Predicting Intraday and Daily Volumes Using ARIMA Model

Volume is an essential, integral market data. However, it receives much less attention in research literature compared to price data. Understanding and being able to model volume dynamics is important because buy-side firms must plan and time their trades to avoid significantly impacting the market, revealing their identities, and incurring excessive transaction costs. Sell-side institutions require knowledge of trading volume to make markets efficiently and need accurate forecasts to implement strategies related to volume, such as those that track some form of volume-weighted average price (VWAP).

Reference [1] studied the dynamics of intraday and daily volume data. The authors pointed out,

In this analysis, we focus on maximizing predictive power of time-series models in forecasting the intraday and daily trading volume of SPY according to the metrics MSE, MAPE, and VWAP error. Our intraday analysis indicates that using SARIMAX with the exogenous variables average directional index, exponential moving average, and momentum give us the optimal forecast, outperforming SARIMA and a spectral representation of the data using m = 3 Fourier frequencies. However, all three models significantly outperform our naïve baselines with respect to tracking VWAP. However, when performing our analysis of daily volume data, we see a lack of seasonality. This is confirmed by R choosing ARIMA and ARIMAX models over SARIMA and SARIMAX models during cross validation. Also, a higher m value of 40 for the FDPR model is optimal. But using exogenous variables with ARIMAX still gives the best predictions for daily data. Overall, we have shown that trading volume can be accurately predicted using ARIMA models with exogenous variables and adding seasonal components when necessary.

In short, the author demonstrated that volume can be predicted with reasonable accuracy using the ARIMA model.

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

References

[1] A Krishnan, M Pollack, A Cooper, Unraveling the Dynamics of SPY Trading Volumes: A Comprehensive Analysis of Daily and Intraday Liquidity Trends, arXiv preprint arXiv:2406.17198, 2024

Originally Published Here: Predicting Intraday and Daily Volumes Using ARIMA Model

source https://harbourfronts.com/predicting-intraday-daily-volumes-using-arima-model/