Everyone always wants to know what makes a good model. One way to think about it is that a good model takes some already valid piece of practical intuition and quantifies it in a way that then leads to an extension of the original intuition, which is then embraced by pracititoners, whose intuition then can be extended by even a better model, which ….

model_i+1 = f(intuition_i)

intuition_i = g(model_i-1)

Sort of something like that.

The good thing about a blog is you don’t have to think too much before you write, said he defensively.

For example:

OAS = f(zero_volatility yield)

zero_volatility_yield = g(forward_rates)

forward_rates = g(yields_to_maturity)

yield to maturity is a simple model of value. That leads to an intuition about forward rates. That leads to the idea of forward rates as contributing to the zero volatility yield. OAS or HJM then add volatility to the notion of zero-volatility yield.

Not quite right, but you get the idea.