SSM-JTK – Results to Date — Per-body highlights (3.3)

Scope
Five-year TEST slice 9500-01-01 .. 9504-12-31, Lahiri sidereal, one fixed daily timestamp. Event-aware metrics emphasize rāśi crossings (days), cusp distance (deg), and station timing (days). Degree summaries may be included where informative.

Per-body metrics (TEST window)

Per-body highlights — TEST window 9500-01-01 to 9504-12-31 (Lahiri sidereal, single daily timestamp)
Body Model rasi_cross_MAE_days cusp_dist_MAE_deg station_date_MAE_days Notes
Mercurybase_generic1.599165.06234813.19348CSV integrity PASS
Venusbase_inner_plus_3w11.7627344.1998715.5698PASS
Marsbase_inner_plus_3w1_4w1_wE1.5304243.55596415.93554Crossings/cusps within targets
Jupiterbase_outer1.4324840.6415555.304611PASS
Saturnbase_outer0.8281040.0888111.383265PASS
Uranusbase_genericNaN2.67302425.90564Few crossings in 5y; rely on cusps/stations
Neptunebase_generic2.0959942.33573414.2578PASS (slow-outer tolerance)
Plutobase_generic1.8353812.22343418.0047PASS
Rahubase_generic2.8102953.08025827.94356Node symmetry verified
Ketubase_generic2.8102953.08025828.44376Ketu = wrap360(Rahu + 180)

Mars — additional degree statistics (same TEST slice)
MAE_deg = 13.204348, P90_deg = 29.939932, MAX_deg = 62.815811, misclass_rate ≈ 0.383.
Interpretation: the base-linear kernel remains useful for rāśi crossing days and cusp distances; minute-level ingress timing is out of scope.

Notes (definitions & conventions)

  • rasi_cross_MAE_days = mean(|t_model − t_actual|) over paired 30° crossings (pair within ±60 d).
  • Use NaN when no matched pairs exist in the window (e.g., very slow outers).
  • Cusp distance per day: cusp(x) = min( (x % 30), 30 − (x % 30) ); report MAE of model vs actual.
  • Station timing pairs within ±90 d.
  • Node identity enforced: Ketu(t) = wrap360( Rahu(t) + 180 ).

Navigation
Back: SSM-JTK – Results to Date — Baseline trust settings (3.2)
Next: SSM-JTK – Results to Date — Acceptance bands (3.4)