Compare the grading methods
Testing changes do not automatically change report scores. Saved input corrections can be reviewed and explicitly applied to the selected quarter using the Fund Data Corrections section. Source: AllFunds_Updated.xlsx; intended quarter: April–June 2026, as supplied by you. The export does not independently certify the as-of dates of every metric.
First hypothesis: cost and implementation characteristics
This first composite tests equity funds only: 50% cost, 25% turnover, 25% concentration. Lower values score higher within the same broad category and Category Name. These are experimental choices, not validated predictors of returns or suitability. Concentration measures the top-ten holding weight; it is not a complete measure of diversification. Low turnover does not necessarily mean low trading costs.
Method and limitations
Each component is a 0–100 relative score using average ranks for ties, with the lowest value receiving 100 and the highest 0. At least 20 valid peers are required for every component. Missing inputs are not treated as zero and weights are not redistributed. Duplicate tickers are excluded from both methods and both peer universes.
The composite is ranked again among complete equity peers (minimum 20), using average tie ranks. The original grade cutoffs are retained: A+ at 95th percentile, A at 90th, B at the selected 60th or 70th percentile, C at 50th, D at 33rd, otherwise F. The original method is recalculated on this export, preserving its own ranking and missing-value rules. Grades are relative to this export, including distinct share classes, and are not a forecast.
Fixed-income, allocation and other categories remain visible for research, with no experimental composite grade. Duration and credit quality describe exposure; they should not automatically earn rewards for being lower-risk. Tracking error, beta and R-squared are context only and have zero grading weight. Tracking-error and R-squared headers specify “vs Peer Group”; their precise calculation frequency, annualization and beta benchmark still need verification before scoring use. No bonus is awarded for a passive label.
Decile boundary correction
Only the out-of-range endpoint is corrected; other boundaries, ties, weights and grade cutoffs stay unchanged. The original baseline stays frozen. Peer rankings can change even for funds without a decile-11 input.
Client funds affected by the correction
Original formula — weight testing
Reweight the original-style decile-point formula by broad category. Duration now uses the confirmed middle-decile pattern: 60, 70, 80, 90, 100, 100, 90, 80, 70, 60 points. The original baseline retains its earlier duration rule. Fixed Income uses the confirmed “Other” rules. This retains the original tie ranks, missing-value behavior and selectable B cutoff; the experimental grading-scale selector below does not affect this test. Test grades: A+ ≥90th, A ≥70th, B ≥ selected cutoff, C ≥50th, D ≥33rd, otherwise F. With B set to 70th, there is no B band because A takes precedence. The untouched original baseline retains its original cutoffs: A+ ≥95th, A ≥90th, B ≥70th. No randomized adjustments are used. Only factors with a nonzero original category weight can be edited.
How deciles earn points
All factors are ranked from lower to higher numeric values within their broad category and category name. Higher numeric values therefore land in higher deciles. Expenses, turnover, price/book, top-ten concentration and standard deviations receive 100 down to 10 points across deciles 1–10. Holdings counts and Sharpe ratios receive 10 up to 100. Market capitalization receives 20, 40, 60, 80, 100, 80, 60, 40, 20, 10. Duration receives 60, 70, 80, 90, 100, 100, 90, 80, 70, 60. Credit quality receives 100, 90, 90, 60, 60, 40, 40, 30, 30, 10: lower YCharts values indicate higher credit quality. A source value of 10 can also be an unrated/unmapped fallback; the number alone cannot distinguish this. The frozen baseline retains its earlier credit scoring.
The inherited workbook decile calculation is retained: truncate(rank ÷ category count × 10 + 1), using the lowest tied rank. This can produce decile 11 at the maximum rank; that factor has no points when the correction is off, or uses decile-10 points when it is on. Missing-value behavior is also retained. Deciles displayed below make those cases visible.
Fixed Income Formula 2
An independent test using the confirmed decile-point rules for five factors, excluding credit quality. Default weights: 20% duration, 40% net expenses, and 5%, 10%, 25% for 3-, 5-, and 10-year deviations. Middle-decile duration earns the most points; lower expenses and deviations earn more points; this is a hypothesis to test, not a claim that lower risk is always better.
Only Fixed Income and the confirmed Other category are scored. Formula 2 updates the Allocation blend when the blend is enabled and Formula 2 is selected. The source labels deviations as standard deviation of quarterly returns; no annualization is added. Original missing-value and ranking rules are retained.Two separate assessments
Structure & cost keeps the adjustable equity-only formula. Risk-adjusted performance ranks the supplied Sharpe ratio against the same broad category and Category Name, with higher ratios ranked higher. Choose one horizon at a time; the default is 5 years. We do not average overlapping periods or substitute a shorter history when the chosen period is missing. Equity and fixed-income funds are eligible for this separate performance assessment.
Performance percentiles use average ranks for ties divided by the valid peer count, with the same letter cutoffs as above and at least 20 valid peers. Negative ratios are retained, but a high relative grade can still have a negative Sharpe ratio. These historical comparisons do not predict returns. Share classes count separately; metric as-of dates and provider definitions remain to be confirmed.
Only experimental B/C boundaries change. Scores, rankings, and original-method grades stay the same. A+ ≥95, A ≥90, B ≥ selected threshold, C ≥50, D ≥33, otherwise F. “Boundary changes” counts grade changes versus the 70th-percentile scale, not changes versus the original method.
Select a fund to inspect its score
Underlying holdings review
Upload a CSV with Symbol, Name and % Weight. Every source row is retained. Matching symbols are combined; missing symbols remain separate unless you approve a suggested name match. Rounded zero-weight rows remain included in the provisional count. Weights are never scaled to 100%.
Diversification assessment pilot
Three-fund pilot: DFISX, DGEIX and DFUSX. These assessments are separate from existing scores. Imported counts alone do not establish diversification. No assessment is pre-approved.
Proposed rubric and spreadsheet workflow
Each of three criteria receives Meets = 100, Partial = 50, or Does not meet = 0. Unknown prevents a completed score. The criteria have equal weight in this initial hypothesis. A completed average of 80 or higher is “Meets criteria”; 50–below 80 is “Concentration considerations”; below 50 is “Does not meet criteria”. These are provisional research choices, not established industry thresholds or an overall investment recommendation.
- Breadth for intended role: evidence shows exposure across a sufficiently broad set of underlying investments for the stated mandate. A wrapper count alone is insufficient.
- Concentration: evidence supports acceptable issuer/security concentration for that role. Missing look-through percentages require alternative documented evidence, or Unknown.
- Overlap and exposure: evidence addresses overlap across sleeves and relevant sector/geographic concentrations. Multiple underlying funds alone do not satisfy this criterion.
Both assessment routes use the same rubric; the evidence differs. There is no automatic fund-of-funds bonus. The spreadsheet register stores reviewed decisions, not invented underlying metrics. Raw YCharts exports continue to supply numerical data elsewhere; they do not automatically answer these review questions.
Register CSV columns: Ticker, Method, Intended Role, Breadth, Concentration, Overlap, Source, As Of, Review Due, Notes. Method is security or structure. Criterion values are meets, partial, fails or unknown. Save an exported register and use it as the editable spreadsheet template.
Earlier alternate-formula experiment (retained for reference)
Saved fund exceptions — alternate formula test
Approve an exception once for funds whose holdings count and top-ten concentration are not comparable. The alternate original-formula test omits both inputs and proportionally redistributes their weights. It does not measure diversification. Peer scores are recalculated using the same alternate formula before assigning the exception fund’s percentile. Results stay separate from the existing grades and reports.
Suggested first reviews: DFISX, DGEIX, DGSIX and DFAW. None is automatically approved. Low holdings count remains only a screening trigger. Current formula weights, decile correction and original-test grade thresholds are used; these settings are not saved by this register.
The register saves in this browser for this local page. Other browsers, moved files or cleared browser storage may not retain it; export a backup. A future quarterly-import connection is still needed—this pilot uses the currently loaded AllFunds dataset.Fund Data Corrections
Review & Apply Corrections to Selected Quarter
Save your manual overrides first. This opens the selected reporting quarter with a preview before you confirm. Test weights and grade thresholds are not applied to the report.
Correct the inputs, not the letter grade. Select a fund in the results list and open Manual override to enter the replacement holdings count or top-ten concentration, source, date and explanation. Corrections recalculate the same test formula and affected peer rankings. Original baseline grades remain unchanged.
Corrections and their revision history save in this browser. Export correction history, then load that JSON under Optional documented corrections before uploading AllFunds in Source Reconciliation. Testing edits do not automatically change reports. Only funds actually held by a household should appear in that household’s disclosure.