids. By the same token, is anyone claiming the market is exactly BSM in all it's statistic properties? Instead, the language of BSM provides a convenient way to talk about volatility "as if" the volatility structure was some amalgamation of BSM Gaussians of different volatilities. BSM or DCT offers a mathematical language for describing some physical thing and as with all languages, the atoms of the language and the atoms of the thing are not one and the same.The one huge discrepancy in the "BSM<->market" is like "DCT<->image" metaphor is that the use of BSM affects the markets in ways that DCT does not affect images. Market participants and especially the market makers do far more than "ride the price waves." "Riding" is too passive a word. In actuality. anyone who rides waves knows that the rider creates a wake so that the wave rider is a wave maker, too. The shape of the wave created by the wave rider is a function of the shape of their surfboard and how they steer it.Every act of buying and selling changes the price. The financial markets have no passive observers (unless ET is listening the CNBC). And every interpretation of price that affects whether someone buys or sells (or how much they buy and sell) ends up affecting the market. The act of hedging, especially dynamic hedging, cannot help but change the market. In this case, a BSM-shaped surfboard creates a BSM-shaped wake in the markets.Under most conditions at one's favorite beach (or bourse), the waves generated by the massive ocean dominate the puny wakes generated by the assembled surfers. But that's not always the case in the financial markets. It would seem that many a crisis (e.g., flash crash, 1987, 2008, LTCM, etc.) arose when the collective waves created by the surfers swamped the exogenous ocean of the economy. In such cases the model (even if used only as a language) matters and the result may be a BSM-shaped tsunami.Is riding the price waves passive? I am not sure.1. Yes, An image is not a movie. 2. Motion control (detect artefacts, anomalies, ...) of a movie of a real scene is an ambitious task for a computer program. Even if you have detailed context information. In the core you use models that take an image at a time and predict what may happen a lttle later and adapt your models trying to close the gap between reality and prediction..3. If you restrict movies to scenes about deformations of cubes under certain forces, it seems easy. One contextual (continuums mechanics) model can make the motion control? But wait, the cubes are of a material thats exact physical properties are unknown. You need to mine them from the deformations you observe (parameter identification). Your model instance improves during the movie.4. Yes, published prices change themselves. This influences the price waves as well as the "weather" and other conditions - but quantitatively unknown. Only the "wave" contains all information. To make a little next step, you need a model how. 5. Can you keep all parameters stochastic? IMO, no (computational complexity and uninformative data). And this is where the philosophy of speculative reality may help: If I do not know more properties of a real behavior than my model describes, then my model is reality. Tomorrow, I may know more.BSM is a hammer and all the discussions of the volatility smile are discussions of how nail-shaped the markets are.What I think is so curious is that the markets are social constructs both in their rules and in their operations. To the extent that enough people believe model X is true and buy when model X says Y is underpriced then Y will rise in price which is consistent with model X! That is, to a first approximation, belief in model X will induce behaviors in buyers and sellers that make the markets consistent with model X! Someone might propose a "better" model but they will likely lose money as long as most participants believe in model X.I'd say that all parameters are stochastic (or integrals of stochastic processes) because we live in a dynamic world. But some are more slowly time-varying than others or are seemingly invariant within a restricted sampling context (e.g., the mass ratio of protons to electrons holds in this universe but maybe not in others or pi is the ratio of circumference to diameter in all non-extreme gravitational fields). That said, we might assess the quality of a model by how not-stochastic its parameters are. A poor model will have highly stochastic parameters in which "real life" diverges from the model over a short time horizon. A good model will have less stochastic parameters and provide a longer horizon of congruence between the model and the evolution of the system.

QuoteOriginally posted by: Traden4AlphaQuoteOriginally posted by: exneratunriskQuoteOriginally posted by: Traden4AlphaTo me, the use of BSM in the markets is somewhat like the use of FFT or DCT (discrete cosine transform) for JPEG images. DCT is a very useful mathematical representation of the structure of an image, but no one would claim that an image "IS" a literal collection of sinusoids. By the same token, is anyone claiming the market is exactly BSM in all it's statistic properties? Instead, the language of BSM provides a convenient way to talk about volatility "as if" the volatility structure was some amalgamation of BSM Gaussians of different volatilities. BSM or DCT offers a mathematical language for describing some physical thing and as with all languages, the atoms of the language and the atoms of the thing are not one and the same.The one huge discrepancy in the "BSM<->market" is like "DCT<->image" metaphor is that the use of BSM affects the markets in ways that DCT does not affect images. Market participants and especially the market makers do far more than "ride the price waves." "Riding" is too passive a word. In actuality. anyone who rides waves knows that the rider creates a wake so that the wave rider is a wave maker, too. The shape of the wave created by the wave rider is a function of the shape of their surfboard and how they steer it.Every act of buying and selling changes the price. The financial markets have no passive observers (unless ET is listening the CNBC). And every interpretation of price that affects whether someone buys or sells (or how much they buy and sell) ends up affecting the market. The act of hedging, especially dynamic hedging, cannot help but change the market. In this case, a BSM-shaped surfboard creates a BSM-shaped wake in the markets.Under most conditions at one's favorite beach (or bourse), the waves generated by the massive ocean dominate the puny wakes generated by the assembled surfers. But that's not always the case in the financial markets. It would seem that many a crisis (e.g., flash crash, 1987, 2008, LTCM, etc.) arose when the collective waves created by the surfers swamped the exogenous ocean of the economy. In such cases the model (even if used only as a language) matters and the result may be a BSM-shaped tsunami.Is riding the price waves passive? I am not sure.1. Yes, An image is not a movie. 2. Motion control (detect artefacts, anomalies, ...) of a movie of a real scene is an ambitious task for a computer program. Even if you have detailed context information. In the core you use models that take an image at a time and predict what may happen a lttle later and adapt your models trying to close the gap between reality and prediction..3. If you restrict movies to scenes about deformations of cubes under certain forces, it seems easy. One contextual (continuums mechanics) model can make the motion control? But wait, the cubes are of a material thats exact physical properties are unknown. You need to mine them from the deformations you observe (parameter identification). Your model instance improves during the movie.4. Yes, published prices change themselves. This influences the price waves as well as the "weather" and other conditions - but quantitatively unknown. Only the "wave" contains all information. To make a little next step, you need a model how. 5. Can you keep all parameters stochastic? IMO, no (computational complexity and uninformative data). And this is where the philosophy of speculative reality may help: If I do not know more properties of a real behavior than my model describes, then my model is reality. Tomorrow, I may know more.BSM is a hammer and all the discussions of the volatility smile are discussions of how nail-shaped the markets are.What I think is so curious is that the markets are social constructs both in their rules and in their operations. To the extent that enough people believe model X is true and buy when model X says Y is underpriced then Y will rise in price which is consistent with model X! That is, to a first approximation, belief in model X will induce behaviors in buyers and sellers that make the markets consistent with model X! Someone might propose a "better" model but they will likely lose money as long as most participants believe in model X.I'd say that all parameters are stochastic (or integrals of stochastic processes) because we live in a dynamic world. But some are more slowly time-varying than others or are seemingly invariant within a restricted sampling context (e.g., the mass ratio of protons to electrons holds in this universe but maybe not in others or pi is the ratio of circumference to diameter in all non-extreme gravitational fields). That said, we might assess the quality of a model by how not-stochastic its parameters are. A poor model will have highly stochastic parameters in which "real life" diverges from the model over a short time horizon. A good model will have less stochastic parameters and provide a longer horizon of congruence between the model and the evolution of the system.Sorry, I have argued too much about computational complexity and traps. I should return to the argument of game rules. BSM can make a (small) part of a market a "fair" game. It lost its innocence because the game was played against the rules?

Thank you all contributors to this thread. And please keep coming with your insightful thoughts and comments, especially related articles. I've just finished reading the article referred by numbersix which was deep and insightful! Thanks.It certainly appears to me that we are all on a journey looking for absolute or relative truth. It reminds me of a quote: "If you immediately know the candle light is fire, then the meal was cooked a long time ago." I feel as if i've already possessed the truth all along, but fails to recognise, and put it into words and formulae.

I sometimes think we need to replace the word "model" with "toy"