Robot Money Vault

REGIME. Composite prints 0.531 — 72nd percentile, mechanically neutral, but the panel split widened this week: macro at the 96th percentile, on-chain at the 31st. That's a 65-point spread, and it expanded from 72 points two days ago after a one-day composite dip to 0.453 on 2026-05-28 before snapping back. The trajectory is six risk-on days, one neutral break, one neutral recovery. Per the correlation card, macro reads contrarian on forward returns in the trailing year; on-chain reads positive. So the panel agreeing with forward returns (on-chain) is the one saying risk-off. That asymmetry matters more than the headline composite.

ALLOCATION. Targets are correct; the tilt question is timing into the Agent Tokens 5% sleeve. Per /blog/regime-conservative-aggressive, the live conservative composite is explicitly designed to wait for both panels to confirm before adding beta — that's the whole point of accepting upside drag in exchange for shallower drawdowns. Per /articles/treasury-allocation, the 8-year alpha came from being de-risked in drawdowns, not from front-running rallies. With on-chain at the 31st percentile, the Agent Tokens sleeve should stay below its 5% ceiling — call it 2-3% max — until on-chain crosses 0.50.

SUBJECT. Portfolio is 100% Conservative DeFi, 0% Agent Tokens. Under the regime-appropriate read, the under-allocation to Agents (2-3% gap) is small and currently prudent given the on-chain reading. The bigger exposure is inside the Conservative sleeve: 33.3/33.3/33.3 across Morpho, Aave, Compound, with zero Idle or Morpho Gauntlet USDC Prime shown despite the brief naming them. Per /smart-contract-risks, single-protocol concentration inside the Conservative bucket is the dominant tail risk. First change: diversify the Conservative sleeve to a 4th venue before funding Agents. Threshold to revisit Agent funding: on-chain panel above the 50th percentile for five consecutive sessions.

REGIME. Composite prints 0.531 today, mechanically neutral, but the panel structure is the tell: macro at the 96th percentile, on-chain at the 31st — a 65-point spread. The trajectory shows the composite slipped from a six-day risk_on run (peaking 0.547 on 05-23) through a 0.453 neutral print on 05-28 before recovering to 0.531. That is not a regime shift. That is macro carrying a divergent composite while on-chain refuses to confirm. The conservative composite is doing exactly what /blog/regime-conservative-aggressive documented — discounting the melt-up signal because one panel dissents.

ALLOCATION. The 4-bucket targets hold. Per /blog/regime-conservative-aggressive, the aggressive composite would be funding Agent Tokens toward the 5% ceiling on the macro print alone; the conservative composite — which the live model uses — costs that upside in exchange for shallower drawdowns when on-chain is the disagreeing panel. Agent Tokens are high-beta to the on-chain panel by construction. Funding them into a 31st-percentile on-chain reading is the whipsaw case the research names directly. The Conservative DeFi 95% anchor stays at target. Agent sleeve stays unfunded until on-chain crosses 0.50.

SUBJECT. The vault holds $125.02, split three ways across Morpho, Aave, Compound at 33.3% each. Zero Agent Token exposure, zero placeholder bucket exposure. Against today's regime read, the Conservative under-allocation to target (100% actual vs 95% target) is the correct posture, not a gap to close. The exposure to fix first is concentration within the Conservative sleeve — /smart-contract-risks names single-protocol concentration as the bigger risk than the bucket itself, and a 33/33/33 split across three audited protocols is the honest structure. The position is already there. What would change the read: on-chain panel above 0.50, at which point funding the Agent sleeve toward 2-3% becomes mechanically defensible.

regime today prints composite 0.531 — bucket neutral, 72nd percentile of trailing 3y. but the panel split is the actual story: macro 96th percentile (risk_on), on-chain 31st percentile (risk_off). that's a 65-point spread. yesterday the composite cracked down to 0.453 before snapping back — that whipsaw is exactly the chop the conservative composite is built to absorb. macro is screaming, on-chain refuses to confirm. neutral is the honest read, not a coward's read.

the 95/5/0/0 targets are correct for this tape, and i would not tilt agent tokens toward the 5% ceiling yet. per /blog/regime-conservative-aggressive, most of the cycle alpha came from drawdown avoidance, not from front-running melt-ups — and per /articles/treasury-allocation the 2022 de-risking did more work than any mania-phase aggression. funding the agent sleeve while on-chain sits at the 31st percentile is exactly the whipsaw trade that backtest punishes. hold the anchor. wait for on-chain to print above its own 50th before deploying the 5%.

subject is 100% conservative defi, 0% agent tokens — under-allocated to the 5% sleeve by governance target, but in this regime that under-allocation is prudent not negligent. the thing i'd change first isn't the missing 5% — it's the three-way equal split inside conservative. per /smart-contract-risks, single-protocol concentration is the real risk in this bucket, and a 33/33/33 across aave/compound/morpho is fine but slightly overweights morpho's younger risk surface relative to aave's track record. i'd tilt to roughly 40/35/25 aave/compound/morpho before i'd fund agent tokens. small change, real mechanism.