I recently experienced imposter-syndrome when I was found out to not know how the Black-Scholes model works 😳.
To fix this problem I wrote up an intro piece on how the Black-Scholes model works where I cover the intuition behind it.
Here's an interesting way of modelling mean reversion in government bonds and estimating the return on ETF spreads.
I made an interactive viz with
@observablehq
to show how it works. Pretty neat.
Immigration is strange.
In order for me to stay in the UK, a fellow Commonwealth country, I have to fly back to Australia, ask permission to fly back and then fly back. I wonder what they're scared of? 🤔
Oh well. 🛫🇬🇧🇦🇺
I love Gaussian processes. The Wiener-Khinchin theorem (sexy!) is one of my favourite tricks for modelling periodic time series. Check out my write up here:
The data visualisation tools built by
@mbostock
&
@mmeckf
at
@observablehq
are awesome! Free JavaScript notebooks that you can share and even embed into your own site.
I've got a cool bit of research coming that uses Observable to understand how Ornstein-Uhlenbeck models work.
There's a really awesome linear algebra trick you can use to understand where your portfolio's risks are: matrix square roots.
I wrote an article about how to do it that uses
@observablehq
to make the equations interactive!
Here's some
#Python
code to model bond ETF returns from interest rates.
There's a bonus section showing how bond returns are negatively skewed when interest rates are low.
#SVBCollapse
#finance
The best thing you can do for your machine learning model is make sure you are optimising the right metric. You get your metric right, everything else happens effortlessly.
I've been estimating statistics for bond ETFs. I wrote up a piece showing how to calculate an ETFs mean and variance based on models of market bond yields.
This is my first use of
@observablehq
to make charts. Very easy, very cool.
A quick reference piece for myself on the moments of a Gaussian distribution.
I am often modelling the expected value of a function of Gaussians. The expected value is a function of the moments of the Gaussians.
If you want to see what alpha looks like, check out this write up of predicting currency prices. Uses some math to create a very consistent forex forecasting model.
Beware, you'll need more alpha signals to overcome transaction costs!
I found a slick derivation of the half-life of a time series based on regressed coefficient. I use this frequently on financial data. I don't have to fit the half-life of mean reversion algorithms.
#timeseries
#finance
#machinelearning
#meanreversion
@PtrPomorski
It's a bit like if you want to learn psychology but the only book you have is a children's bible. There's some valid points in there, but largely misses the mark.
@AlexRBucknall
@NotionHQ
It is! I'm now doing most of my work out of . Really hits the spot. Missing a few critical elements, but leagues ahead of Evernote