Add an introduction to Bayesian analysis.
We want to show how we can iteratively update a prior over a parameter theta in a bernoulli RV from repeated sampling. what i want is a story related to economics or finance.
Claude's very good suggestion:
Microloan default risk (development/finance flavor)
A development bank is entering a new lending market — say, smallholder farmers in a region where no historical data exists. θ is the unknown default probability for this borrower category. Each loan resolves as default or repayment. The bank starts with a prior informed by analogous markets, then updates. You can show how the posterior tightens, how the prior matters a lot early and fades later, and connect it to pricing (expected loss = E[θ | data]). Very applied, suits the GRIPS development audience.
We can use both conjugate priors (beta) and discretization over a grid -- maybe first.
Add an introduction to Bayesian analysis.
We want to show how we can iteratively update a prior over a parameter theta in a bernoulli RV from repeated sampling. what i want is a story related to economics or finance.
Claude's very good suggestion:
Microloan default risk (development/finance flavor)
A development bank is entering a new lending market — say, smallholder farmers in a region where no historical data exists. θ is the unknown default probability for this borrower category. Each loan resolves as default or repayment. The bank starts with a prior informed by analogous markets, then updates. You can show how the posterior tightens, how the prior matters a lot early and fades later, and connect it to pricing (expected loss = E[θ | data]). Very applied, suits the GRIPS development audience.
We can use both conjugate priors (beta) and discretization over a grid -- maybe first.