How Live Rankings and Editorial Picks work

The Live Rankings table ranks assets by the selected asset class, action, horizon, and return treatment. It is designed to show the most important decision fields in one row: consensus score, rating, expected return, dividend contribution, valuation, risk pressure, financial quality, volatility, and forecast price context.

The table is intentionally column-driven. Some columns are universal across asset classes, while others only appear when the underlying asset has suitable data. For example, Financial Health is equity-only, and P/E ratio is meaningful only when earnings fundamentals are available.

iPulse AI Engine V7 is the public name for the governed Top Picks pipeline. It covers asset preselection, independent AI-advisor forecasts, the asset-level consensus, risk and fundamental diagnostics, cross-asset scoring, ranking, and the final current-event editorial review. It is an end-to-end research system name, not the name of one language model.

1. Stage 1 · Candidate discovery

   Approximately 100,000 instruments are screened

   iPulse AI scans catalog and market metadata to find eligible candidates. This is the top of the funnel—not a count of assets that received AI-advisor forecasts or full Deep Analysis.

   ~100,000

   screening candidates only

   ~99,000

   Stocks

   ~200

   ETFs and ETPs

   ~50

   Crypto assets

   ~100

   Indices

   ~150

   Commodities and macro markets

3. Stage 2 · Actual Deep Analysis

   426 assets in Batch 7

   These selected assets—not the 100,000-instrument screening universe—were processed through the governed AI-advisor workflow, producing 5,964 individual advisor forecasts.

   Assets analyzed

   426

   Advisor forecasts

   5,964

   Analyzed cohort grew batch by batch

   Batch 1

   204

   assets

   Batch 2

   212

   assets

   Batch 3

   247

   assets

   Batch 4

   312

   assets

   Batch 5

   312

   assets

   Batch 6

   377

   assets

   Batch 7

   426

   assets

5. Stage 3 · Eligible ranking cohort

   The count changes for every ranking view

   Only deeply analyzed assets that match the selected asset class, action, horizon, and data-eligibility rules enter that ranking. Each public edition reports its actual assessed, eligible, and excluded counts; no fixed number is borrowed from the screening universe.

7. Stage 4 · Live Rankings

   Usually 50–70 assets per ranking table

   Every eligible asset in a Live Rankings table is scored with the same versioned iPulse AI Consensus Score formula. The table then sorts those comparable scores for its selected asset class, action, and forecast horizon—without asset-specific formula overrides.

   The exact count is disclosed on each table and can change as eligibility and data coverage change.

9. Stage 5 · Published Editorial Top Picks

   Up to 10 assets, depending on the page

   • 5 in a Top 5 snapshot.
   • Up to 10 in a specialist Editorial Top Picks table.
   • Up to 10 in the Best Long-Term Investments editorial table.

   A page can show fewer when fewer assets pass its quantitative and editorial requirements.

The default ranking uses the canonical dividend-inclusive iPulse AI Consensus Score when dividend treatment is enabled. If a view is switched to price-only where diagnostic fields exist, the table can show ex-dividend score and signal values for comparison.

A live ranking is a quantitative output, not an editorial endorsement. After the ranking layer updates, the Editorial Review Workflow creates a frozen candidate snapshot and tests the selected assets against evidence published after the underlying Deep Analysis run. This keeps current-event review separate from both model generation and formula-based scoring.

Editorial review entered the public operating system with V6 and remains active in V7. It was introduced after the raw Batch 6 Deep Analysis execution. It therefore governs publication and current-evidence validation; it did not influence the original Batch 6 model outputs.

Net Dividend Contrib. describes the dividend income component used in dividend-inclusive total return. It is not a marketing yield estimate; it is the pipeline input used to adjust expected return for dividend-paying assets.

dividend_adjusted_return =
  ((1 + price_return) * (1 + annual_net_dividend_contribution)^years) - 1

annualized_return =
  (1 + dividend_adjusted_return)^(1 / years) - 1

Assets without usable dividend history receive no dividend uplift. For some views, price-only diagnostics may still exist, but the canonical Top Picks ranking uses the dividend-inclusive basis when dividend treatment is enabled.

The Event Risks column is the user-facing label for the underlying 0-100 event-risk pressure score. The score is based on the SEO/GEO summary layer after repeated advisor concerns have been consolidated into consensus friction factors and consensus tail risks. It does not include historical volatility, Financial Health, or advisor disagreement, because those are separate diagnostics.

friction_pressure =
  sum(abs(friction_impact_pct) * rank_weight)
  where rank weights are 1.00, 0.75, 0.55, 0.40

tail_risk_pressure =
  sum(abs(downside_impact_pct) * probability_pct / 100 * rank_weight)
  where rank weights are 1.00, 0.65

raw_event_pressure =
  0.60 * friction_pressure + 0.40 * tail_risk_pressure

risk_score =
  percentile(raw_event_pressure), 0 to 100, higher means riskier

The Asset Snowflake uses the display-friendly inverse, Risk Resilience = 100 - the underlying Event Risk score.

Financial Health is shown only when enough equity fundamentals are available. It is not used as a universal score for crypto, commodities, FX, or indices because those assets do not share the same balance sheet and income statement structure.

financial_health_score =
  0.20 * liquidity
  + 0.25 * leverage_safety
  + 0.25 * profitability
  + 0.20 * cash_generation
  + 0.10 * operating_trend

The score can be AVAILABLE, PARTIAL, INSUFFICIENT_DATA, or NOT_APPLICABLE depending on how much usable fundamental data exists for the asset.

Top Picks avoids showing blank or misleading columns where possible. Event Risks appear when event-risk pressure is available. Financial Health appears when equity fundamentals are sufficient. P/E appears when earnings data supports a meaningful multiple. Voice-count columns appear in detailed table modes.

The main table focuses on dividend-inclusive expected return and annualized return. Price-only return remains part of forecast price diagnostics, but it is not the primary column for Top Picks ranking.