Prospective scoring without post-hoc frame selection
The empirical protocol required to compare tropical, Lahiri-sidereal, combined, and ordinary baselines without letting outcomes choose the winning system.
Frame comparison · released 2026-08-18
Shared fact substrate
Every model receives the same eligible tasks, outcome definitions, information cutoff, scoring rule, blinded evaluation schedule, and prospectively fixed stopping point.
Required shared inputs
- Locked target and outcome horizon
- Frozen tropical model
- Frozen sidereal model
- Ordinary or random baseline
- Analysis and multiplicity policy
Tropical view
Tropical predictions are issued and persisted under a fixed model version before the outcome becomes available, including abstentions and confidence where allowed.
Lahiri-sidereal view
Sidereal predictions follow the identical submission contract while retaining their own feature namespace, rules, training history, and abstentions.
Where they agree
When both models issue the same verdict, it is recorded as agreement but remains one paired task outcome—not two independent observations.
Where they disagree
Opposed verdicts, differing confidence, and one-model abstention remain explicit. The system does not resolve them narratively after the result.
Preservation policy
Score each model separately, report paired differences against the same baseline, and evaluate any ensemble only if its weights and conflict policy were frozen in advance.
Prohibited synthesis
Do not report the better retrospective score as the prospective policy, drop discordant tasks, tune weights on the test set, or call agreement proof of truth.
What empirical comparison requires
Treat tropical and sidereal outputs as separate candidate models. Any performance claim requires preregistered features, outcomes, horizons, baselines, model-selection policy, multiplicity control, and prospective scoring that retains misses and abstentions.
Method sources
These sources define calculation and namespace policy. They do not determine which astrology system is true.
- Tropical versus sidereal astrology
Maha Celestial
The policy that tropical and sidereal charts remain parallel declared models with separate feature and rule namespaces.
Boundary: The guide is a methodology declaration, not an adjudication between traditions.
- Corporate and mundane methodology library
Maha Celestial
Evidence-bound organization events, uncertainty handling, frame comparison, preregistration, and outcome refusals.
Boundary: Corporate chart methodology cannot establish valuation, revenue, survival, investment return, or guaranteed outcomes.