Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Final Presentation 070219

Methodologies: Heatmaps, Alphalens, Pyfolio
Factor: Volatility
Session returns: Intraday
Time period: 2014/01/01 - 2017/11/06

Idea: Building a long-short strategy depending on the volatility factor assuming that the stocks which are highly volatile will perform bad intraday, while the stocks which are less volatile will perform well intraday => high quantiles are shorted and low quantiles are longed. In here, volatility is defined as volatility of daily returns (the annualized standard deviations of the last 20 days). This strategy goes against the idea "Higher risks, higher returns", which could be beneficial for risk adverse investors.

Heatmaps: Intraday returns drop significantly (0-9q). In lower quantiles, there more positive returns than in higher quantiles. The number of observations tend to concentrate on the inverse diagonal of the heatmap.

1 response

Factor: Volatility (daily returns) (in the code I used -Volatility)
Universe: QS500
Quantiles: 0-4q (5 quantiles)
Returns: Intraday for 1 day, 5 days, 10 days

Alphalens
Returns Analysis: Annual alpha is positive for three types of returns, which means the stocks outperformed the benchmark index. Top quantiles have postive returns, low quantiles have negative returns, but since -Volatility is used in the code, this means low quantiles has better performance than higher quantiles.
Information Analysis: Information coefficient shows how closely our forecasts are to actual results. IC mean increases as the period of three return types increases.

Companies per day: the plot shows high fluctuation but actually only within small range 484 - 500.

Pyfolio
Annual returns: 8.9%, Cumulative returns: 60.5%. Sharpe ratio: 0.98 (risk adjusted returns). Negative beta means inverse relation to the market.