Reflexivity in protocol earnings

Pitfalls in using P/E analysis to value protocols

[post edited/updated on 2/21]

There are several big differences between protocol earnings per token, or EPT (fully diluted protocol marketcap divided by annualized protocol revenue for that particular day) and the EPS of a public company.

One is that annualizing a daily EPT is extremely subjective. You’re holding factors constant (velocity of money, number of active wallets) which never stay constant, to a degree unheard-of in traditional companies.

Another big difference is that protocols tend to look “cheapest” — or at least cheapish — closest to the top. This was obvious in hindsight, but didn’t dawn on me until pretty recently. Below is the TokenTerminal chart of Ethereum’s annualized current P/E (daily earnings accruing to ETH holders * 365), going back 180 days, relative to its market cap.

Note: TokenTerminal P/E’s have some issues, like not being able to measure certain types of revenues accruing to tokenholders (such as token buybacks/’burning’).

Eth peaked at around 4800 on Nov. 30th. During its 25% fall (4800 → 3700) in December, its annualized daily P/E grew from ~24x to over 35x. In January, otoh (3700 → 2750), ETH’s multiple compressed from 35x to a low of just under 20x, and perfectly coincided with the bottom. This price-to-earnings reflexivity ( (35/24)/.75 , or 65%) is pretty unique to crypto, but it makes sense when you think about it.

Protocols make money from transaction taxes (gas fees). Gas fees are a function of the demand for network computing power: it’s the velocity of money on the network. As velocity plunges from a frenzied high to a local low, the “earnings multiple” of the currency goes up as the value of the currency goes down.

Eventually, after low gas fees spur increased network activity, the velocity of the network finds a natural bottom. At this stage, any further multiple compression should become positively correlated with annualized present earnings. Only at this stage, IMO, is it safe to consider going long.

A cryptocurrency bottoms when degens are gone, and whales who don’t sell are the only ones left. Whales, by definition, have very low velocity of money. Therefore, the aggregate velocity of the network should be lowest at a local low.

Similarly, velocity of a network should peak at the beginning of the end of the bubble. Over the course of a cryptocurrency bubble, the % share held by whales will go down, and a rising proportion is held by fast-trading, low-conviction degens. Then, when the on-network velocity and leverage have veered to speculative excess, a contraction happens that’s much faster than the rise up. This creates a massive flush of liquidations, a trading frenzy as traders’ risk-management goes into overdrive, and a peak in protocol revenues as the protocol needs to flush a crescendo of transactions in a short amount of time.

So are P/E’s garbage when looking at protocols?

Nah, I don’t think so.

Protocols that get cheaper during selloffs are clearly more interesting than protocols that get more expensive during selloffs, because they imply that velocity is entering an over-correction stage.