aeon-token-pick

At most one token pick and one prediction-market pick per run. Each pick must have:

• Falsifiable thesis — one sentence.
• Entry — price + venue.
• Kill criterion — the specific signal that says the thesis is wrong.
• Sizing — small / medium / large vs the operator's stack.
• Time horizon — hours / days / weeks.

If no candidate clears the bar — named, dated catalyst plus asymmetric setup — the skill returns NO_PICK with the top three near-misses and what would have tipped them over.

Sample output

TOKEN: $XYZ — entry $0.42 on Base (Aerodrome WETH/XYZ pool)
Thesis: ProductHunt launch May 15 + commit-velocity surge signals real shipping
Kill: launch delayed past May 22 OR daily commits drop below 5 for 3 days
Sizing: medium · Horizon: 2-3 weeks

MARKET: NO on "EIGEN unlock delayed past Q2" @ 0.31
Thesis: on-chain timelock activity confirms unlock; delay risk overpriced
Kill: official delay announcement before market resolves
Sizing: small (binary) · Resolution: July 1

No-pick output:

NO_PICK — 2026-05-12

Top near-misses:
1. $XYZ — catalyst unclear, asymmetry weak (upside +30%, downside -50%)
2. NO on "EIGEN delayed past Q2" — mispriced but no dated resolution

Would tip the call: a dated catalyst on $XYZ; a specific committee date for EIGEN.

Rules

• Falsifiable thesis or no pick.
• Cite the catalyst by name and date. "Sentiment turning" is not a catalyst.
• Crowdedness penalty — if every newsletter is recommending it, the asymmetry is gone.
• NO_PICK is a valid output. Manufactured picks burn capital.

Pairs naturally with aeon-narrative-tracker (narrative fit), aeon-monitor-runners (momentum candidates), aeon-unlock-monitor (supply-side risk), and Bankr Submit / AgenticBets for execution.