Alpaca is a securities broker with a free-to-use API for stock trading. In this project we use it to paper trade with Python.

The algorithm/strategy used in this project is the Double Exponential Moving Average. The idea is that we compare 2 double exponential moving averages with different windows over time. We buy when the "fast" DEMA line (the one with a shorter window) crosses above the "slow" DEMA line (longer window) and sell when the fast crosses the slow below. In this code, the DEMA lines are updated every minute during market open (hence the time.sleep(60)).

1. Any moving average will lag behind the actual data.
2. Volatility might result in false crossovers.

We could schedule this to run daily in the cloud (e.g. pythonanywhere) or locally with task scheduler (Windows).

Our fast and slow values/windows seem quite arbitrary. I could possibly dynamically adjust them with some equation or use machine learning to do so. Such a solution would probably take into account stock volatility, market volatility, beta, etc.