Protocol

Benchmarks

Alpha here is LP-and-harvest alpha, not token alpha. The book is scored against a holder who never trades and against a trader who is always 50/50.

Three books, one start

All three start with the same dollars in the same two tokens at t = 0.

HODL

Buy the opening mix (usually 50/50). Do not trade. Value is just the two spots. This is the inventory you would have if you never touched an AMM.

Continuous rebalancer

Always restore 50/50. Every move pays bid–ask, impact, and gas. This is the “keep delta-neutral inventory” book without earning LP fees.

The TRUSS LP book is the third path: in-range it is a concentrated LP; out of range it sits in 50/50 (today) and waits for w* to exist again.

Figure 4. Ranking is versus HODL and versus a continuous 50/50 rebalancer. The desk is not forecasting which token goes up.

What “alpha” means on this desk

α_HODL = V_LP − V_HODL, α_reb = V_LP − V_reb

Neither number is a forecast of which token appreciates. If token A doubles and token B does nothing, HODL wins on the A-heavy residual; a 50/50 rebalancer sold A all the way up. The LP book did something in between: it sold A into the range (fee income, less A) and, in high vol, harvested more of that two-way flow. The simulation asks which of those three inventory rules won, not whether A was a good buy.

We are not in the business of calling token direction. If you want directional exposure, HODL. If you want to be paid for two-sided flow without pretending you know the next print, use the LP book when Π(w) > 0.

Volatility harvesting

A symmetric concentrated position is mechanically short the outperforming asset and long the laggard. As the mid walks up, the vault sells the winner and buys the loser so that, while in range, inventory is pulled back toward 50/50. That is the same identity as a CFMM’s gamma: you are paid (fees, emissions) to warehouse that flow.

Figure 5. Inside a symmetric range the position is always selling the stronger asset and buying the weaker one. That is the volatility-harvest, not a directional bet.

When volatility is high enough that sitting in the pool fails the Π test, the desk currently swaps residual inventory toward 50/50 and waits. That is the harvest in its strongest form: you have already sold strength, and you stop paying gamma once it is too expensive.

Future: vs-HODL sit-out

The next sit-out rule does not recentre on 50/50. It recentres on the mix a pure holder would still have — the vs-HODL weights. High-vol flow is then closed without forcing a sale of the winner back to half the book.

Figure 8. Today a high-vol sit-out recentres on 50/50. The vs-HODL path will instead match the mix a never-traded holder would still have.

That path will show up as a fourth line on the simulation when it ships. Until then, treat today’s sit-out as “harvest and wait,” not as a directional overlay.