Reference no: EM133797066

Task  - Forecasting

The task involves analyzing the asset pricing of the German stock exchange. And forecast pil prices with the usage of futures and spot prices.

Key Responsibilities:
- Conduct Fama-Macbeth regression analysis on oil the German stock exchange and acess with GRS test.
- Forecast future price movements in oil, such as point forecast, random walk, out of sample forecasting
- Evaluate forecasting errors
- Assess the Capital Asset Pricing Model (CAPM)

Question 1. Continue with the data from the previous task. Treat front-month futures prices as point forecasts of the spot price one month ahead.

Question 2. What is the distribution of forecast errors from the futures prices? Plot their histogram.

Question 3. Evaluate the forecasts statistically; refer to points a, b and e from the lecture slides:

• Are the futures prices unbiased forecasts of the subsequent spot prices? (What is the point estimate of the bias, what is the standard error and what is the p-value?)
• Are they optimal in terms of autocorrelation at various lags?
• What about the Mincer-Zarnowitz test? Beware of the unbalanced regression problem (forecast errors likely being stationary while the dependent variable likely being integrated).

Question 4. Suppose an economic actor suffers a loss from each nonzero forecast error. The loss is a scalar multiple of the absolute value of the forecast error. Choose any positive scalar as the multiple (and specify it explicitly).

• What is the estimated expected loss?
• What is the estimated distribution of losses? (Draw a histogram. Book Our Assignment Help Service Now!)
• What is the estimated distribution of losses conditional on the realized price being higher than the forecasted price? (Draw a histogram.)

Question 5. Consider a random walk model for the spot price. (Ignore the model's obvious flaw.) Obtain point forecasts derived from this model. Are they superior to the futures prices in terms of the (estimated) expected loss? Interpret the point estimate, but also use the Diebold-Mariano test.

Question 6. Split your dataset so that you have a training-and-validation subset and a test subset. Use the former subset to design and estimate (and perhaps tune) a forecasting model for the spot price. It can be as simple or as intricate as you wish. Iterate model estimation (and tuning) and forecasting in a rolling window fashion.

• Compare your model's Demand forecasting performance on the test subset with the futures and random walk forecasts in terms of estimated expected loss and distributions of losses (plot histograms, describe their differences).