41:14 most dramatic part of the talk "the regime has changed, cointegration has been lost ! the gamma, the gamma has changed! " this quote gets the oscar
This is absolutely amazing, currently an undergraduate studying mathematics with limited statistical knowledge however your concepts especially discussing co-integration and using Kalman filter for pairs trading was incredibly insightful. Took plenty of notes and will definitely be returning to this video. Huge thanks Professor!
Your energy, research and presentation!🔥 Thank you so very much Sir! I'm an EC engineer in the last year interested in finance. You just gave me the perfect thing for my final year project!
porfavor nunca deje de subir videos! desearia poder seguir estudiando en universidad para tomar estas oportunidades, lo bueno es que pude encontrarlo en youtube y aprender de sus papers, libros y videos, gracias por compartir lo que sabe por aqui! espero poder seguir viendo mas de usted
Thanks for the material. In the slide "LS Regression for Pairs Trading" (1) what are the dotted horizontal lines? I'm assuming these lines are used as thresholds for buy/sell signals. (2) When you say 'gamma' is this the same thing as the coefficient between each pair after fitting a linear regression?
hey i have observed there as is cointigration and cointegration if u use lenghth instead of a fixed point as u used as from zero due to this slowly slowly it is not cointegreating . i used dynamic length so we will always use latest data instead of fixed point .
Fantastic talk, really enjoyed it. In today's world is it plausible that a lay person can successfully compete in pairs trading? I've a background as an RF engineer and now working in data science. Thanks
6:35. SP500 log returns are locally stationary but that the argument for assigning it a non-stationary designation via the existence of volatility clustering of any type is not valid. Volatility clustering doesn't not necessarily imply non-stationary. Slowly varying unconditional volatility implies non-stationarity but GARCHian class volatility for most parameters does not.
Crazy how people spend upwards of $100k or more on undergrad & masters tution to learn about all these information and it just sits here free. What a world we live in.
But to get a job offer or post doc positions, they will still require formal qualifications. People are essentially paying for the formal qualification (and networking of course)
Some material appears here: Yiyong Feng and Daniel P. Palomar, A Signal Processing Perspective on Financial Engineering, Foundations and Trends® in Signal Processing, Now Publishers, 2016. [palomar.home.ece.ust.hk/papers/2016/Feng&Palomar-FnT2016.pdf
@@sultanalshirah The Perspective book is much shorter indeed. Of course you can also read selected chapters in the new longer book (e.g., chapters 1,2,3,6,7,8,...). Up to you.
Is there anything you can point to in terms of implementation details for the Kalman strategy? Mainly, 1) do you track position and volatility or just position, and 2) is there a process for setting the initial parameters for the model? And apart from the Kalman element, is there a recommended procedure for determining the pairs? Very interesting video.
Great presentation sir. I have a question. The efficiency of kalman filter against rolling regression doesnt depend on the signal/noise ratio? I ve seen terrible results of KF in this context...
It all depends on the choice of the parameters. If properly chosen, Kalman is always superior than a rolling regression. For more information, please check Chapter 15 in my new book: portfoliooptimizationbook.com
He repeatedly misspelled 'Markov' in his slides. What a weird mistake to make for someone who works in his field. (that aside the talk was very interesting and he presented it masterfully ...)
Incredible talk! Timestamps:
1:46 Start of talk
4:17 Signal processing perspective on financial data
21:00 Robust estimators (heavy tails / small samples)
30:39 Kalman in finance
46:44 Portfolio optimization
57:34 Summary
59:05 Questions
Thanks for the timestamps!
I think that all are amazing how do we put them into practice is an another issue
41:14 most dramatic part of the talk "the regime has changed, cointegration has been lost ! the gamma, the gamma has changed! " this quote gets the oscar
students and theorists dont understand how almost all of that is fake un real life trading
I wish my professor would give a presentation like you, so enthusiatic that my attention didnt move away for the whole hour
Thanks!
Lecture is very beautiful…. No time
This is absolutely amazing, currently an undergraduate studying mathematics with limited statistical knowledge however your concepts especially discussing co-integration and using Kalman filter for pairs trading was incredibly insightful. Took plenty of notes and will definitely be returning to this video. Huge thanks Professor!
A working quant here , This talk definitely motivates me to check out signal processing side of theory.
Your energy, research and presentation!🔥 Thank you so very much Sir! I'm an EC engineer in the last year interested in finance. You just gave me the perfect thing for my final year project!
Magnificent talk. If only all lecturers were as enthusiastic and engaging as yourself. Bravo!
porfavor nunca deje de subir videos! desearia poder seguir estudiando en universidad para tomar estas oportunidades, lo bueno es que pude encontrarlo en youtube y aprender de sus papers, libros y videos, gracias por compartir lo que sabe por aqui! espero poder seguir viendo mas de usted
Thanks!!
I really appreciate you, prof.. I have solid finance and statistics background, but your codes have been very helpful🎉🎉🎉🎉🎉
I cant count how many times I watched this lecture. It is like an interesting movie!
Engaging speaker and excellent presentation! Love this video. Thanks for posting.
Thank you Prof. Daniel for very attractive talk
My pleasure!
Wow, I just wanna say this is absolutely beautiful work.
Thank you sir! watching this in 2022 and still very relevant!
I can confirm his book is still available for free on the U of HK site.
Thanks for the material. In the slide "LS Regression for Pairs Trading" (1) what are the dotted horizontal lines? I'm assuming these lines are used as thresholds for buy/sell signals. (2) When you say 'gamma' is this the same thing as the coefficient between each pair after fitting a linear regression?
hey i have observed there as is cointigration and cointegration if u use lenghth instead of a fixed point as u used as from zero due to this slowly slowly it is not cointegreating . i used dynamic length so we will always use latest data instead of fixed point .
33:36
This is impressive! Thanks
Glad you like it!
This is a very excellent talk. I love the contents, and more love the professor's passion :)
This is a gold mine. Keep up the good work.
s/o from South Africa
amazing lecture sir!
Fantastic talk, really enjoyed it. In today's world is it plausible that a lay person can successfully compete in pairs trading? I've a background as an RF engineer and now working in data science. Thanks
really good information provided in short amount of time. Kudos 👍
6:35. SP500 log returns are locally stationary but that the argument for assigning it a non-stationary designation via the existence of volatility clustering of any type is not valid. Volatility clustering doesn't not necessarily imply non-stationary. Slowly varying unconditional volatility implies non-stationarity but GARCHian class volatility for most parameters does not.
You are correct indeed! 👍
Excellent presentation. Will take a look at your book!
Awesome, thank you!
Great talk! Why wouldn't you just retrain your kalman filter nightly instead of using a fixed training period?
Indeed, you can retrain the model whenever convenient. For more information, please check Chapter 15 in my new book: portfoliooptimizationbook.com
Thanks for sharing! Love your enthusiastic way of teaching!
Amazing lecture made it very easy to understand!
Amazing talk!
Thanks for the video lecture!
In slides @8:50, where the histograms are shown, how was the histogram bin width chosen? Thank you.
Amazing lecture. Thank you!👍🏾👍🏾
Great talk
Excellent!
Just wow
Gracias Daniel. ¡Buenísima presentación! Ahora habrá que implementar
Where was signal processing in the talk?
Crazy how people spend upwards of $100k or more on undergrad & masters tution to learn about all these information and it just sits here free. What a world we live in.
But to get a job offer or post doc positions, they will still require formal qualifications. People are essentially paying for the formal qualification (and networking of course)
and Havard, MIT open cources are all free, great time for learners indeed!!!
Absolutely fantastic lecture!
Thanks for this talk, it was great! What was the name of the talk was given by your student?
Thx for the content!
Can anyone recommend me a book that details the contents of this presentation to further study the topic?
Some material appears here:
Yiyong Feng and Daniel P. Palomar, A Signal Processing Perspective on Financial Engineering, Foundations and Trends® in Signal Processing, Now Publishers, 2016.
[palomar.home.ece.ust.hk/papers/2016/Feng&Palomar-FnT2016.pdf
@@danielpalomar Hello Sir,
Link is not working.
@@gs-e2d Fixed!
In this webpage you can find my new book as well as slides, R code, and Python code that I will be adding: portfoliooptimizationbook.com
Are there any good books that also go over these topics on signal processing?
I am happy to say that I have just released my new book, available online for free: portfoliooptimizationbook.com
@@danielpalomar thank you!
@@danielpalomar thank you very much Prof. Daniel. If am starting, should I start with this one or the Perspective book?
@@sultanalshirah The Perspective book is much shorter indeed. Of course you can also read selected chapters in the new longer book (e.g., chapters 1,2,3,6,7,8,...). Up to you.
Is there anything you can point to in terms of implementation details for the Kalman strategy? Mainly, 1) do you track position and volatility or just position, and 2) is there a process for setting the initial parameters for the model? And apart from the Kalman element, is there a recommended procedure for determining the pairs?
Very interesting video.
For more information, please check Chapter 15 in my new book: portfoliooptimizationbook.com
Great presentation sir. I have a question. The efficiency of kalman filter against rolling regression doesnt depend on the signal/noise ratio? I ve seen terrible results of KF in this context...
It all depends on the choice of the parameters. If properly chosen, Kalman is always superior than a rolling regression. For more information, please check Chapter 15 in my new book: portfoliooptimizationbook.com
This is really interesting 😃
Thanks for a great talk. where can one download the slides ?
In this webpage you can find my new book as well as slides, R code, and Python code that I will be adding: portfoliooptimizationbook.com
brillant
Sir, can i get the PPT?
In this webpage you can find my new book as well as slides, R code, and Python code that I will be adding: portfoliooptimizationbook.com
He repeatedly misspelled 'Markov' in his slides. What a weird mistake to make for someone who works in his field. (that aside the talk was very interesting and he presented it masterfully ...)
No
I used to think like this, too, until I grew up
No hay que dar por hecho que la audiencia sabe todo y sugiero que al presentar estes un poco mas calmado. De todas maneras gracias
When he likened intraday charts to the quantum world, I died fam...
Very charming man
37:44
😉
There is no time in this conference XD
Incredible talk, I get sinister giggles when smb treats Gaussian like a slut!
free alpha
Shit it does look like speech recording, he's on to something...
Kalman Rushdie !
ta loco
Fcuk rocket science. This blows one's mind 😭😍😍
Palomar is sweet 🫡