Panel vs timeseries

Answers

What Mode.PANEL vs Mode.TIMESERIES mean, when each is dispatched, and the sample-guard contract for each. For the conventions table (column names, alignment), see Timeseries-mode conventions. For the mode= parameter on evaluate(), see evaluate. For sample-guard error surfacing (InsufficientSampleError, ModeAxisError), see Errors.

Sample guards¶

Time-series length n_periods and asset count n_assets are gated independently — factrix does not use a combined n_periods × n_assets observation count, because per-date statistic variance is driven primarily by n_assets, while time-series aggregation power is driven by n_periods.

Two-axis guard structure¶

Axis	Hard block	Soft warning	Clean
`n_periods` (T)	T < 20 → `InsufficientSampleError`	20 ≤ T < 30 → `UNRELIABLE_SE_SHORT_PERIODS`	T ≥ 30
`n_assets` (N)	none	N < 10 → `SMALL_CROSS_SECTION_N`; 10 ≤ N < 30 → `BORDERLINE_CROSS_SECTION_N`	N ≥ 30

n_assets is never hard-blocked because the cross-asset t-test on E[β] is mathematically well-defined for N ≥ 2 — only its statistical power degrades. A hard block would force users to choose between "can't run" and "don't know there's a problem"; the warning provides the result while surfacing the issue.

Behaviour matrix by factory and N¶

Factory	N=1	N=2..9	N=10..29	N≥30
`individual_continuous(IC)`	raises `ModeAxisError`	all dates dropped (MIN_ASSETS_PER_DATE_IC=10) → NaN	normal PANEL	normal PANEL
`individual_continuous(FM)`	raises `ModeAxisError`	per-date guard; low df	normal PANEL	normal PANEL
`common_continuous`	TIMESERIES single-series β	emits `SMALL_CROSS_SECTION_N`	emits `BORDERLINE_CROSS_SECTION_N`	normal PANEL
`individual_sparse` / `common_sparse`	`_SCOPE_COLLAPSED` → TS dummy; `SCOPE_AXIS_COLLAPSED` info	normal PANEL CAAR	normal PANEL CAAR	normal PANEL CAAR

Aggregation order¶

PANEL procedures split into cross-section first (cs-first — individual_continuous information coefficient (IC) / FM, sparse CAAR) and time-series first (ts-first — common_continuous, common_sparse). The order determines small-sample failure modes and the N=1 collapse behaviour: common_continuous at N=1 degenerates to a single-series β test (null: β=0, not E[β]=0), which is still well-defined; individual_continuous at N=1 has no cross-section to aggregate over, so it raises. The cs-first / ts-first / ts-only / static-cs / per-event shorthand is the canonical vocabulary used across the metric matrix and list_metrics() output — see Reference § Metric pipelines § Aggregation vocabulary.

Full per-procedure pseudocode for all 7 registered pipelines lives in Development § Procedure pipelines.

Introspection¶

# heuristic config suggestion + risk warnings
result = fx.suggest_config(panel)
print(result.reasoning, result.warnings)

# list all cells and their MIN_PERIODS thresholds
print(fx.describe_analysis_modes())