The article develops a method to estimate parameters in models that combine nonlinear autoregressive patterns with varying levels of volatility. The researchers show that their estimation technique is reliable and accurate, even when the data is complex. This is the first time such results have been achieved in this type of model.