Week 6 Problem Session: Bayesian updating workshop

Rebuilding the flagship Bayes update from scratch, then updating on a second signal to see the posterior move again

29 March 2027

Before the session#

Read the lecture page. Bring the 60% prior, the 0.8 / 0.5 bet-sizing likelihoods, and the ≈70.6% posterior — the session rebuilds all three numbers by hand.

In the session#

The group recomputes the flagship posterior from the raw prior and likelihoods, then recomputes EV(call) from that posterior, checking that it lands near −$5.88 and flips the week 5 call into a fold.

Then a second signal is introduced (villain also checks their timing tell a fixed way depending on hand strength), and everyone updates again — using the first posterior as the new prior — to see a range narrow over two signals instead of one, which is closer to how a real hand actually plays out street by street.

Afterwards#

Nothing is marked this week. Going into week 7: comfort computing a Bayesian posterior from a prior and a likelihood pair without the lecture’s numbers already in front of you.