Quant Casual Update

I just added RSS to quantcasual.com at quantcasual.com/rss.

There's still a few days of the best quirky idiosyncratic news from Finance Twitter before it gets more, 'educational' and straight quant oriented...

In any case, news today leads with 'fuck off essays on Facebook'.


What we know about Paris terrorists -Not Syrian -Not refugees -No encryption What the US is focusing on -Syrians -Refugees -Encryption

A new version of Python's Pandas has been released!

Another thing that has been in the works for a while, but getting ever more extreme:

The top 10 stocks in the S&P 500 are +13.9%. The other 490 are -5.8%. Largest spread since the late 1990s.

Next up a paper on 'Sequence to Sequence' machine learning which overcomes a limitation of Deep Neural Networks, namely that you need to know the dimensionality of the problem beforehand. I.e. with speech recognition or language translation, you don't necessarily know how long a sequence of words will have to be until you can translate the meaning - sentences and ideas are expressed in varying amounts of words.

Strongly suspect that could come in handy in financial time series analysis also, where the length of history one needs to come up with a buy or sell signal could fluctuate.

I really need to brush off my scepticism and get into machine learning.