How to handle the volume spikes near expiration for futures contracts?

Question

Are there any common practices to handle the volume spikes that occur near expiration of a futures contract?

I intend to use volume of futures contracts as a predictor. However, due to the rolling activity near the expiration, there are spikes around expiration. These tend to be asset specific (quarterly contracts, monthly contracts etc.). While time series techniques like:

1. ARIMA for seasonality
2. FFT transform

do sound promising, have been impractical in my experience. Are there any common techniques used around for this ?

Answer 1

I worked for a major systematic trading hedge fund. We traded futures extensively. We would set specific windows of time relative to either first notice date or last trade date where we would avoid trading contracts. Ie, we would reduce our position to zero when the contract got too short. Exactly what those windows were was obviously a research topic.

If your strategy intends to use trade volume as a signal, but the volume around expiration is not the market process you are looking to sample, then don't sample it. Roll your positions forward to further contracts ahead of time.

Answer 2

As other answers say you normally don't. You avoid trading it normally a couple of days before expire and you tend to find liquidity dries up on the day of expiring. Volume-based roll is a common way to know when to roll.

ARMIA assumes some regression to the mean and normally isn't used to predict volume.

Links:

https://www.sierrachart.com/index.php?page=doc/ContinuousFuturesContractCharts.html#ContinuousFuturesContractVolumeBasedRollover

https://adamhgrimes.com/how-to-calculate-futures-rolls/