Curriculum·S205 On-Chain and Fundamental Data·about 30 min
Building a thesis from data
By the end of this lesson you can
- →Write a thesis as a claim with named metrics that would falsify it
- →Distinguish a thesis from a description, and both from a position
- →Compute the point at which a published mechanism fails, from published inputs
- →Build a dashboard whose purpose is to disprove your own thesis rather than to display it
Sophomore · enrolled learners
This lesson opens with Anchor's yield, computable from published figures.
- What happened
- TerraUSD reached about $17.5B in supply, with roughly 75 percent held in Anchor, a protocol paying around 19.5 percent on deposits. The yield was not earned from borrower demand; it was subsidised from a reserve. In February 2022 that reserve was topped up by $450M to a total of about $507M, a level the Luna Foundation Guard assessed as sufficient to maintain the rate for approximately a year. By April the subsidy was reported to be depleting at around $6M a day. Every one of those figures was published: the rate, the deposit base, the reserve balance and the burn rate. The peg broke on 7 May 2022, and the defense mechanism took LUNA's supply from 343 million to 6.53 trillion within a week.
- The decision point
- The thesis that Anchor was sustainable had a falsifier that was public, quantitative and updated continuously: the reserve balance divided by the burn rate. That quotient was about 85 days against a stated expectation of a year. Nobody needed private information, a model, or a forecast about the market. They needed to divide one published number by another and to have decided in advance that the answer would change their position.
- Recorded loss
- $40,000,000,000
What you will be able to answer
- →What distinguishes a thesis from a description?
- →What makes a falsifier usable?
- →What was Anchor's public falsifier?
- →What is a dashboard for?
Orientation and Year One are open: anyone can read them without an account. From Year Two onward the lessons are for enrolled learners, because progress through the later years only means anything if it is tracked against a record.
It is free. We do not sell the list and there is nothing to buy at the end of it.
Sources and review
- https://cointelegraph.com/news/terra-injects-450m-ust-into-anchor-reserve-days-before-protocol-depletion
- https://corpgov.law.harvard.edu/2023/05/22/anatomy-of-a-run-the-terra-luna-crash/
- https://docs.llama.fi/
- https://krebsonsecurity.com/2025/03/feds-link-150m-cyberheist-to-2022-lastpass-hacks/
Confidence medium·Volatility high·Reviewed 2026-08-06·Owner unassigned
Contested
F102-04 and the F102 course autopsy use Terra to teach stablecoin models and subsidised yield. This lesson uses it as a worked example of a falsifiable thesis with public inputs, and deliberately does not re-derive the stablecoin taxonomy. Keep the split.
The burn rate and reserve figures are from contemporaneous reporting and moved over the period. The arithmetic is presented to demonstrate that a falsifier existed and was computable, not to claim any specific person should have predicted the date.