Forums  > General  > NN Taleb: σ needs to be retired for mean deviation  
     
Page 2 of 2Goto to page: 1, [2] Prev
Display using:  

dbt


Total Posts: 4
Joined: May 2020
 
Posted: 2020-05-22 19:23
-

dbt


Total Posts: 4
Joined: May 2020
 
Posted: 2020-05-22 19:23
Have to chime in that I side with Cliff and tip my hat to him for speaking up.

Story time. One of my mentors from college helped develop a simplified pricing model, which became widely used. But like Fama, he was always well aware of the imperfections of the model assumptions and empirical properties of data.

Never mind that my mentor also broadened the appeal for jump-diffusion models to correct for fat tails, in recognition of the the original model's limitations. This was in 1976 - long before NNT somehow established himself as the flagbearer of the populist rebranding of fat tails. To this day, reading through some of these tweets, it seems NNT hasn't stopped with the misinformation campaign against him.

My view is limited because I haven't read any of NNT's books. But it seems from multiple accounts that NNT recycles the same template. Start a strawperson argument that some highly respectable individual blindly follows some naive model, then slander said person for a position that the person never held. With enough repetition, somehow the layperson becomes indoctrinated with the fictionalized version of events. Doesn't help that Asness, Gene Fama, my mentor, etc. are all respectable and helpful people who have a wide cross-section of publications that can be subject to this sort of creative misinterpretation. And they all have better time on their hands than to speak up to their detractors.

It's quite a shame too because I presume NNT was once, too, helpful in interfacing academia with practitioner wisdom.

ronin


Total Posts: 594
Joined: May 2006
 
Posted: 2020-05-24 19:41
It's hard to sympathise with either of them tbh.

It's the battle of the overweights.

In the red corner, Nassim Nicholas "Right Once Every Decade" Taleb.

In the blue corner, Clifford Cliff "We Crash When the S&P Crashes" Asness.

The final installment of this epic sage is expected to take place some time in the 2070s. The running title is "I told you so - no you didn't".

Riveting stuff.

"There is a SIX am?" -- Arthur

quantie


Total Posts: 903
Joined: Jun 2004
 
Posted: 2020-05-27 02:25
Aaron weighs in

AlmostEvil665


Total Posts: 5
Joined: Dec 2016
 
Posted: 2020-06-08 14:24
Well, one thing that I learned reading Tableb's books it was about the "sterilization" of probability.

The guidance using probability/statistical from a computer, it can be (or it is) different from the real world. It's a pretty concept to grasp.

From the experience of other books and myself, it's intrinsic for ourselves to simplify everything (find patterns, when they don't exist, ring any bells?) - maybe we did with probability as well. Well, it's a concept tho

doomanx


Total Posts: 88
Joined: Jul 2018
 
Posted: 2020-06-08 17:13
Alright guys it's not funny any more, stop training neural nets on the Incerto and letting them lose on NP like this @AlmostEvil665

did you use VWAP or triple-reinforced GAN execution?

pj


Total Posts: 3530
Joined: Jun 2004
 
Posted: 2020-06-08 19:27
Maybe it's the inverse Turing test?
People are becoming more and more stupid?

The older I grow, the more I distrust the familiar doctrine that age brings wisdom Henry L. Mencken

AlmostEvil665


Total Posts: 5
Joined: Dec 2016
 
Posted: 2020-06-08 20:37
lol. My uni teachers at time didn't receive with a warm welcome this concept as well.

Nowadays I think as a way to never trust enough the models. But take care to Sisyphus not laugh at you.


About the Turing Test: I'm not a native english speaker, so some texts can sound as written by some NLP... Well, that's what a machine would say. Big Smile

injectivenn


Total Posts: 1
Joined: Jun 2020
 
Posted: 2020-06-08 22:03
Talleb's books were a breath of fresh air in graduate school. Basically it is an exercise in the construction of improved results. Is possible to model closed-form systems that make sense of the non-antifragility of open-form contrapositive systems?

Talleb's idea of the canonical category is that we can address classical questions about locality by treating every price series as a Grassmann fractal.

She also showed us how to dynamically price every interest rate swap using long-tailed ideas of Huygens and Torricelli.

But is the nullification of robustification also the same as the continuity of negative antifragility? To this day experts like Aaron still don't know.

Also her idea about degenerate hyper-stabilized tails. Domain experts make wrong model assumptions about the general topology of time series forecasting. She call this idea the intrisification bench press. This is why LTCM failed.

nikol


Total Posts: 1172
Joined: Jun 2005
 
Posted: 2020-06-09 09:49
Started to listen audiobook of "Antifragile"

Although Nassim has coded there many narcissistic clauses and reverences of "ah, I did it all my life, believe me", I should say that the text spawns dozens of parallel ideas. Not necessarily in exact direction of where Taleb has meant it to be. The book is a good fertilizer. Time to time I struggle to agree with his arguments, hence start to generate my own. So, overall the result is positive.

That's what good book should be about - not only teaching, but encouraging to explore the subject yourself.
Previous Thread :: Next Thread 
Page 2 of 2Goto to page: 1, [2] Prev