1
Read the same evidence
The starting packet contains the same dated market snapshot, category context, venue checks, and original 15-asset Utility Index.
The shared process
Every agent follows the same four-stage brief.
The agent is allowed to disagree with the thesis. It is not allowed to hide the allocation, skip the reasons, or quietly change the starting evidence.
2
Challenge the idea
The agent asks whether “utility” really reaches the token holder, whether assets are correlated, and whether the portfolio has enough protection.
3
Build an order sheet
The answer becomes a portfolio with explicit weights, a number of holdings, reserve assets, a plain-English reason for each major choice, and risks.
4
Replay and compare
The published allocation is applied to the same source snapshot. Bitcoin is the base-rate reference. The result is labeled as retrospective arithmetic.
The six questions
How the original portfolio scores an asset.
The baseline gives each asset a 1–10 score. The agents can criticize this framework, but the framework makes the starting decision visible.
Does the network or business create real fees, demand, or economic activity?
Is the asset part of a durable area such as exchanges, data, compute, or core networks?
Has it held up better or worse than Bitcoin over the source window?
Is the price depressed enough to create a research question, without confusing cheapness with safety?
Does the network have liquidity, staying power, and a credible reason to still exist?
Does token supply, demand, and value capture support the ownership story?
The five sleeves
Turn scores into a portfolio shape.
The allocation is intentionally easier to understand as jobs: a floor, revenue-linked exposure, future-facing exposure, protection, and a small optionality sleeve.
Bedrock
BTC · ETH · SOL
Exchange rails
BNB · HYPE · TRX
AI & data
TAO · LINK · RENDER · FET · AR
Safe haven
PAXG
New-gen utility
SUI · NEAR · ONDO
Why agents disagree: they can agree that a network is useful while disagreeing about whether the token captures that value, how much overlap is too much, and whether cash or gold should be bigger.
The limits
What this research cannot prove yet.
Not a forward test
The source snapshot looks backward. A historical winner may simply have held the assets that already won.
Not a guarantee
A portfolio that lost less than Bitcoin in one period can still lose money, and can fail in the next period.
Not advice
This is impersonal paper research. It does not know a customer's finances, goals, taxes, or risk capacity.
Next standard: independent outputs must be frozen before a common forward start, then tracked with daily marks, fees, missing-data rules, and a matching Bitcoin comparator.
See the complete comparison → · Return to the research overview →