Rebuilding cash-flow indices and testing portfolio choices
This report checks daily index returns using available constituent snapshots, then compares research portfolios over 80 monthly intervals. Historical data and trading assumptions still need verification.
Evidence record reviewed ·
Findings and limits
The available snapshots support a partial rule replay. Cash-flow, dividend and custom portfolios have aggregate results. Complete official-history replication and production qualification remain open.
- Daily returns track the official series closely within the available snapshot sample. Historical constituent coverage remains incomplete.
- 980092 is the current research index, 932365 is the broad cash-flow reference, and 000922 is the dividend reference. Keep their roles separate.
- Further work needs rule changes and financial reports available at each historical date, attribution to size, industry, valuation and quality, and a common cost model.
Evidence limits
- The full ten-year history of official constituents and methodology is incomplete.
- The public record contains aggregate results only. It excludes stock-level signals, portfolio weights and execution details.
- Production qualification and full out-of-sample validation remain incomplete.
Official-snapshot calibration
Reading the correlations
Correlation measures how closely changes in the two daily return series move together. A value near 1 indicates close co-movement. Return differences and historical coverage need separate checks.
- Sample and scope
- Available snapshots; holding-count research covers 80 monthly intervals.
- Scope
- Official-snapshot checks and separate research portfolios
- Baseline
- Compare calibration to official daily returns, and portfolio choices within a common research setup.
- Input basis
- Partial constituent records and reconstructed financial history
The replay uses available constituent snapshots. The full ten-year official history remains incomplete.
| Object | Daily return correlation in available replay |
|---|---|
| Dividend benchmark | 0.997 |
| Broad-market free-cash-flow benchmark | 0.992 |
Which historical records are available
These dates mark different records. The nine membership intervals do not supply historical weights.
15 August 2024
The current method takes effect. Membership from this day through 17 September is missing from the export.
From 18 September 2024
Nine official adjustment intervals, each with 100 constituents. Membership changes are available with no historical weights.
30 September 2026
The official weight export provides this one snapshot. Other dates still need weight records.
Shadow portfolio comparisons
Terms used in the charts
- Research portfolio
- A portfolio replayed for research using selected rules and reconstructed inputs. Its results depend on the rules and input history described here.
- Weighting
- A stock’s share of portfolio capital. Equal weight gives each holding the same share. Cash-flow weighting with a cap uses cash-flow inputs and limits weights. The public chart does not specify every construction parameter.
- One-way turnover
- A measure of how much of a portfolio is replaced at rebalancing. This chart reports its average. Higher turnover means more trading, and costs depend on execution assumptions.
- Point-in-time inputs
- Inputs must have been available at the historical decision time. This is the PIT requirement in the original record. Later financial revisions need their own availability checks.
How returns change with holding count
980092 standard research replay over 80 monthly intervals, using mixed historical rules and reconstructed data. Returns belong to these research portfolios. Official index history is separate and remains incomplete.
Annualized research portfolio return
- Equal weight
- Cash-flow weighting with a cap
Bar length starts at zero. Both weighting methods share the same scale.
Equal weight leads at 10, 50 and 100 holdings. Capped cash-flow weighting leads at 20 and 30. Fresh-data tests and production qualification remain open.
Exact chart values
| Number of holdings | Equal weight | Cash-flow weighting with a cap |
|---|---|---|
| 10 | 16.99% | 15.12% |
| 20 | 15.74% | 16.54% |
| 30 | 14.22% | 14.48% |
| 50 | 13.02% | 11.53% |
| 100 | 13.07% | 12.24% |
Source records · decision
How turnover changes with holding count
The same equal-weight research family retains some existing holdings to reduce replacements. Only buffered results are shown. A matching unbuffered series is unavailable here.
Average one-way turnover
- Equal weight with holding buffer
Bar length starts at zero. This chart contains only buffered portfolios.
Turnover is 16.90% at 50 holdings and rises to 19.49% at 100. This chart cannot quantify the reduction from buffering or establish extra return.
Exact chart values
| Number of holdings | Equal weight with holding buffer |
|---|---|
| 10 | 24.00% |
| 20 | 21.50% |
| 30 | 19.33% |
| 50 | 16.90% |
| 100 | 19.49% |
Source records · decision
Comparison scope and review date
Compare weights under the same research protocol and selected holding count. Official history is recorded separately.
Evidence record reviewed · 2026-10-02
cash-breadth · equal · breadthResearch record
The four steps below track what was checked and how far each result reaches.
- 01
Available snapshots
Check the daily movements
- Question
- Can available constituent snapshots reproduce daily returns?
- Observation
- Snapshot replay correlations are 0.997 for the dividend benchmark and 0.992 for broad free cash flow.
- Interpretation and decision
- Use the result to check this sample. Missing historical constituents still need to be recovered.
Source records · status
- 02
2026-09-04
Compare how many stocks to hold
- Question
- Does a more concentrated portfolio consistently produce higher returns?
- Observation
- Results change with both holding count and weighting. Returns do not rise or fall consistently as the count changes.
- Interpretation and decision
- Choose holding-count rules before testing on fresh data. Selecting the highest entry after inspecting this table would use the result to choose the rule.
Source records · decision
- 03
2026-09-04
Check turnover and trading costs
- Question
- Does keeping some existing holdings reduce trading, and is extra return independently established?
- Observation
- The source reports lower turnover with buffering. Every cost-stress variant earned less as assumed costs rose.
- Interpretation and decision
- These results describe portfolio construction and cost sensitivity. Official replication and tradable capacity still need separate checks.
Source records · decision
- 04
2026-10-06
Check which official records are available
- Question
- Which parts of official index history are available?
- Observation
- The 980092 adjustment export covers nine intervals from 2024-09-18, with 100 constituents in each interval and no historical weights. The weight export contains only the 2026-09-30 snapshot.
- Interpretation and decision
- The current method took effect on 2024-08-15. Membership is missing through 2024-09-17 and historical weights remain unavailable. The ten-year official replay is incomplete. Earlier snapshot correlations and research portfolios remain separate evidence.
Source records · overview
What still needs verification
Announcement availability
PartialSome early fields are proxies; field-level strict PIT review remains open.
Historical financial versions
UnverifiedNot independently qualified in the cited record.
Historical constituents and eligibility
PartialOfficial constituent snapshots are partial and historical rules changed.
Corporate actions and terminal prices
PartialCorporate actions, dividends and temporary replacements are not fully reproduced.
Execution timing
PartialRule-shadow and official-index calibration are separate evidence.
Transaction costs
PartialA common executable cost comparison remains open.
Reproduction status
Partial
Recorded run · cashflow-shadow-standard-80-months
Research portfolio replays and official constituent calibration are recorded separately.
Remaining reproduction gaps
A complete rerun must match original input availability, dependencies and historical eligibility. Source links and file hashes cover only part of that verification.
Code revision, runner hash and artifact hash were not recorded.
Next research question
Can the cash-flow rule hold up when historical rules are consistent, inputs were public at each date and execution assumptions match?
Research only · Verify historical rules and constituents first. Then use common dates, weights and execution costs for comparison.
Verify historical rules and constituents first. Then use common dates, weights and execution costs for comparison.
Qualification gate
Resolve the named gaps and retain an immutable audit record. Keep the existing research-only decision if qualification fails.
How the interpretation changed
Available snapshots
- Earlier interpretation
- The available snapshots showed high correlations with official daily returns. The scope of that result needed to be stated.
- What changed the interpretation
- Historical constituents are missing and methodology changed over time.
- Current interpretation
- Keep snapshot calibration and current-rule research portfolios as separate results, each with its own limits.
Sources and publication record
Publication moved on 23 October 2026. Figures, original source revisions and the 14 October evidence review are retained; this move adds no experiment or new source verification.
Data comes from the public research entry in the quantitative research repository. The official index, the rule-reconstructed shadow index, and the self-built portfolio are separate evidence types.
- Authoritative study overview
doc:quant-research.research.cashflow-indices-map - Cash-flow current status
doc:quant-research.research.cashflow-indices.current-status - Research decision receipt, 2026-09-04
doc:quant-research.decision.long-term-fundamental-v2-20260904
Original source versions
- Research record version
e1984591be37073d9e1b3b1dee8a7e7813652cb3- Evidence audit version
e83f2d10a5c45aa4b4bc4b28a288887f86500aa8
Source status at the prior review · matched · 2026-10-14T10:05:57+00:00
Download the complete public records
File fingerprints and publication contract
study.jsond46a414b421bca87c23d93aaefb1e5e582e9f66dc86d312a89488c1e530b492b
exploration.json7bad8aab1696827156debe7d7cd0d580c6d68b5329e19026eff507fd6eebf08a
navigation.json02843520ffd7525ec2e76aaba81b8389d52e52c74211661d4932051cf9782d87
publication.jsonf714542644ec5007e668d84c729539a67c2d90ba182b492e2ef77cba0fdd7a0d
doc:quant-research.research.fundamental-evidence-navigation
publication.json retains the original publication contract and source hashes. Missing run hashes remain unrecorded.
Related factor evidence
- pb-roe
Valuation and quality checks can help identify low-priced stocks with weak fundamentals in cash-flow portfolios.
- hk-fundamental-archive
Both records add cash-flow criteria, with their own industry and market-period limits.
- fundamental-family-shadow
Comparing fundamental factor families helps distinguish valuation and quality effects from cash-flow effects.