If I interpret correctly, autoregressive means past has impact to today, which is the opposite of mean reversing. In the example of one lag volatility, means todays volatility has yesterday's volatility and other components. In a simple term, it can be said that if today's volatility is high (low), tomorrow's volatility is more likely to be high too (low).
convergence of implied and historic vols is decent signal in fx markets...not sure about equites...might be worth looking at... just my 2 cents
(I'm reactivating this old thread about volatility forecasting) Forecasting the volatility of the underlying, and also the price direction of the underlying, are two of the most important criteria when entering a trade. Which forecast methods are mostly in use nowadays? To which degree is volatility forecastable? How is the usual success rate? Which one is harder: volatililty forecast or weather forecast? What experiences have you made with volatility forecast? Which experiences have you made with directional forecasts, ie. price direction of the underlying?
Here are some charts with historical volatility plotted: How to best predict the volatility? Could LR or NonLR be useful here?