Predictive algorithm selection for time series - Python

- Build Database with Forex Broker's EUR/USD prices vs Datetime (Broker's historical limit is 10000 price points per week)
- Identify in-database best algorithm for EUR/USD price prediction, at the lowest possible computing cost/time.
- Tune model to find optimum amount of past price points for the prediction calculation, as well as the ideal price difference (higher or lower) for the prediction.
- Adapt given robot to ensure orders can be placed based on the chosen algorithm, feature optimisation and tuning.