CNN's "Fear and Greed Index" is a popular metric for measuring the general mood of investors. As we saw in Gus's popular community post (https://www.quantopian.com/posts/using-the-cnn-fear-and-greed-index-as-a-trading-signal), use of the Fear and Greed index as a trading signal varies. Mean reversion strategies buy stocks when the crowd is afraid. Momentum strategies buy when the crowd is greedy.
Unfortunately, CNN doesn't release the raw data sources used in their Fear and Greed index. They also don't provide CSV of index values, let alone an API, that would allow the index to be easily used as a signal in an algorithm.
My hope is to eventually evaluate the Fear and Greed Index as a trading signal. However, I don't want to rely on the hacky scraping of a chart from CNN.com. I'd like to recreate the Fear and Greed Index with a combination of Quantopian's price data and easy to access external data sources.
While I feel like I am on the right track, many parts my NB could certainly be improved. I could use some help finding the best way to equally weight all the metrics and account for the components' past values (as CNN does). Right now I am using rolling z-scores with decay.
Any suggestions on the best way to test the predictive power of this index and its components in a regression or ML algo are welcome! I'll keep posting what I find.
