After my first post about Front Running S&P 500 Index Funds I got a lot of awesome feedback on ways that I could improve it. Below are a couple things I tried to address:
- What is with that huge jump in the 2007-2008 backtest?
- Look ahead bias
- Poor trade execution
- Inconsistency in the data
In response I've cleaned up the data, dialed down trade execution (see the notebook for a further explanation), and I included plots of the gross leverage to help explain some of the larger jumps. I did however leave the actual strategy and portfolio balancing mechanism the same so the results to the first post can still be comparable.
ORIGINAL POST
I read "The Hugely Profitable, Wholly Legal Way to Game the Stock Market" by Yuji Nakamura of Bloomberg and decided to try my hand at a quick and dirty version of the strategy/phenomna presented in the article: front running index funds.
Recent trends in investing are showing a huge influx in the amount of money people are investing in index funds. Attracted by their passive management style, and consequently low fees, investors look to index funds as a source of consistent performance that neither beats nor comes up short relative to the market.
For example take the S&P 500.
The S&P 500 effect on the price of companies is nothing new and neither is front running. While "front running" is generally considered illegal, the strategy of buying stock in a company who is about to be added to the S&P 500 or any index is not. Naturally, index funds track the underlying assets of their indices, so when a new security is added there is a large influx of buy orders from funds trying to fulfill their mandate. This wave of orders raises the price of the security, and potentially yields profit for the savvy investor.
To delve into this further we need data on the index changes
We looked to our friends at EventVestor and they provided us a sample of data containing index addition and deletion events for the last 8 years (2007-present).
You will need to upload the data above into the new Quantopian Research platform to fully interact with this notebook.
You can request your own sample data from EventVestor.