Liminal / Structure
Fragility Monitor
Thesis
Mega-cap concentration and stretched valuation are outweighing calm credit and vol pricing
The composite fragility score blends seven independently z-scored reads, each weighted and none forecasting a drawdown on its own: index concentration (mega-cap share of total market movement), average pairwise correlation across a cross-sector basket, the gap between implied and realized volatility, a Buffett-style market-cap-to-GDP valuation ratio, bond-market signals (the 2s10s curve plus high-yield spreads), macro momentum (CPI trend and jobless-claims trend), and a breadth proxy (the share of tracked instruments above their own 50-day average). Together they measure how little absorption the market has left if a shock arrives -- not whether one is coming.
Evidence
Each of the seven components is weighted 15% except breadth at 10%, z-scored against its own trailing history (six months of daily data for the market-based components, several years of FRED history for the macro/rates components), then combined into the composite. A component whose data source fails to resolve -- most commonly FRED_API_KEY being unset, or a single series failing -- drops out of the average entirely rather than being filled with a guess, and the remaining weights are renormalized so they still sum to 100% of whatever's available.
Fragility measures shock absorption, not timing. A reading in the elevated band means a given catalyst would travel further than it would in a calmer regime, not that a catalyst is arriving; composite scores like this one have sat above the elevated threshold for multi-month stretches in prior cycles without any event following.
Each of the seven components is weighted 15% except breadth at 10%, z-scored against its own trailing history (six months of daily data for the market-based components, several years of FRED history for the macro/rates components), then combined into the composite. A component whose data source fails to resolve -- most commonly FRED_API_KEY being unset, or a single series failing -- drops out of the average entirely rather than being filled with a guess, and the remaining weights are renormalized so they still sum to 100% of whatever's available.
Market Read
Tail-hedge pricing across index options isn't unusually rich for where this score sits in its own history, which suggests the market hasn't singled out the current reading as something worth defending against yet. That gap between the score and options positioning is the closest thing this module has to a live edge.
The Gauge's Own Limits
Composite fragility gauges flatter the analyst: every input is real and independently sourced, but the weighting across seven components is a judgment call, and in backtests elevated composite readings have preceded nothing more often than they've preceded actual stress. The honest use of this score is sizing and hedging context, not prediction.
Tape
Composite · Concentration + Correlation + Vol Pricing