webclinic017 / project_sp500_analysis_forecast

Where to put your money with COVID-19 on the loose. In this volatile year is the stock market a safe bet? Predicting the next 6 months.

PredictingTheSP500

What to do with your money in hard times - a look at some money options

Gordon D. Pisciotta

Money Options:

• Low Returns
  • “Under your mattress”
  • High Yield Savings Account
  • Treasury or Government Bill/Bond
• Risky Assets
  • Cryptocurrency
  • High Capital Investments
  • Real Estate
  • Alternative Assets
  • Private Equity
• Alternative Options ~ A Potential Better Route:
  • Stock Market (S&P 500 Index) - measures performance of 505 publicly listed companies

Historical Price Of SP500 Index:

• SP500 - 10 years - daily prices

Stationary Time Series Data

• Price movements tend to drift towards some long term mean either upwards or downwards.
• Stationary TS has constant mean/variance/autocorrelation over time
  

Augmented Dicker-Fuller Test

Null = time series contains unit root and is non-stationary

• p-value > 0.05 = reject null (nonstationary)

Alternative - time series does not contain unit root and is stationary

• p-value ≤ 0.05 = fail to reject null (stationary)

ARIMA & SARIMA

• Fit by Grid Search to find lowest AIC - Fit data to Model

PREDICTION:

• Upper & Lower Bounds of Fitted Parameters (7 year Rolling Perdictions - Trailing)
• 95% Confidence Interval (shaded Section)
• Forcast 6 months into future
  • perdiction widens to reflect loss of certainty in the market outlook

SUMMARY:

1. Over time the market outperforms other assets
2. Every year your money in your savings account is losing money due to inflation

LOOKING FORWARD:

1. Generate portfolio from components of SP500

2. Build Automated Trading Model to Trade based on Perdictions/Moving Averages

(1) Kernel-PCA

• Use z-score to normalize data
• Perform Kernel-PCA on the 505 component stocks within the S&P 500 Index
• Generate Outputs:
  • EigenVectors - direction of principal component line
  • EigenValues - amount of variance each component stock (eigenvector) of the S&P 500 Index generates onto the overall price fluctuations over time

Reconstruct Kernel-PCA vs SP500 Index (Price)