❌ [Analysis] What's wrong with cryptoasset valuation models today? – Qiao Wang

The goal of the post is to encourage people to critically evaluation valuation models, papers, and blogs before using them for investment decisions. There are five obvious types of problems with existing models:

1. Lack of empirical analysis – Many models don't have rigorous empirical evidence and proven track records (ex. NVT model)
2. Clarity of input data – Many research papers don't clarify the origin and consistency of the input data (ex. the "active addresses" data input in Metcalfe's Law + LPPLS Model)
3. Misunderstanding of model assumptions – When models for other situations are applied to new situations, the underlying assumptions can sometimes no longer work (ex. In "An Institutional Investor's Take on Cryptoassets" the PQ assumption and the MV assumption do not describe the same economy)
4. Overfitting: model complexity – Complex models tend to be overfitted and perform poorly in future datasets (ex. "Valuing cryptoasset from the ground up" uses a high-dimensional formula to predict network value)
5. Overfitting: data mining – Spurious correlations are often made because people try to overfit datasets (ex. "BTC will go to X because mining cost is Y" thesis is a spurious correlation)