A naive model is a simple model that uses basic heuristics, benchmarks, or simple statistics.According to the content learned in the course and my understanding, a naive model can be:
- Previous Day Close as Prediction: A simple naive model could be using the previous day's closing price as the prediction for the next day. This approach assumes the market conditions don't change significantly day over day.
- Moving Average: Use a simple moving average of the past few days (e.g., 1-week or 4-week moving average) as the prediction. This model smooths out day-to-day fluctuations and reflects longer-term trends.
- Constant Mean Model: Use the historical average of Bitcoin prices as the prediction. This model assumes that the best guess for tomorrow's price is the average of what it has been in the past.
We can use metrics such as Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), or Mean Absolute Percentage Error (MAPE). A model that significantly outperforms these naive benchmarks in these metrics can be considered to have predictive value beyond just making simple assumptions.