Family swept: jump / tail / fragility ignition detectors · campaign regime-invariant-orthogonality · append-only
The ignition axis. The prior iterations mapped who is exposed to a cascade (fragility, connectedness, crowding); this one harvests the detectors that fire when a cascade starts: discontinuous price jumps (BNS bipower, Huang–Tauchen, Lee–Mykland, Aït-Sahalia–Jacod, Jiang–Oomen), jump-robust vol baselines (Mancini threshold, MedRV/MinRV, preaveraging), signed jump variation and realized higher moments (downside/"bad" vol, negative realized skew = crash-risk), self-exciting criticality (Hawkes branching-ratio n→1, Filimonov–Sornette reflexivity/endogeneity index), fat-tail EVT (Hill, Pickands, Dekkers–Einmahl–de Haan, POT-GPD), and the bubble/crash singularity (LPPLS).
All 18 are spot-legal — none need quotes/L2 or derivatives. Two spot sub-axes: [P] from HF prices/returns (ODB bar OHLC or aggTrade mid), [T] from trade/event timestamps. Generic Hawkes intensity was already seeded (S16); the genuinely new addition here is the branching-ratio / reflexivity criticality (n→1) reading — H46/H47 cite S16 as parent.
| # | Name | Family | Measures | Cascade / ignition mapping | Spot? | Param | Stable-ID | FOSS |
|---|---|---|---|---|---|---|---|---|
| H35 | BNS bipower-variation jump test | realized-jump-test | RV−BV gap ⇒ presence & share of jump variation | jump ignition (discontinuous moves) | [P] | PP | Barndorff-Nielsen & Shephard 2004/2006 — DOI 10.1093/jjfinec/nbh001; 10.1093/jjfinec/nbi022 | VER highfrequency::BNSjumpTest (R, GPL≥2) |
| H36 | Huang–Tauchen relative-jump / z-stat | realized-jump-test | RJ=(RV−BV)/RV; ratio/log/max z-statistic | jump ignition; variance fraction from jumps | [P] | DIM | Huang & Tauchen 2005, JFEc — DOI 10.1093/jjfinec/nbi025 · parent H35 | VER highfrequency (R) |
| H37 | Lee–Mykland jump test | realized-jump-test | return / local bipower spot-vol ⇒ intraday jump times & sizes | jump ignition, per-event timing | [P] | PP | Lee & Mykland 2008, RFS — DOI 10.1093/rfs/hhm056 | VER YalDan/hf.econometrics + QuantLet/JumpDetectR (R/Py, crypto-native) |
| H38 | Aït-Sahalia–Jacod jump test | realized-jump-test | power-variation ratio at two scales → 1 (jumps); jump-activity index | jump ignition; infinite- vs finite-activity fragility | [P] | PP | Aït-Sahalia & Jacod 2009, Ann.Stat — DOI 10.1214/07-AOS568; arXiv:0903.0226 | VER highfrequency::AJjumpTest (R); hf.econometrics (Py) |
| H39 | Mancini threshold / truncated RV | jump-robust-vol | continuous-part IV by truncating returns above θ(Δ) | separates diffusion from jump; jump-size ID | [P] | PP | Mancini 2009, Scand.J.Stat — DOI 10.1111/j.1467-9469.2008.00622.x | VER highfrequency (R) threshold measures |
| H40 | MedRV / MinRV | jump-robust-vol | nearest-neighbour-truncation jump-robust integrated variance | jump-robust IV baseline for the RV−IV jump gap | [P] | PF | Andersen, Dobrev, Schaumburg 2012, J.Econometrics — DOI 10.1016/j.jeconom.2012.01.011 | VER highfrequency::medRV / minRV (R, GPL≥2) |
| H41 | Jiang–Oomen swap-variance jump test | realized-jump-test | variance-swap replication-error statistic (3rd+ return moments) | jump ignition; more powerful than BV in tails | [P] | PP | Jiang & Oomen 2008, J.Econometrics — DOI 10.1016/j.jeconom.2008.03.011 | VER highfrequency::JOjumpTest (R) |
| H42 | Preaveraging noise-robust (JLMPV) | jump-robust-vol | pre-average returns to kill microstructure noise → consistent IV / jump tests | jump ID robust to noise (dense aggTrades) | [P] | PP | Jacod, Li, Mykland, Podolskij, Vetter 2009, SPA — DOI 10.1016/j.spa.2008.11.004 | VER highfrequency (R) preaveraged cov |
| H43 | Bollerslev–Law–Tauchen co-jump test | co-jump | cross-sectional common-jump detector across a panel | systemic co-jump ignition / contagion across symbols | [P] multi-symbol | PP | Bollerslev, Law, Tauchen 2008, J.Econometrics — DOI 10.1016/j.jeconom.2008.01.006 | UNV highfrequency::rBPCov blocks; dedicated BLT UNVERIFIED |
| H44 | Realized semivariance / signed jump variation | signed-jump-semivariance | RS⁺, RS⁻; SJV = RS⁺−RS⁻ ("good"/"bad" vol) | crash precursor: bad vol & negative SJV predict higher future vol | [P] | PF | Barndorff-Nielsen, Kinnebrock, Shephard 2010; Patton & Sheppard 2015, REStat — DOI 10.1162/REST_a_00503 | VER highfrequency::rSV / rSemiCov (R, GPL≥2) |
| H45 | Realized skewness / kurtosis | realized-higher-moment | intraday-return realized skewness & kurtosis over a window | fat-tail fragility; negative realized skew = crash-risk loading | [P] | DIM | Amaya, Christoffersen, Jacobs, Vasquez 2015, JFE — DOI 10.1016/j.jfineco.2015.02.009 | VER highfrequency::rSkew / rKurt (R, GPL≥2) |
| H46 | Hawkes branching-ratio criticality (n→1) | self-exciting-criticality | branching ratio n = kernel L¹-norm; n→1 ⇒ near-critical self-excitation | reflexive feedback; one event triggers a near-diverging cascade | [T] | DIM | Hardiman, Bercot, Bouchaud 2013, EPJ B — DOI 10.1140/epjb/e2013-40107-3; arXiv:1302.1405 · parent S16 | VER X-DataInitiative/tick (Py, BSD-3); hawkesbook; hawkeslib |
| H47 | Filimonov–Sornette reflexivity index | self-exciting-criticality | endogeneity fraction = n (self-generated vs exogenous-news activity) | flash-crash precursor; rising endogeneity ⇒ approaching criticality | [T] | DIM | Filimonov & Sornette 2012, PRE 85:056108 — DOI 10.1103/PhysRevE.85.056108; arXiv:1201.3572 · parent H46 | UNV tick / hawkesbook (calibrate → n); dedicated FS pkg UNVERIFIED |
| H48 | Hill tail-index estimator | tail-index-EVT | ML tail exponent α from k upper order statistics (power-law tail) | fat-tail fragility; small α ⇒ heavy crash tail | [P] | PP | Hill 1975, Ann.Stat — DOI 10.1214/aos/1176343247 | VER evir::hill (R, GPL≥2); POT; extRemes |
| H49 | Pickands tail-index estimator | tail-index-EVT | order-statistic estimator of EVT shape ξ (all domains of attraction) | fat-tail fragility; ξ>0 ⇒ heavy tail | [P] | PP | Pickands 1975, Ann.Stat — DOI 10.1214/aos/1176343003 | VER evir::pickands (R); POT |
| H50 | Dekkers–Einmahl–de Haan moment estimator | tail-index-EVT | moment-ratio estimator of EVT index γ (Hill generalized to real γ) | fat-tail fragility (bounded & heavy tails) | [P] | PP | Dekkers, Einmahl, de Haan 1989, Ann.Stat — DOI 10.1214/aos/1176347397 | UNV evir / extRemes (R); exact fn UNVERIFIED |
| H51 | Peaks-Over-Threshold GPD fit | tail-index-EVT | exceedances over threshold u ~ Generalized Pareto (Balkema–de Haan–Pickands) | fat-tail fragility; tail-VaR / ES; ξ = crash-tail heaviness | [P] | PP | Balkema & de Haan 1974, Ann.Prob — DOI 10.1214/aop/1176996548; Pickands 1975 | VER POT (R, GPL≥2); extRemes; scipy.stats.genpareto (Py, BSD-3) |
| H52 | LPPLS log-periodic power-law singularity | bubble-crash-singularity | fits faster-than-exp growth + log-periodic oscillations → finite-time critical t_c | bubble/crash precursor; endogenous critical point t_c | [P] | PP (7 params, ill-cond.) | Johansen, Ledoit, Sornette 2000, IJTAF — DOI 10.1142/S0219024900000115; calib. Filimonov & Sornette 2013 — DOI 10.1016/j.physa.2013.04.012 | VER Boulder-Investment-Technologies/lppls (Py, MIT) |
BNSjumpTest, AJjumpTest, JOjumpTest, medRV/minRV, rSV/rSemiCov, rSkew/rKurt, spotVol. cran.r-project.org/package=highfrequency · (NB: older R realized pkg archived → absorbed here)hawkesbook, hawkeslib. github.com/X-DataInitiative/tickscipy.stats.genpareto (BSD-3). cran.r-project.org/package=POTaxis-2-cascade-free/CANDIDATE-CATALOG.md · back to loop hub