I am forced to exist side by side with an HJM type model which in the current crazy market environment produces weird/unrealistic risk analytics. Please, give me some advice on how to straighten the model out. I am sure the issues I'll describe are not new and somebody faced them already.My model is very sensitive to the choice of short rate (it's HJM), so given that front end of the curve has been very volatile lately, it's caused problems. Is that a good idea to throw away the front end of the curve completely and start the curve with, say, 1 year maturity? My model is also normal, so lately in this very low rate environment my short rate had tuned negative quite a bit. We bounce short rate off a certain minimal level, but it creates too much noise in the model. Besides, the volatility function in the model depends on the level of rate, so once those rate paths get near zero, they can't get out.Do these issues ring any bell?

Using short rate in the HJM framework isn't very common, is it?

Forgive my ignorance, if I'm wrong, but I thought that all model derived from HJM framework can be interpreted as short rate models. Isn't that true?

Did they recently invent any products referring to the short rate? So far everything was based on Libor and swap rates.

they invented overnight rate, but that's as close as they could get to the short rate. you're not answering my question though. looks like you're trying to wiggle you way out, but no no no, I cornered you. be a man (or woman) and give me a straight answer!

Yura, the reason behind inventing HJM was to abandon an artificial concept of the short rate. So skip it, work on forward rates and everything will be OK.

I am kind of not sure about your question on "choice of short rate", would you be more specific? Under the HJM framework, why would you try to choose any short rate? Are you trying to construct current forward rate curve (which is assumed to be given in HJM) ?