Axis-2 · Cascade-Free Harvest — Iteration 2 20 NEW (+2 OOS)

Family swept: liquidity-cascade / fire-sale / feedback-loop / systemic-fragility · campaign regime-invariant-orthogonality · append-only

20new (spot-legal/adaptable)

+2flagged out-of-scope

0dupes

48running total (spot)

The fragility axis. This iteration harvests the measures that quantify how a market becomes cascade-prone: tail-spillover / contagion (ΔCoVaR, MES, LRMES, CATFIN), connectedness networks (Diebold–Yilmaz, Baruník–Křehlík frequency spillover, Billio et al. Granger network), coupling / PCA-fragility (Absorption Ratio, RMT market-mode eigenvalue, Turbulence Index, and its 2020 nonlinear autoencoder generalization), correlation-dynamics crowding (DCC-GARCH, DECO, Correlation Surprise, CISS), herding (CSAD/CSSD), and the two liquidity-spiral proxies (commonality-in-liquidity, liquidity-adjusted VaR).

Spot-legality was the load-bearing screen. Almost all of these run on a panel of spot returns or on our existing spot illiquidity/spread estimators — no derivatives needed. Two are catalogued but out-of-scope: SRISK (needs equity/debt leverage — no clean token analogue) and the Brunnermeier–Pedersen funding/market spiral (a model whose funding leg is off-limits). Their return-only shadows — LRMES and commonality-in-liquidity — are kept in-scope. Billio et al. is spot-legal despite the word "network": its graph is estimated from returns, not from exogenous holdings.

New candidates (iter 2)

#	Name	Family	Measures	Cascade / fragility mapping	Spot?	Param	Stable-ID	FOSS
H13	ΔCoVaR	systemic-risk / tail	system VaR conditional on symbol i at its VaR, minus median state (quantile reg)	directional tail-spillover → contagion	spot	PP	Adrian & Brunnermeier 2016, AER — DOI 10.1257/aer.20120555	VER frds (Py); TimoDimi/SystemicRisk (R); SystemicR (R)
H14	MES	systemic-risk / tail	mean return of i on worst q% market days	crowding / co-crash exposure	spot	PP	Acharya, Pedersen, Philippon, Richardson 2017, RFS — DOI 10.1093/rfs/hhw088	VER frds (Py); Belluzzo (MATLAB)
H15	LRMES	systemic-risk / tail	GARCH-DCC-simulated long-run MES (SRISK return-only leg)	fire-sale / co-crash contribution, no leverage wrapper	spot	PP	Brownlees & Engle 2017, RFS — DOI 10.1093/rfs/hhw060 · parent H14	VER frds (Py) `LRMES`
H16	SRISK	systemic-risk	expected capital shortfall = f(size, leverage, LRMES)	fire-sale amplification / capital-shortfall buildup	OOS needs-fundamentals	MN	Brownlees & Engle 2017, RFS — DOI 10.1093/rfs/hhw060	frds; Belluzzo — flagged, not usable
H17	Diebold–Yilmaz connectedness	connectedness/spillover	total & directional shares of generalized H-step FEVD across a VAR	return-derived contagion network; rises in crises	spot	PP	DY 2012, IJF — DOI 10.1016/j.ijforecast.2011.02.006; 2014, J.Econometrics — DOI 10.1016/j.jeconom.2014.04.012	VER frequencyConnectedness (R, GPL-2); ConnectednessApproach (R, GPL-3)
H18	Baruník–Křehlík freq. connectedness	connectedness/spillover	spectral decomposition of DY into short/med/long-run bands	transient vs persistent (structural) contagion	spot	PP	Baruník & Křehlík 2018, J.Fin.Econometrics — DOI 10.1093/jjfinec/nby001 · parent H17	VER tomaskrehlik/frequencyConnectedness (R, GPL-2)
H19	Absorption Ratio	market-fragility / PCA	fraction of return variance absorbed by top-K covariance eigenvectors	tight coupling → fragility; ΔAR spikes precede drawdowns	spot	DIM	Kritzman, Li, Page, Rigobon 2011, JPM — DOI 10.3905/jpm.2011.37.4.112	VER frds (Py); Belluzzo (MATLAB)
H20	Turbulence Index (Mahalanobis)	market-fragility	statistical unusualness of a return vector vs historical mean/cov	fragility/stress flag; co-moves with cascades	spot	PF	Kritzman & Li 2010, FAJ — DOI 10.2469/faj.v66.n5.3	VER frds (Py); Belluzzo (MATLAB)
H21	Correlation Surprise	co-movement / correlation	turbulence split into magnitude vs correlation-conditioned surprise	unusual co-movement structure = crowding onset	spot	DIM	Kinlaw & Turkington 2013, J.Asset Mgmt — DOI 10.1057/jam.2013.27	UNV Belluzzo (MATLAB)
H22	DCC-GARCH	correlation-dynamics	time-varying conditional correlation matrix (correlation-targeting)	rising dynamic correlation = diversification breakdown / crowding	spot	PP	Engle 2002, JBES — DOI 10.1198/073500102288618487	VER alexiosg/rmgarch (R, GPL-3); frds (Py)
H23	DECO (dynamic equicorrelation)	correlation-dynamics	single time-varying average correlation across all pairs	scalar "market coupling" index; spikes = systemic co-movement	spot	PP	Engle & Kelly 2012, JBES — DOI 10.1080/07350015.2011.652048 · parent H22	VER alexiosg/rmgarch (R, GPL-3)
H24	RMT largest-eigenvalue / market-mode	market-fragility / PCA	empirical correlation eigenvalues vs Marchenko–Pastur null; λ_max = collective mode	deviation from RMT null = genuine coupling → fragility; λ_max grows pre-crisis	spot	PF	Laloux, Cizeau, Bouchaud, Potters 1999, PRL 83:1467 — DOI 10.1103/PhysRevLett.83.1467	UNV numpy / pyRMT
H25	Billio–Getmansky–Lo–Pelizzon	connectedness / fragility	PCA cumulative-variance + Granger-causality network from returns	return-based causal-network density + PCA absorption rise pre-crisis	spot	PP	Billio, Getmansky, Lo, Pelizzon 2012, JFE — DOI 10.1016/j.jfineco.2011.12.010	UNV SystemicR (R); Belluzzo (MATLAB)
H26	CATFIN	systemic-risk / tail	cross-sectional catastrophic VaR (avg of GPD/GEV/non-parametric tail VaR)	aggregate cross-section tail-risk/fragility; leads downturns	spot	PP	Allen, Bali, Tang 2012, RFS — DOI 10.1093/rfs/hhs094	UNV Belluzzo (MATLAB)
H27	Critical Slowing Down / early-warning signals	market-fragility (tipping-point)	rising rolling variance + lag-1 autocorrelation (+skew) near a tipping point	fragility / loss-of-resilience before regime shift or crash	spot	PP	Scheffer et al. 2009, Nature — DOI 10.1038/nature08227; Dakos et al. 2012, PLoS ONE — DOI 10.1371/journal.pone.0041010	VER earlywarnings (R, BSD-2); EWSmethods (R)
H28	Commonality in liquidity	liquidity-spiral (spot shadow)	co-movement of individual illiquidity with market-wide illiquidity	spot-legal shadow of the fire-sale illiquidity spiral; systemic-liquidity risk	spot	PP	Chordia, Roll, Subrahmanyam 2000, JFE — DOI 10.1016/S0304-405X(00)00057-3; Karolyi, Lee, van Dijk 2012, JFE — DOI 10.1016/j.jfineco.2011.12.008	UNV panel regression
H29	Liquidity-adjusted VaR (LVaR)	liquidity-spiral	VaR augmented by the bid-ask-spread distribution	illiquidity add-on; spread blows out in fire-sales → tail widens	spot	PP	Bangia, Diebold, Schuermann, Stroughair 1999, Wharton FIC WP 99-06 (DOI 10.1007/978-1-4615-0791-8_1)	UNV
H30	Brunnermeier–Pedersen liquidity spiral	liquidity-spiral (model)	margin-spiral × loss-spiral mutual reinforcement (theory)	canonical fire-sale / illiquidity-spiral mechanism	OOS needs funding/margin	model	Brunnermeier & Pedersen 2009, RFS — DOI 10.1093/rfs/hhn098	NONE (model) — spot shadow = H28
H31	Dynamic conditional tail dependence	tail-dependence (crypto SOTA)	time-varying lower/upper tail-dependence across the crypto cross-section	"crashing together": persistent systemic core; asymmetric tail connectedness	spot	PP	"Crashing Together, Rallying Apart" — arXiv:2606.16840 (2026)	UNV
H32	Autoencoder Reconstruction Ratio	market-fragility / PCA (SOTA)	nonlinear generalization of the Absorption Ratio via autoencoder reconstruction error	co-movement/coupling shifts a linear PCA absorption ratio misses	spot	PP	"Detecting Changes in Asset Co-Movement…" — arXiv:2002.02008 · parent H19	UNV
H33	Herding — CSAD / CSSD	co-movement / crowding	cross-sectional dispersion of returns vs the market return	dispersion collapses when the crowd herds → directional-crowding flag	spot	PF	Chang, Cheng, Khorana 2000, JBF — DOI 10.1016/S0378-4266(99)00096-5; Christie & Huang 1995, FAJ — DOI 10.2469/faj.v51.n4.1918	UNV OLS
H34	CISS (Composite Indicator of Systemic Stress)	systemic-risk / correlation	portfolio-theoretic aggregation of segment stress weighted by time-varying cross-correlation	weights simultaneous cross-segment stress = systemic; correlation-aware fragility	spot adaptable	PP	Holló, Kremer, Lo Duca 2012, ECB WP 1426 (SSRN 2018792)	UNV method only

FOSS anchors verified this iteration

tomaskrehlik/frequencyConnectedness — R, GPL-2, CRAN — Diebold–Yilmaz + Baruník–Křehlík frequency connectedness. github.com/tomaskrehlik/frequencyConnectedness
GabauerDavid/ConnectednessApproach — R, GPL-3, CRAN — DY/BK + TVP-VAR/QVAR superset, joint & extended-joint connectedness. github.com/GabauerDavid/ConnectednessApproach
alexiosg/rmgarch — R, GPL-3, CRAN — DCC / DECO / GO-GARCH / Copula-GARCH engine. github.com/alexiosg/rmgarch
earlywarnings (Dakos/Lahti) — R, BSD-2, CRAN — critical-slowing-down / generic early-warning signals. Sibling: EWSmethods. cran.r-project.org/package=earlywarnings
bashtage/arch — Python, NCSA — univariate GARCH/EGARCH/HAR backbone for MES/CoVaR/LRMES (NOT DCC — use rmgarch/frds for correlation). github.com/bashtage/arch
mgao6767/frds — Python — AbsorptionRatio / Turbulence / MES / LRMES / CoVaR / GARCH-DCC (LICENSE file present, exact string to re-verify). github.com/mgao6767/frds

Scope & provenance notes

Out-of-scope (principle #8): SRISK (H16, needs leverage/balance-sheet) and Brunnermeier–Pedersen spiral (H30, needs funding/margin). Return-only shadows LRMES (H15) & commonality-in-liquidity (H28) kept in-scope.
Billio et al. (H25) is spot-legal — its Granger network is estimated from returns, not from exogenous holdings/exposures.
Higher-moment DY/BK spillovers (co-skewness / co-kurtosis, 2024-25 crypto lit) run through the same ConnectednessApproach machinery — a family extension, not a separate row.
Provenance caveat: DOIs for H21/H22/H26/H33 and every UNVERIFIED FOSS were citation-confident but not byte-re-fetched this session (CRAN/GitHub direct fetch rate-blocked) — re-verify before promotion.
No numbers claimed — literature + FOSS harvest only; nothing evaluated on our data.

Next angle → jump / tail / fragility ignition detectors (realized jumps, bipower/threshold jump tests, Hawkes branching-ratio criticality, tail-index/Hill, self-exciting cascade ignition). · Data SSoT: axis-2-cascade-free/CANDIDATE-CATALOG.md · back to loop hub