Trading Equity Indices Using Time Series Models

Time series models like ARIMA, or Autoregressive Integrated Moving Average, and VAR, or Vector Autoregression, are essential tools for forecasting sequential data points over time, making them invaluable for investment analysis and decision-making. These models can capture and analyze historical patterns and trends in financial markets, helping investors identify potential opportunities and risks.

Reference [1] investigates the use of ARIMA and VAR models for forecasting the prices of the DJIA, NASDAQ, and NYSE indices. Specifically, it examines three hypotheses. The authors pointed out,

Three hypotheses were stated at the beginning of this paper. The first one was: ARIMA and VAR models have similar forecasting power. The second one stated: VAR models should be more robust to the changes than ARIMA models. The third one: The model with more accurate forecasts might not perform better when applied to an algorithmic investment strategy. Based on the delivered results, the first hypothesis is rejected as ARIMA model had lower forecasting errors than VAR model. The second hypothesis can also be rejected as the results showed that the VAR model was less robust to the changes and obtained a higher standard deviation of information ratios compared to the ARIMA model. Although ARIMA performed remarkably well during volatile periods, VAR based strategies mainly outperformed ARIMA based strategies in terms of its lower portfolio risk and higher risk-adjusted return measures. Regarding the third hypothesis, the obtained results seem to be consistent with this hypothesis. ARIMA had lower forecasting errors while the performance statistics showed that VAR based investment strategies outperformed ARIMA based investment strategies. It can be concluded that we failed to reject the third hypothesis, meaning that there is not enough evidence to reject the statement that the error metrics may not be a reliable measure to evaluate the performances of models.

In short, the paper concludes that

• Models with more parameters aren't always better.
• Additionally, relying on error metrics to evaluate a forecasting model can lead to inaccurate results. It's preferable to integrate the forecast model into trading strategies and then assess the performance of those strategies.

The authors' conclusions are consistent with our practical observations.

Let us know what you think in the comments below or in the discussion forum.

References

[1]  Sahil Teymurzade, Robert Ślepaczuk, Predicting DJIA, NASDAQ and NYSE index prices using ARIMA and VAR models, Working Papers 2023-27, Faculty of Economic Sciences, University of Warsaw

Originally Published Here: Trading Equity Indices Using Time Series Models

source https://harbourfronts.com/trading-equity-indices-using-time-series-models/