iteration 4 · 2026-06-03

Iteration 4

Angles

Raw candidates

Accepted

0 / 2

consec-dry

5-angle remaining-cluster sweep; yield tapering (13 vs iter-3 24); 4 forex-prior Haar drops; scout crypto-prior tags rejected by git-anchor (-py); no new families

Accepted this iteration (13)

#	Candidate	FOSS implementation @ license	Paper · provenance	Coverage
#96	`sign_markov_entropy_production_rate_threestate` Markov-chain entropy-production (time-irreversibility) rate of the 3-state sign transition matrix: the Kullback-Leibler divergence rate between the fo ⚠ lag-1 3-state sign Markov chain (structural alphabet) · forex-roadmap-mentioned (listed as an UNTESTED inventory item in 🔎 FOREX-MENTIONED-ONLY (roadmap inventory, not evaluated)	`github.com/networkx/networkx (graph build) + clo` BSD-3-Clause	arXiv:1201.5613 git-anchor: absent @c803ef96e + @2d308f1c	not-covered
#97	`sign_markov_slem_spectral_gap_mixing_threestate` Second-largest-eigenvalue-modulus (SLEM) / spectral-gap mixing rate of the 3-state sign transition matrix: \|lambda_2\| where lambda_2 is the eigenvalue ⚠ lag-1 sign Markov chain	`github.com/TommasoBelluzzo/PyDTMC (MarkovChain.m` MIT	DOI:10.1137/S0036144503423264 git-anchor: absent @c803ef96e + @2d308f1c	not-covered
#98	`sign_markov_inout_strength_asymmetry_flux` Net stationary-flux asymmetry of the 3-state sign transition graph: for each directed sign-pair the antisymmetric current J_ij = pi_i P_ij - pi_j P_ji ⚠ lag-1 sign Markov chain	`github.com/networkx/networkx + numpy closed-form` BSD-3-Clause	arXiv:2404.01978 git-anchor: absent @c803ef96e + @2d308f1c	not-covered
#99	`groeneveld_meeden_b3_skewness` Groeneveld-Meeden global skewness b3 = (mean - median)/E\|X - median\| = (mean-median)/MAD-from-median. The mean-median location gap normalized by the m	scipy/scipy (mean, median, median_abs_deviation primitives) BSD-3-Clause	DOI:10.2307/2987742 git-anchor: absent @c803ef96e + @2d308f1c	not-covered
#100	`hl_pseudomedian_minus_median_gap` Location-shape gap between the Hodges-Lehmann pseudomedian (median of all Walsh pairwise averages (xi+xj)/2) and the ordinary median, normalized by a	scipy/scipy (median primitive) + numpy Walsh-average constru BSD-3-Clause	DOI:10.1214/aoms/1177704172 git-anchor: absent @c803ef96e + @2d308f1c	not-covered
#101	`bickel_halfsample_mode_skewness` Bickel modal skewness = (mean - mode)/scale, where the mode is the robust Half-Sample Mode (HSM): recursively keep the shortest half-interval containi ⚠ half-sample-mode structural (dense ticks needed)	cran/modeest (paulponcet/modeest mirror) hsm() + skewness GPL-3	DOI:10.1016/S0167-9473(01)0005 git-anchor: absent @c803ef96e + @2d308f1c	not-covered
#102	`left_medcouple_lower_tailweight` Brys-Hubert-Struyf Left Medcouple (LMC): the medcouple skewness kernel evaluated on the lower structural half {x <= median}, i.e. LMC = -mc(x[x<=media	cran/robustbase lmc() GPL-2.0-or-later	DOI:10.1016/j.csda.2004.09.012 git-anchor: absent @c803ef96e + @2d308f1c	not-covered
#103	`right_medcouple_upper_tailweight` Brys-Hubert-Struyf Right Medcouple (RMC): medcouple kernel on the upper structural half {x >= median}, RMC = mc(x[x>=median]). Robust per-tail kurtosi	cran/robustbase rmc() GPL-2.0-or-later	DOI:10.1016/j.csda.2004.09.012 git-anchor: absent @c803ef96e + @2d308f1c	not-covered
#104	`l_correlation_serfling_xiao_bidpath_askpath` Serfling-Xiao L-correlation: the 2nd-order L-comoment ratio ρ_L(X\|Y)=λ2[12]/λ2[11], built from the concomitants of one intra-bar series (bid mid-price	cran.r-project.org/package=lmomco (Lcomoment.matrix + Lcomoment.correlation, Willi GPL-2.0-or-later	DOI:10.1016/j.jmva.2007.01.008 git-anchor: absent @c803ef96e + @2d308f1c	not-covered
#105	`l_comoment_directional_asymmetry_bidpath_askpath` The Serfling-Xiao directional gap Δ = λ2[12] - λ2[21]: the difference between the 2nd L-comoment of bid-path-given-ask-path and ask-path-given-bid-pat	cran.r-project.org/package=lmomco (Lcomoment.matrix off-diagonal entries [12] vs [ GPL-2.0-or-later	DOI:10.1016/j.jmva.2007.01.008 git-anchor: absent @c803ef96e + @2d308f1c	not-covered
#106	`symmetric_gini_covariance_jointrank_bidpath_askpath` Sang-Dang-Sang symmetric Gini covariance/correlation: a rank-based bivariate dependence built from the JOINT rank function (spatial-rank) rather than	xdang/SymmetricGini (Ole Miss Xin Dang reference code; also reimplem GPL-3.0-or-later	DOI:10.1002/cjs.11287 git-anchor: absent @c803ef96e + @2d308f1c	not-covered
#107	`etc_effort_to_compress_logret_sign` Effort-To-Compress complexity of the per-bar log-return sign string {+,0,-}: run Non-Sequential Recursive Pair Substitution (NSRPS) — at each step rep	pranaysy/ETCPy Apache-2.0	DOI:10.1140/epjst/e2013-01888- git-anchor: absent @c803ef96e + @2d308f1c	not-covered
#108	`sequitur_grammar_compression_ratio_logret_sign` Run the Sequitur (Nevill-Manning–Witten) online grammar-inference algorithm on the log-return sign string. Sequitur is fully deterministic — it enforc	markomanninen/pysequitur MIT	arXiv:cs/9709102 (Nevill-Manning & Witte git-anchor: absent @c803ef96e + @2d308f1c	not-covered

Dropped this iteration (38)

Candidate	Drop reason
`sign_transition_graph_reciprocity_garlaschelli`	gate: not-parameterless
`sign_transition_graph_global_efficiency_latora`	gate: not-parameterless
`sign_markov_stationary_distribution_entropy`	gate: not-parameterless
`sign_markov_entropy_rate_threestate`	gate: not-parameterless
`sign_markov_edgeweight_entropy_transition_matrix`	gate: not-parameterless
`sign_transition_graph_degree_assortativity`	gate: not-parameterless
`pearson2_mode_skewness_iqr_variant`	gate: not-computable(PARTIAL)
`crow_siddiqui_quantile_kurtosis`	gate: not-parameterless
`yule_kendall_index`	gate: forex-dup:bowley_quantile_skewness (taken) —
`median_quantile_cross_covariance_bidpath_askpath`	gate: forex-dup:blomqvist_beta_bidpath_askpath (me
`expectile_codependence_tau_half_signedreturn_spread`	gate: not-parameterless
`lower_vs_upper_gini_md_asymmetry_signedreturn_spread`	gate: forex-dup:gini_mean_difference + gini_correl
`countermonotonicity_index_lower_frechet_bidpath_askpath`	gate: forex-dup:koch_deschepper_comonotonicity_coe
`quantile_cross_covariance_nonmedian_level`	gate: not-parameterless
`haar_l1_detail_l_skewness_tau3_shape`	gate: forex-dup:l_skewness_tau3 (on raw series; th
`allan_variance_single_natural_lag_ratio`	gate: forex-dup:vonneumann_ratio (dedup — avar at
`haar_l1_detail_l_kurtosis_tau4_shape`	gate: forex-dup:l_kurtosis_tau4 (raw-series; fines
`haar_l1_approx_detail_blomqvist_sign_beta`	gate: forex-dup:blomqvist_beta_bidpath_askpath (de
`multiscale_haar_detail_energy_partition_entropy`	gate: not-parameterless
`wavelet_leader_multifractal_singularity_width`	gate: not-parameterless
`wavelet_scattering_first_order_mean`	gate: not-parameterless
`modwt_detail_variance_ratio_adjacent_scales`	gate: not-parameterless
`rle_compression_ratio_logret_sign`	gate: forex-dup:tickrule_run_length_entropy / trad
`t_complexity_titchener_logret_sign`	gate: not-computable(PARTIAL)
`kaspar_schuster_complexity_signs`	gate: forex-dup:lziv_complexity_signs (kaspar-schu
`lzw_dictionary_growth_rate_signs`	gate: not-parameterless
`sign_markov_kemeny_constant_threestate`	internal-dedup: eigenvalue-based mixing ≈ slem_spectral_gap (kept)
`sign_markov_transition_determinant_predictability`	internal-dedup: det = eigenvalue product ≈ slem mixing
`hotelling_solomons_skewness`	internal-dedup: (mean-median)/sd — same numerator as groeneveld_meeden_b3 (kept); ship one normalizer
`l_coskewness_serfling_xiao_signedreturn_spread`	deferred: higher-order L-comoment; core L-correlation kept (re-proposable)
`l_cokurtosis_serfling_xiao_midreturn_duration`	deferred: higher-order L-comoment; core L-correlation kept (re-proposable)
`spearman_footrule_psi_bidpath_askpath`	dup-class: footrule already dropped iter-2/iter-3; dominated by Kendall/L-correlation
`koch_deschepper_comonotonicity_coefficient_bidpath_askpath`	internal-dedup: comonotonicity ≈ monotone-dependence already dense (Kendall/Schweizer-Wolff)
`gzip_self_compression_ratio_logret_sign`	soft-knob: DEFLATE level/window are library defaults (inert but analyst-pinnable) — fails strict iter-4 rubric; ETC+Sequitur kept
`haar_l1_approx_detail_sign_concordance`	forex-prior: wavelet research already conducted on forex (schema.sql:333 wavelet_low_freq_bias_bar; scales 1-5 rejected \|r\|≥0.92 vs RV)
`haar_l1_detail_sign_entropy_finest`	forex-prior: wavelet family already evaluated on forex (schema.sql:333)
`haar_l1_detail_zero_crossing_rate_finest`	forex-prior: wavelet family already evaluated on forex (schema.sql:333)
`haar_l1_detail_classical_skewness_shape`	forex-prior: wavelet family already evaluated on forex (schema.sql:333)