Commodity · Global · CFDs, ETFs, futures · Monthly · Global Free

Forecasting Crude Oil Prices

Academic paper

Forecasting Crude Oil Prices: Does Global Financial Uncertainty Matter?

AuthorsYong Ma; shuaibing li; Mingtao Zhou

Institute

Hunan University of Finance and Economics
?Hunan University - School of Finance and Statistics
?Hunan University, School of Finance and Statistics
Hunan University

Paper linkhttps://papers.ssrn.com/sol3/papers.cfm?abstract_id=4565543

Strategy in a nutshell

Invest in crude oil derivatives (futures or spot). Use a univariate regression with GFU as predictor to forecast next-month oil returns. Buy or short monthly based on GFU signals. Portfolio: single position, monthly rebalancing. Suitable for investors with risk aversion coefficient of 3.

Economic rationale

GFU is a robust alternative data predictor for crude oil returns. It improves forecasting accuracy, remains statistically and economically significant, and enriches existing literature on commodity market timing strategies.

Backtest performance

Annualised return4.65%

Volatility19.38%

Beta-0.206

Sharpe ratio0.24

Win rate31%

Full Python code

1from AlgorithmImports import *
2import data_tools
3import statsmodels.api as sm
4# endregion
5
6class ForecastingCrudeOilPrices(QCAlgorithm):
7
8def Initialize(self):
9self.SetStartDate(2000, 1, 1)
10self.SetCash(100000)
11
12leverage:int = 3
13self.oil:Symbol = self.AddData(data_tools.QuantpediaFutures, 'CME_CL1', Resolution.Daily).Symbol
14self.Securities[self.oil].SetLeverage(leverage)
15
16self.gfu:Symbol = self.AddData(data_tools.GFU, 'GFU', Resolution.Daily).Symbol
17
18# self.test_size:float = 0.2
19
20self.min_period:int = 12
21self.price_storage:Dict[Symbol, List] = {}
22for symbol in [self.oil, self.gfu]:
23    self.price_storage[symbol] = []
24
25def OnData(self, data: Slice) -> None:
26rebalance = False
27
28# store daily price
29if self.gfu in data and data[self.gfu]:
30    self.price_storage[self.gfu].append(data[self.gfu].Value)
31    self.price_storage[self.oil].append(self.Securities[self.oil].Price)
32    rebalance = True
33
34# rebalance once new GFU data arrived
35if rebalance:
36    last_update_date = [c.get_last_update_date() for c in [data_tools.QuantpediaFutures, data_tools.GFU]]
37    if all(self.Securities[symbol].GetLastData() and last_update_date[i] > self.Time.date() for i, symbol in enumerate([self.oil, self.gfu])):
38        if all(len(self.price_storage[symbol]) >= self.min_period for symbol in [self.oil, self.gfu]):
39            oil_prices:np.ndarray = np.array(self.price_storage[self.oil])
40            x:np.ndarray = np.array(self.price_storage[self.gfu])[1:]
41            y:np.ndarray = np.log(oil_prices[1:] / oil_prices[:-1]) # oil monthly log returns
42
43            # run regression
44            model = data_tools.multiple_linear_regression(x[:-1], y[1:])
45            predicted_oil_return:float = model.predict([1, x[-1]])[0]
46
47            # test_size:int = int(len(self.price_storage[self.oil]) * self.test_size)
48            # model = data_tools.multiple_linear_regression(x[:-1][:-test_size], y[1:][:-test_size])
49            # predicted_oil_return:float = model.predict(sm.add_constant(x[-test_size:]))[-1] if test_size > 1 else model_.predict([1, x[-test_size:]])[-1][0]
50
51            # trade based on prediction
52            if predicted_oil_return > 0:
53                self.SetHoldings(self.oil, 1)
54            else:
55                # self.Liquidate(self.oil)
56                self.SetHoldings(self.oil, -1)
57    else:
58        self.Liquidate()
59        return