Another interesting decision, now for ‘Beyond Nelson-Siegel and Splines: a model-agent machine learning framework for discount curve calibration, interpolation and extrapolation’ | R-Bloggers

Another Interesting was made this time for my paper entitled: “Beyond Nelson-Siegel and Splines: a model-agent machine learning framework for discount curve calibration, interpolation and extrapolation”. By means of European Actuarial Journal. The first time, 7 years ago, it was for Swap Curve Construction for insurance prices, based on no arbitration short interest models – that works perfectly. I could decide to publish the review one day.

I think, as it was suggested to me (after I, institutional and not alone, have been brought down several times), that even the subtle intimidation is important, especially when my reality (and the reality in general) is interrogated to make automatic prophecies.

What is amazing (so far only for me) is that people don’t hesitate Write down absurditiesEven when they are confronted with compelling vouchers that will continue to be written almost forever. Is this the era of post-truth I heard about? Or is there another hidden reality that I am not aware of (who does not even justify the gigantic institutional, and not alone, gaslighting)?

Paper Introduces a general machine learning framework for yield curve modeling, in which classic parametric models such as Nelson-Siegel and Svensson serve as special cases within a wider class of functional regression pensions. By linearizing the bond prices/SWAP valuation comparison, I reformulate the estimate of spott rates as an accompanying regressing problem, in which the response variable is derived from observed bond prices and cash flows, and the regressors are constructed as flexible functions of time-to-be-ventures. I show that this formulation supports a wide range of modeling strategies – including polynomial extensions, laguerre polynomes, kernel methods and regularized line arm models – all within a uniform framework that could retain economic interpretability. This not only makes crooked calibration possible, but also static interpolation and extrapolation. By abstracting a fixed parametric structure, my framework bridges the gap between traditional yield curve modeling and modernly subjected learning, offering a robust, expandable and data-driven tool for financial applications, ranging from asset prices to regulatory (?) Reporting.

The reviewer says:

“

1. The approach is based on a linearization of the exponential function, which can lead to considerable mistakes

2. The Regression uses errors correlated with the regressorswhich leads to model specification

That is why I cannot recommend this article for publication. ‘

Ok, so:

1. which can lead to considerable mistakes: In theory yes. But in Pratice? Have you read the newspaper? Have you seen the residues? Have you seen the images? Have you seen the comparison with other published articles? The residues are less than 1st-18.
2. What if I use independent Gaussian or what specification of errors instead? Or completely remove the errors? The reviewer does not even suggest that. He only says “what leads to model specification”.

Based on the article (at least on Swaps data, for binding data, can be discussed: what if I use independent Gaussian or any specification of errors instead? chatty Published articles, can anyone see what they are talking about? That can’t be improved?

At least recognize the novelty and suggest ways to improve, instead of opposing hatred and shadow, random rules. Have you seen that before? Bootstrapping the yield curve + interpolation + extrapolation (almost the same as I did in Swap Curve Construction for insurance prices, based on no arbitration short interest models) All at the same time? With a model-agent approach, which means a lot of flexibility based on different model capacities? Could not be this article at least improved?. This is a generalization of many existing approaches. Although it is grotesque to name names, which I will not do, I can see many papers in the same diary that can easily be Master’s dissertations. A lot of “[Place any existing- sophisticated-model-created-by-someone else here] Applied to insurance “indeed.

https://www.researchgate.net/publication/392507059_beyond_nelson-siegel_and_plines_a_model-_agnostic_machine_learning_framework_discount_curve_calibration_interpolation_and_Etrapolation

Uit p.22 in: “@article {Andersen2007Discount, title = {kortingscurve constructie met spanningssplines}, auteur = {Andersen, Leif}, Journal = {Review of Derivatives Research}, Volume = {10}, Number = {3}, Pages = {227–267}, Year = {2007}, Publisher = {Springer} “

By My paper:

From P. 242 in “@Misc {Andersen Interest, Title = {interest rate modeling, 2010}, author = {Andersen, L and Piterbarg, v}, Publisher = {Atlantic Financial Press: London}}”

By My paper:

Can you see the model wrong specification?

Related