View Code? Open in Web Editor
NEW
Can a smallish feedforward neural network learn from 15+ years of stock market data and predict asset price movements?
finance-nn's Introduction
- Fetch fundamental financial data
- Create training data set (exclude 2020)
- Analyze data (correlations, descriptive statistics)
- Predict S&P daily change with neural nets
- Validate performance
- If successful: Test robustness for 2020
- Stock markets: S&P500, Nasdaq, Dax, FTSE 100, Hang Seng, SSE, Nikkei, Ibovespa (8)
- Volatility: VIX (1)
- Bond rates 3M, 5Y, 10Y, 30Y (4)
- Commodities: Gold, silver, copper, crude oil, gas, corn, lumber, cocoa, cement (9)
- Inflation 5Y Breakeven, 10Y Breakeven, 5Y FWD (3) # (not implemented yet)
- Currencies: USD-EUR, USD-GBP, EUR-CHF, USD-JPY, USD-HKD, USD-CNY, USD-SGD, USD-INR (8)
- Day of the week (5)
- Month (12)
finance-nn's People
Contributors
Watchers