Databases Reference
In-Depth Information
2. Create the time series of daily log returns of the stock price.
3. Just for comparison, do the same for volume data (i.e., create the
time series of daily log changes in volume).
4. Next, try to set up a linear regression model that uses the past two
returns to predict the next return. Run it and see if you can make
any money with it. Try it for both stock returns and volumes.
Bonus points if you: do a causal model, normalize for volatility
(standard deviation), or put in an exponential decay for old data.
5. Draw the cumulative P&L (forecast × realized) graphs and see if
they drift up.
