Skip to main content
ORVANTHIS
Sign inStart free trial
METHODOLOGY

A transparent path from discovery to follow-up.

Orvanthis looks broadly, prioritizes carefully, keeps validation gaps visible, and preserves the reasoning for the next review.

RESEARCH INTELLIGENCEMETHODOLOGY
01
RESEARCH STAGE

Broad discovery

Review multiple discovery paths without treating any single feed as complete.

  • Market movers and fast-developing behavior
  • Actively traded equities
  • Press releases, earnings, and source-backed events
  • New, returned, pinned, and active setups
02
RESEARCH STAGE

Eligibility and normalization

Make unlike inputs comparable before ranking.

  • Resolve ticker and company identity
  • Separate provider event time from fetch time
  • Distinguish forming from closed candles
  • Keep unavailable evidence explicitly missing
03
RESEARCH STAGE

Deterministic prioritization

Eligible setups enter a ranked research queue based on bounded, versioned rules.

  • Current evidence and participation
  • Freshness and source quality
  • Risk and weakening conditions
  • Outstanding validation needs
04
RESEARCH STAGE

Independent validation

The user reviews what supports the setup and what could disprove it.

  • Open original and authoritative sources
  • Confirm timestamps, units, and reporting periods
  • Review conflicting evidence
  • Write the next research question
05
RESEARCH STAGE

Durable follow-up

Carry the reasoning into the next session instead of rebuilding it from memory.

  • Save the thesis and evidence
  • Pin active research
  • Monitor defined conditions
  • Generate a structured brief
06
RESEARCH STAGE

Bounded historical context

Completed tracked outcomes may add sample-aware context while current evidence stays primary.

  • Disclose sample size
  • Avoid causal or return claims
  • Version outcome definitions
  • Do not let an AI model rewrite detector rules
07
RESEARCH STAGE

Customer-facing performance record

Relative research performance appears only after an immutable release record and matched benchmark observations satisfy the customer-display gate.

  • Keep every eligible release in the record; do not retroactively select winners or remove losses
  • Measure Orvanthis research and SPY, QQQ, DIA, and IWM over the same timestamps and outcome window
  • Disclose the date range, qualified-release count, mature-observation count, and maturity status
  • State entry and exit conventions, transaction-cost assumptions, and whether dividends are included
  • Show negative absolute returns, underperformance, partial windows, temporarily outdated data, and unavailable benchmarks plainly
CURRENT EVIDENCEPrimary

Fresh, source-backed context should drive the present review.

HISTORICAL CONTEXTBounded

Small or incomplete samples must remain limited and clearly labeled.

AI ROLEAssistive

AI may organize research, but it does not rewrite ranking rules or guarantee accuracy.

USER ROLEValidate

The user verifies sources, judges suitability, and makes every decision.

Current evidence remains primary. Time alignment does not prove causation; events remain source context unless stronger evidence supports a causal claim. Research-model performance is not an investment-account return, and historical results do not guarantee future outcomes.

GO DEEPER

Learn how to read the evidence.

Open the research library →