Notebook

Supercharged resources/packages¶

A compendium of great python packages you can use, now that you have the skills needed to access the info below.

Please suggest others if you see something cool!

Package	Description	Where
streamlit, plotly, and shiny	Building interactive graphs, dashboards, and apps!	streamlit, plotly, shiny
OpenBloomberg	85% of a Bloomberg terminal for free, see the tweet below for a preview.	https://github.com/OpenBB-finance/OpenBBTerminal
pandas-datareader	Import a ton of data	https://pydata.github.io/pandas-datareader/remote_data.html
auto-sklearn	Automated Machine Learning with scikit-learn	https://github.com/automl/auto-sklearn
Open Source Asset Pricing	Data for asset pricing tests and code to replicate or use. If you are interested in asset pricing or trading, this is a MUST visit. It's so cool that this is freely available, it blows my mind.	https://www.openassetpricing.com/data/
PyQL	QuantLib's Python port.	https://github.com/enthought/pyql
QuantPy	A framework for quantitative finance In python.	https://github.com/jsmidt/QuantPy
vollib	Calculating option prices, implied volatility, and greeks	https://github.com/vollib/vollib
Finance-Python	Python tools for Finance	https://github.com/alpha-miner/Finance-Python
ffn	A financial function library for Python.	https://github.com/pmorissette/ffn
pynance	Lightweight Python library for assembling and analysing financial data.	https://github.com/GriffinAustin/pynance
pysabr	SABR model Python implementation.	https://github.com/ynouri/pysabr
FinancePy	Focuses on the pricing and risk-management of Financial Derivatives, including fixed-income, equity, FX and credit derivatives.	https://github.com/domokane/FinancePy
gs-quant	Goldman-Sachs toolkit for quantitative finance	https://github.com/goldmansachs/gs-quant
willowtree	Willow tree lattice for derivatives pricing.	https://github.com/federicomariamassari/willowtree
financial-engineering	Applications of Monte Carlo methods to financial engineering projects	https://github.com/federicomariamassari/financial-engineering
optlib	A library for financial options pricing written in Python.	https://github.com/dbrojas/optlib
tf-quant-finance	High-performance TensorFlow library for quantitative finance.	https://github.com/google/tf-quant-finance
Q-Fin	A Python library for mathematical finance.	https://github.com/RomanMichaelPaolucci/Q-Fin
Quantsbin	Tools for pricing and plotting of vanilla option prices, greeks, and various other analysis around them.	https://github.com/quantsbin/Quantsbin
finoptions	Complete python implementation of R package fOptions with partial implementation of fExoticOptions for pricing various options.	https://github.com/bbcho/finoptions-dev
AssayingAnomolies	Quickly check if your signal generates alpha	Coming Soon
usaddress and probablepeople	Parsing messy names (people, address, firms)	https://github.com/datamade?q=&type=all&language=&sort=stargazers
SRAF	Textual analysis resources, designed for business text.	https://sraf.nd.edu/

Chat GPT and Generative AI

Chat GPT and other generative AI are at least, of course, fun novelties. Who doesn't love this:

{image}
:alt: kingjames
:width: 75%
:align: center

Hilarious and silly!

But generative AI also have the potential to truly supercharge how work is done across many tasks and industries.

Developments in this area are happening so fast, what I write here will be outdated very soon, so I'll keep it brief here and spend more time on the topic during the semester.

But as of November 2022, here are some thoughts on OpenAI's ChatGPT:

View it as having an infinite supply of interns with an 8th-grade IQ. Meaning:
- Low-cost way to brute force some tasks (brainstorming, rough drafts, outlining, editing, starting code)
- Its creativity is limited (bad at: "write a shocking surprise that was foreshadowed earlier")
- But it can be harnessed: "write a rap about monetary supply in the style of Common", "draft a book proposal/sales copy/etc on topic X", "write a corporate memo addressing X, Y, and Z", "suggest 3 options for X", "provide more detail on the second point"
- Responses are often vague and lack specifics
Do not assume its output is correct or the truth
- Meaning: It will confidently lie and make up things (so-called "hallucinations" and "bullshiting")
- It's not all wrong though - see the note in the margin

{margin}
Counterpoint: [@pythonprimes](https://twitter.com/pythonprimes/status/1601664776194912256) had ChatGPT take the
- LSAT (result: 149)
- bar exam (passed!)
- MPRE legal ethics (60%)
- Medical boards (70%)
- NY high school English (91%) and Chem tests (78%)

"[Everyone is (correctly) worried](https://twitter.com/emollick/status/1601993380619513857) about the fact that AI gets stuff wrong, but it still gets enough right to pass the bar, our standard for how right HUMAN lawyers need to be."

Those points imply that ChatGPT is only useful and accurate when paired with

Your guidance (via iterative conversation)
Your subject matter expertise (to correct errors)

Thus, ChatGPT does not look anything like a replacement for high-skill knowledge workers. It is a complement to our workflow - I always have it open as a tab on my computer now.

There are many implications for this technology - mostly exciting! But it certainly looks like a world-changing tool. Examples:

Makes new business formation/expansion easier and democratizes access
- Start-up time on (coding) projects will decline drastically
- Building websites and prototyping production code rapidly
- Makes outputs achievable for relative novices
Writing workflow will change dramatically
- Shift in emphasis to idea creation, editorial choices, truth checking
- Useful in many domains: journalism, memos, marketing, consulting, etc.
- Editing - overall writing quality should increase
The cost of BSing and cheating will fall
- Typed essays decline in favor of oral evaluations, handwritten essays
- Spam and fishing attempts will become much better

OpenBBTerminal

This Twitter thread has a nice preview of the OpenBBTerminal package, and you can find more in the official GitHub documentation.

In [1]:

import requests

class Tweet(object):
    def __init__(self, s, embed_str=False):
        if not embed_str:
            # Use Twitter's oEmbed API
            # https://dev.twitter.com/web/embedded-tweets
            api = 'https://publish.twitter.com/oembed?url={}'.format(s)
            response = requests.get(api)
            self.text = response.json()["html"]
        else:
            self.text = s

    def _repr_html_(self):
        return self.text

Tweet("https://twitter.com/thebuoyantman/status/1558792415263088641?s=27&t=BgGJLwO4zpRdwJkH90T5QA")

Out[1]:

A Bloomberg subscription costs $24,000 per year.

But you can get nearly 85% of its features for free.

It's through an app called Open Bloomberg (aka OpenBB)

Here's how:

🧵
— Shravan Venkataraman 🔥🚀💰 (@theBuoyantMan) August 14, 2022