I am now working on a time series data and I would like to quantify the series' volatility.

I have think of an approach which is first model it as some auto-regressive model like ARCH and compute the sum of squared residuals between the model and the real data.

Does it make any sense?