In a backtest I did, I'm selling a call option and buying a delta amount of the underlying (calculated using implied vol). Now I know in the limit case of continuous hedging I end up paying a PnL which is a gamma weighted integral of volatility differences, but in discrete hedging (daily) backtest, I'm getting slightly positive PnL's. I'm unable to reconcile this with an intuition. What must drive such a result?
I expect to lose PnL because implied vol seems to be consistently smaller than the realized vol (if calculated as sample standard deviation on a rolling basis)