Bayesian PCA with one informative prior
- the process of turning a given yield curve into a set of component scores
-
To look at in the future:
-
no arbitrage constraints I am particularly concerned by this paper link which states arbitrage opportunities in PCA and gives examples leading to lower capital requirements when mis-matching ... although to be fair would need to recreate the analysis to see for myself that the point raised is correct.
- why do all of this ? 49. 50.
Soundbites
- why use bayesian PCA: the posterior probabilities are proportional to the prior probabilities and the likelihoods