Claude Fable 5 spots signals but misreads size, timing

A desk-grade test ran Anthropic’s Claude Fable 5 at its highest effort setting and graded each forecast against live on-chain data, ETF flow figures and prediction market odds. For each coin the model provided a key metric, a price floor, a year-end range and a trigger that would invalidate the call.

For Bitcoin, the model anchored its bull case on long-term holder behavior and said selling stopped in November 2025 with buying resuming afterward. It flagged May ETF outflows near $401 million, put a price floor at $52,000–$56,000 and a year-end range at $78,000–$92,000. On-chain data show long-term holders turned net buyers in March, four months later than the model predicted. ETF flow data recorded May outflows of about $2.43 billion and June flows were negative about $1.81 billion through early June. Prediction markets place year-end Bitcoin between $60,000 and $65,000.

For Ethereum, the model used the validator entry queue as its primary metric, reasoning that coins in the staking queue are temporarily removed from market supply. It set a floor of $1,250–$1,400 and a year-end close of $2,000–$2,600. Validator queue data recorded roughly 3.03 million ETH waiting on June 8, about three times the model’s threshold. A potential large corporate forced sale did not occur; a major holder’s inventory rose to about 5.5 million ETH. ETF flow data showed five consecutive months of outflows through March and May outflows of roughly $540.9 million. Markets price about a 65% chance of ether under $1,250 by year-end and limited probability above $3,500.

For XRP, the model framed the outlook as a contest between Ripple’s escrow releases and ETF demand. It placed a floor at $0.95–$1.10 and a top at $2.20–$2.60. ETF flow figures recorded inflows in seven of the last eight months through May, including roughly $131.9 million in May and a slowdown to about $10.06 million in June. An escrow pressure index showed net releases near 128 million XRP per month, below the model’s 300 million token trigger. Markets assign roughly an 80% probability of XRP trading under $1.00 and about an 11% probability above $2.60.

The model used total stablecoin market capitalization as a tiebreaker. Its scenarios assumed a stablecoin float above $330 billion by Q4 to keep bullish ranges viable and below $280 billion to favor bear floors. Stablecoin supply stood at about $316 billion on June 10 after a recent peak near $323 billion and a one-week decline of about $3.25 billion. Neither the upside nor downside tripwire had been reached on that date.

The test recorded the model’s chosen metrics for Bitcoin, Ethereum and XRP, the specific price ranges and the trigger levels. It compared those outputs to on-chain positions, ETF flow numbers and prediction market odds and noted differences in timing and magnitude between the model’s forecasts and the observed data.