How much organised structure sits in the recent price path β between pure order and pure noise (ordpy). Β· row #16 Β· ordpy Β· rolling-200 on close Β· LΓ³pez-Ruiz 1995 / Rosso 2007; embedding m=3/Ο=1 = PR#513
BAN
Verdict (worst valid cell)
0.97
Max βΟβ vs panel
0.29
Max RΒ²
1.00
Min h_norm
210
Valid cells
140 / 65 / 5
PASS / WATCH / BAN
10 / 10
Slices done
fast
Cost tier
How this verdict was reached
5 cell(s) crossed a BAN line β βΟβ>0.95 (a near-duplicate of an existing feature), RΒ²>0.95 (rebuildable from the panel), or h_norm<0.05 (went nearly constant). Worst: βΟβ 0.969 vs bar_petrosian_fd. The whole-candidate verdict is the worst cell, so the candidate is BAN.
What this feature measures
In plain terms: How structurally complex the recent price path is. Not just how random it is (entropy), but how much genuine organised structure sits between two boring extremes: a dead-flat line and pure coin-flip noise.
Analogy: A flat line scores ~0 (no structure). Pure random noise also scores ~0 (no structure, just chaos). A richly patterned signal in between scores high.
High vs low: HIGH = the path carries organised, non-trivial structure. LOW = either dead-flat or essentially pure noise.
Tunable knobs (Axis-2): Only the ordinal embedding (m=3, Ο=1) β the already-ratified PR#513 project-wide constant. No tunable magic numbers.
How we test it
We slide a 200-bar window across real ClickHouse bars and compute this feature at every bar, for each of 21 (symbol Γ threshold) cells across 6 crypto symbols, repeated over 10 historical 2-month windows ("slices") spanning 2021β2026 β 210 independent tests.
In each test we ask one question only: does this feature carry information the live production panel does not already have? (This is Axis-1 orthogonality β no profit/signal claim is made here.)
Three checks per cell: rank-correlation vs the single closest feature (βΟβ), joint reconstructability from the whole panel (RΒ²), and whether the feature actually varies (h_norm).
Read-only ClickHouse Β· against the live panel + the landed Petrosian/Katz fractal features.
What each number means
Max βΟβ (rank correlation): how strongly the feature tracks the single most-similar existing feature, in its worst cell. 0 = unrelated, 1 = a rename. Lower is better. β€0.85 good Β· 0.85β0.95 watch Β· >0.95 banned (duplicate).
Max RΒ² (joint redundancy): how much of the feature can be rebuilt from all existing features together. 0 = fully new, 1 = fully reconstructable. Lower is better. Same thresholds.
Min h_norm (liveness): how much the feature actually moves (normalised entropy of its own values). Near 0 = basically constant / collapsed = useless. Want β₯0.05.
Worst-valid-cell rollup: the candidate inherits the verdict of its worst of 210 cells β deliberately strict, so a feature only PASSES if it is orthogonal everywhere.
Per-slice (regime) orthogonality
Each row is one 2-month historical window. "Worst βΟβ" is the most-correlated existing feature in that regime's worst cell β lower is better.
Slice
Valid
P / W / B
Worst βΟβ
Top competitor
slice012026-01 β 02
21
17 / 4 / 0
0.89
bar_petrosian_fd
slice032025-07 β 08
21
17 / 4 / 0
0.92
bar_petrosian_fd
slice052025-01 β 02
21
14 / 7 / 0
0.92
bar_petrosian_fd
slice072024-07 β 08
21
13 / 8 / 0
0.94
bar_petrosian_fd
slice092024-01 β 02
21
13 / 6 / 2
0.97
bar_petrosian_fd
slice122023-04 β 05
21
11 / 8 / 2
0.97
bar_petrosian_fd
slice142022-10 β 11
21
11 / 10 / 0
0.92
bar_petrosian_fd
slice172022-01 β 02
21
14 / 7 / 0
0.95
bar_petrosian_fd
slice202021-04 β 05
21
15 / 5 / 1
0.96
bar_petrosian_fd
slice212021-01 β 02
21
15 / 6 / 0
0.94
bar_petrosian_fd
Most-similar existing features (the candidate's nearest neighbours)
Which already-shipped features this candidate most often resembles. If one feature dominates here and βΟβ is high, the candidate may be re-deriving it.
bar_petrosian_fd β most-similar existing feature in 198 cell(s)
bar_katz_fd β most-similar existing feature in 8 cell(s)
duration_us β most-similar existing feature in 3 cell(s)
Resource utilization
Hard cap: 2 CPU cores (CPUQuota 200%), 2 GB RAM, spare-only β the loop runs at the lowest priority (CPUWeight 10, nice 19, ionice idle) and only starts when β₯2 cores and β₯2 GB are free, so it never competes with production (ClickHouse / the live sidecar).
This sweep: wall time 39m 07s Β· peak RAM 1912 MB of 2048 MB.
Data volume: 6,333,627 bar-rows processed across 210 cells. 28 of 210 cells were sub-sampled to the most-recent 100,000 bars (largest raw cell β 1,058,962 bars). Sub-sampling keeps giant cells tractable and does not change the correlation estimate.
Safety: the sweep completing means it stayed within 2 GB β had it exceeded, the cgroup would have killed only the loop, never the box, and the cap-worked event would be recorded (never raised).
Worth flagging to the operator
π΄ βΟβ 0.969 crosses 0.85 in β₯1 cell (closest existing feature: bar_petrosian_fd) β that cell is BAN (duplicate).
βΉοΈ 28 cell(s) sub-sampled to the 100,000-bar cap (estimate unaffected).