So, thanks to my former boss, and head of direct indexing at BNY Mellon, Vijay Vaidyanathan, and his Coursera course, along with the usual assistance from chatGPT (I officially see it as a pseudo programming language), I have some more software for the Python community now released to my github. As wordpress now makes it very difficult to paste formatted code, I’ll be linking more often to github files.

In today’s post, the code in question is a python file containing quite a few functions from R’s PerformanceAnalytics library outside of Return.portfolio (or Return_portfolio, as it may be in Python, because the . operator is reserved in Python…ugh).

So, here is the “library” of new functions, for now called edhec_perfa.py, as a nod to Vijay Vaidyanathan’s Coursera course(s).

Here’s the link to the latest github file.

Now, while a lot of the functions should be pretty straightforward (annualized return, VaR, CVaR, tracking error/annualized tracking error, etc.), two functions that I think are particularly interesting that I ported over from R with the help of chatGPT (seriously, when prompted correctly, an LLM is quite helpful in translating from one programming language to another, provided the programmer has enough background in both languages to do various debugging) are charts_PerformanceSummary, which does exactly what you might think it does from seeing me use the R function here on the blog, and table_Drawdowns, which gives the highest N drawdowns by depth.

Here’s a quick little Python demo (once you import my file from the github):

import yfinance as yf
import pandas as pd
import numpy as np

symbol = "SPY"
start_date = "1990-01-01"
end_date = pd.Timestamp.today().strftime("%Y-%m-%d")
df_list = []

data = pd.DataFrame(yf.download(symbol, start=start_date, end=end_date))
data = data["Adj Close"]
data.name = symbol
data = pd.DataFrame(data)

returns = Return_calculate(data)

charts_PerformanceSummary(returns)
table_Drawdowns(returns)

This results in the following:

And the following table:

The NaT in the “To” column means that the drawdown is current, as does the NaN in the Recovery–they’re basically Python’s equivalent of NAs, though NaT means an NA on *time*, while an NaN means an NA on a numerical value. So that’s interesting in that Python gives you a little more information regarding your column data types, which is kind of interesting.

Another function included in this library, as a utility function is the infer_trading_periods function. Simply, plug in a time series, and it will tell you the periodicity of the data. Pandas was supposed to have a function that did this, but due to differences in frequency of daily data caused by weekends and holidays, this actually doesn’t work on financial data, so I worked with chatGPT to write a new function that works more generally. R’s checkData function is also in this library (translated over to Python), but I’m not quite sure where it was intended to be used, but left it in for those that want it.

So, for those interested, just feel free to read through some of the functions, and use whichever ones you find applicable to your workflow. That said, if you *do* wind up using it, please let me know, since there’s more value I can add.

Lastly, I’m currently in the job market *and* have a volatility trading signal subscription available to offer for those interested. You can email me at ilya.kipnis@gmail.com or find me on my LinkedIn profile here.

Thanks for reading.