Every dataset behind Capmark, where it comes from, how it was checked, and every correction made. Core data trust: 92 / 100 · Overall: 75 / 100.
Automated since Nov 9, 2026: the Daily Brief tape, the Screener and Watchlist closes and indicators, the sector heatmap and movers refresh automatically after each trading day — prices from Massive end-of-day bars, cross-checked against Finnhub; index levels and economic data from FRED. The method below (14-day Wilder RSI, 50-day average, 52-week high, 20-day relative volume) is unchanged. The ledger records the original Nov 6, 2026 audit.
People make financial decisions from pages like this, so Capmark is built to be checked, not believed. Every figure carries its source, its date, and a credibility badge with a one-line reason. Thin samples are flagged Low instead of averaged into a headline. Conflicts between reputable datasets are shown, not resolved in our favor. Where no sourced figure exists, Capmark prints “unavailable” or “no clean analogue” — it never estimates one into existence. No figure on this page is a prediction, and nothing here is a buy, sell or hold call.
Capmark now publishes two numbers, both computed live from the per-dataset scores in the Credibility Ledger below — neither is typed in, and the Overall score is never hidden.
Core Data Trust — — / 100 (7 datasets): the figures a reader actually trades on — the Daily Brief tape & prices, the Screener, the Watchlist, the stock detail view, the midterm price record, the primary six-combination price returns (RBC), and the peer-reviewed economy scorecard. Every price dataset in Core is now second-source verified: all 39 Screener / Watchlist names match Finnhub on every field (close, day, week, 1-month and YTD) as of the Nov 7, 2026 completion pass, and every index close matches CNBC / AP / FRED.
Overall — — / 100 (20 scored datasets): everything above plus the context & history datasets and the tools. Tools & illustrative — — / 100 (3 scored datasets: the one-stock simulator, Ask It Anything, crypto) are scored separately and stay visible with their own honest scores; they no longer drag Core down, but they do count in Overall. The simulator's ±6-point band is an illustrative construction, not data, so it is excluded from both averages and labeled illustrative wherever it appears.
How each average is built — openly: within each group it is the unweighted mean of that group's per-dataset scores — every dataset counts once. Capmark does not weight by importance, because weighting toward the strongest datasets would flatter the numbers. Splitting is not weighting: both numbers are shown, with their membership stated here and per-dataset group tags in the ledger.
Freshness is the biggest remaining drag — stated plainly. Every price on this page is the Friday, Nov 6, 2026 close, timestamped as such and never presented as live. A snapshot can be perfectly sourced and still go stale: each passing day lowers what these scores can honestly claim. What would lift them further is a daily refresh — rebuilding the tape, Screener and Watchlist from that morning's closes and re-running the second-source check — not a formula change. This page does not fake a live feed.
What these scores are — and are not: they measure data trust — sourcing, method and honesty — not predictive certainty. Some samples are irreducibly small: 19 midterms is 19, and the Scenario Lab holds 30 tagged years; no sourcing can enlarge history, so those datasets cap below perfect by nature. A true 100/100 is not honestly reachable for a history-based market tool, and Capmark does not claim it. Last audited: Nov 7, 2026 (figures as of the Nov 6, 2026 close; second-source validation completed Nov 7, 2026 — CNBC, AP, Reuters, FRED, Finnhub, CME FedWatch reports and Cisco Investor Relations; disagreements found — a 0.01pt S&P rounding difference, a 7-cent SPY vendor difference, MarketBeat outlier prints on several names — are recorded in the ledger, not averaged away).
Reproduce this: pick any figure on the page → find its dataset's row in the Credibility Ledger → the row names both sources and the check date → recompute the group averages from the per-dataset scores printed there; the header numbers must match. The methodology for every calculation is in plain English in the next card, and every correction ever made is in the corrections log.
Full audit, dataset by dataset: the Credibility Ledger — its per-dataset scores are in this expanded view, directly below · every source named: Sources & method notes.
Every dataset on this page, audited in November 2026 and second-source validated on Nov 7, 2026 against CNBC, AP, Reuters, FRED, Finnhub, CME FedWatch reports and primary company sources — each row records both sources and the check date where a check was made, and says “single-source” where none was. Ratings weigh five things: source authority (primary/official vs secondary), sample size, date range & freshness, method transparency, and whether each figure is directly sourced, calculated, or estimated. Anything unsourced, estimated, stale or conflicting is flagged unverified here rather than presented as fact. Ratings: High Medium Low
| Section / dataset | Source | Range · sample | Rating | Caveat / unresolved |
|---|---|---|---|---|
| Daily Brief — prices & tape | Yahoo Finance daily closes; validated Nov 7, 2026 vs CNBC closing reports, AP and FRED (S&P series) | Close of Fri Nov 6, 2026 | High | Second-source validation — Nov 7, 2026: every tape level confirmed exactly by a second reputable source — S&P 500 7,376.86, Nasdaq 24,579.42, Dow 51,285.73 (CNBC closing reports, echoed by AP), VIX 25.64, −9.01%, 10-year 4.86% (4.859%, CNBC/Yahoo), Thursday's closes 7,353.73 / 24,315.00 (CNBC; FRED SP500 series), weekly changes −2.7% / −3.4% / −0.9% (CNBC). Two flags, kept open: (1) Friday's S&P % prints +0.31% at CNBC vs +0.32% at AP — the level is identical; Capmark keeps AP's +0.32% and discloses the 0.01pt rounding difference rather than averaging it away. (2) SPY's close prints $735.29 (Yahoo) vs $735.36 (Nasdaq.com) — a 7-cent vendor difference; Capmark keeps its stated Yahoo figure. Correction logged: the 2-year yield was corrected 4.46% → 4.51% (it rose 5bp Friday; 4.46% was Thursday's print). Dated snapshot — stale by design after Nov 9; no live/intraday claim. |
| Daily Brief — week narrative & calendar | AP, CNBC and Reuters reporting; BLS Employment Situation & BLS / Federal Reserve published schedules; Cisco Investor Relations; CME FedWatch via dated Barron's / Investopedia / Seeking Alpha reports; PwC post-election note | Nov 2–13, 2026 | Medium | Trust-audit fixes stand: exact election seat counts are still not stated (AK & MS Senate races uncalled at compile time — only AP's call of chamber control, with source and date). Second-source validation — Nov 7, 2026: two removed items were restored only because a primary source now confirms them — Cisco: Q1 FY27 results Thu Nov 12, 2026, confirmed by Cisco Investor Relations' own release (Nov 3); December-hike odds: CME FedWatch read 72.8% pre-jobs and ~77–79% post-jobs on Nov 6 (Barron's / Investopedia / Seeking Alpha, all citing CME; CNBC live coverage ~70% earlier Friday) — quoted in the Brief as a timestamped range, because the odds move intraday. Jobs (+165K vs 59K CNBC consensus / 65K Reuters poll; unemployment 4.1%) corroborated by CNBC, Reuters and Bloomberg reporting of the BLS release. Desk S&P levels stay removed: the only levels found in this pass were an unnamed newsletter's self-described own estimates — not a named desk. CPI consensus 3.9% is CNBC's survey figure, labeled in place. |
| Screener (39 names) | Calculated from Yahoo Finance daily closes; second vendor: Finnhub exchange data + CNBC / closing-table reports (validation pass, Nov 7, 2026) | 1 year of daily bars to Nov 6, 2026 · 39 names | High | All figures are formula outputs (14-day Wilder RSI, 50/200-day averages, 20-day relative volume) — reproducible from the stated inputs. Cross-check record — Nov 7, 2026 (Yahoo vs Finnhub, an independent vendor, checked Nov 7): 8 names match on every field — close, day %, week %, 1-month % and YTD % identical: AMD (449.04, +11.04%, −11.05%, −27.65%, +109.68%; RSI 38.4 and −31.8% from its 52-week high also match), AAPL (342.37, −0.52%), MSFT (530.61, −0.99%), TSLA (358.22, +4.23%), GOOGL (284.78, +0.78%), JPM (334.49, −0.84%), UNH (394.15, −0.40%), XOM (165.87, +0.34%). 8 more match on close and day % in CNBC-linked closing tables and market recaps: NVDA (186.64, +2.78%), META (610.15, +0.84%), PLTR (160.38, +3.48%), AVGO (293.69, +0.74%), AMZN (259.30, −0.35%), SMH (+3.62%), WMT (−1.72%), JNJ (+1.43%). Completion pass — Nov 7, 2026 (Yahoo vs Finnhub, checked Nov 7): the remaining 23 names now also match on every field — close, day %, week %, 1-month % and YTD % identical: CRM (213.65, −3.94%, −6.77%, −6.21%, −19.35%), ORCL (116.10, +0.82%), HD (343.47, −1.85%), NKE (38.59, −2.45%), SBUX (105.41, −0.71%), NFLX (63.41, −3.28%), DIS (116.75, −0.40%), BAC (49.42, −0.68%), GS (865.82, −0.36%), V (422.65, +0.01%), MA (652.70, −0.49%), BRK-B (570.89, −0.25%), COIN (134.76, −1.51%), CVX (212.61, −0.15%), COP (136.27, +0.60%), LLY (1,354.00, −0.79%), PFE (29.39, −1.64%), PG (165.40, −1.02%), COST (1,053.56, −1.13%), KO (95.67, −0.06%), PEP (141.34, −0.96%), CAT (558.63, +0.90%), UNP (290.16, −2.61%) — all 39 of 39 names are now second-source verified on every field; no figure changed, because every figure matched. Flagged, not hidden: one MarketBeat compare page prints JPM at $334.67 / −0.79% and XOM at $165.81, and MarketBeat's own pages print LLY at $1,372.15, MA at $653.45, BAC at $49.54, CAT at $559.32, UNP at $290.50 and DIS at $116.72 — Finnhub and the closing tables agree with Capmark's figures on every one of these names, so Capmark keeps its figures and records the outliers here rather than averaging them in. Snapshot only; not a live screen. Opens on the Oversold screen; screen counts and per-row “why it's here” hits are computed from the same data. |
| Midterm case study | Capital Group / RIMES / S&P, via Greenbush Financial Group; 2018–19 from Capital Group's later edition; 2022–23 calculated from S&P 500 closes (3,828.11 → 4,382.78) | 1950–2023 · all 19 midterms | High | Complete sample for its span and method disclosed (Election Day start; Nov 5 proxy in other years). Price return only — dividends excluded. Second-source validation — Nov 7, 2026: the pattern — higher 12 months after every midterm since 1950, averaging ~15% — is corroborated by four independent reputable sources: J.P. Morgan Asset Management (positive after every midterm since 1950, ~15% average), Nationwide (never lower one year after a midterm since 1950), Fidelity (+14.7%, 1950–2023) and Capital Group's own later editions (15.1% as of Dec 31, 2025; 15.4% in its 2026 update). Capmark keeps its stated 15.0% — the mean of its own 19 bars, 1950–2023 — and discloses the 15.0–15.4% edition spread openly: the differences are window and edition, not error, and they are not averaged together. 6.8% “all other years” remains the original source's own figure, not independently recomputed here. The 2026 midterm's window closes Nov 2027 and is correctly absent. |
| Six combinations — RBC (builder, simulator base, ranked chart) | RBC Wealth Management / Bloomberg, 1953–2025 — corroborated verbatim Nov 7, 2026 by multiple independent republications of the RBC table and RBC's own post-election commentary | 1953–2025 · 6 combinations | Medium | Designated PRIMARY in the November 2026 trust audit — the only six-combination dataset the builder, simulator and headlines draw on. Second-source validation — Nov 7, 2026: all six averages match the RBC/Bloomberg table exactly in every republication checked: Dem pres · Divided +17.0%, Dem pres · Rep Congress +16.3%, Rep pres · Rep Congress +13.3%, Dem pres · Dem Congress +8.0%, Rep pres · Divided +7.3%, Rep pres · Dem Congress +4.9% — and RBC's own Nov 6, 2026 commentary confirms the +4.9% weakest-combination figure and the post-election setup. This row's score rose because the dataset is no longer “via secondary summaries” only — the table itself is now multiply corroborated. Still price-only returns, and some combinations rest on barely a decade of years. It conflicts with the 1926–2023 alternative on unified governments; the two are never blended or averaged together. |
| Six combinations — long dataset | Retirement Researcher data via WT Wealth Management | 1926–2023 · 4-way split only | Medium | Labelled ALTERNATIVE — shown for comparison in Section 03 only, with the method difference stated (period from 1926 incl. the Great Depression; four-way split, so it cannot separate all six combinations). Year counts per bucket vary widely (13–36 years). Reached via secondary summary, not the primary publication. Never blended with the primary. |
| Scenario simulator band | Base: RBC averages · Band: this page's construction | — | Low | The ±6-point band is an illustrative estimate — unverified as a statistic. It is labeled as such everywhere it is used; do not quote it as a confidence interval. Excluded from the Trust Center average — an illustrative construction is not data, so it no longer drags (or pads) the score; its exclusion is stated here and in the Trust Center methodology, not done silently. |
| One-stock simulator | Calculated from Yahoo Finance dividend- & split-adjusted closes | 1986–2025 · 12 stocks | Medium | Prices are real and adjusted; per-combination samples are only 2–8 years per stock, so results are era-dominated. “Confidence” is a bootstrap frequency, not a probability — labeled in place. |
| Scenario Lab analogues | S&P returns: Slickcharts / S&P DJI series · Force tags: hand-assigned from NBER dates, Fed history, BLS record | 1973–2024 · 30 of 52 years, hand-selected | Medium | Returns sourced; tags are editorial and the year set is selected, not exhaustive. Most real-world mixes match ≤6 years. Sector led/lagged notes are qualitative summaries, not a per-year return table. Second-source validation — Nov 7, 2026: the return series itself was spot-checked against a second independent price-return table and matched exactly on every year checked — 1974 −29.7%, 2008 −38.5%, 2022 −19.4%, 2023 +24.2%, 2024 +23.3% (Slickcharts price returns vs the Plutux / daily-close compilation and SimianX's two-vendor-checked year-end levels). The force tags remain editorial — that, not the returns, is what holds this score down. |
| Policy focus — “How policy moves markets” | No new data: a regrouping of the Scenario Lab dataset above by policy tag (Fed stance, fiscal/tax, deregulation, trade war, debt-ceiling, shutdown), plus the Sector Playbook's qualitative tilts where no return exists | Same tagged years, 1973–2024 · slices of n=1 to n=12 | Medium / Low slices | Every average is printed with its n. Since the November 2026 trust audit, slices with n=1–2 carry no headline average at all — QT (n=2), tax hikes (n=1), deregulation (n=1) and tariffs (n=2) are shown as individual episodes with their n stated; shutdown/debt-ceiling (n=3) keeps its warning. Regulation tightening and isolated government spending have no tagged sample — shown as “no clean historical analogue” with no figure, by design. Averages for multi-force years cannot isolate the named policy's own effect; the cards say so per episode. Second-source validation — Nov 7, 2026: this section inherits the Scenario Lab return validation above (episode years 1974, 2008, 2022, 2023, 2024 confirmed against a second price-return source) — its score rose with the Lab's, and no further. The policy tags themselves are still hand-assigned. |
| Chamber by chamber (Dow) | Mike Patton / Integrity Wealth Mgmt via Forbes; Forecast-Chart compilation; AEA 2025 paper as counterweight | 1946–2020 · 75 years | Medium | Single-analyst compilation, not peer-reviewed; rows overlap (chambers usually align). A 2025 AEA paper finds no reliable single-chamber premium — the conflict is shown in the section. |
| President by president | TD Economics / Bloomberg via Cabot and secondary term tables | Since 1946 · 6 modern terms ranked | Medium | Two compilations differ by a few points on identical terms (e.g. Obama +182% vs +189%) — method conventions, disclosed in place. Tiny sample; starting-point luck dominates. |
| 13-era explorer | Compiled from the sources in Section 10 | 1945–present · 13 eras | Medium | Era market totals are approximate. Trump II era is in progress — provisional, unverified as a completed record, and excluded from completed-term rankings. |
| Economy scorecard | Blinder & Watson, American Economic Review (2016); EPI update; NBER | 1949–2023 | High | Strongest sourcing in the app. Interpretive caveat stands: the authors attribute ~half the partisan gap to luck (oil shocks, productivity, world conditions). |
| Sector playbook & leaderboard | Franklin Templeton; Ameriprise/Bloomberg; Fundstrat; sector-ETF study | Windows from 1930 (medians) to 1998–2022 (ETF era, 6 cycles) | Medium | Window-dependent winner (financials lead Q4, finish last over 12 months). Per-combination sector returns are not measured anywhere — presented as policy tilts only. Communication Services exists as a sector only since 2018. |
| Crypto — BTC & ETH | Exchange daily-close compilations (cryptocurrencyhistory), cross-checked vs Coin Metrics / CoinMarketCap series; drawdowns: Plutus; GrowYourEthereum (Coinbase) | BTC 2011– · ETH 2015– · 4 presidencies, 3 midterms | Medium | Early-year figures differ a few points between series — disclosed. Term totals are calculated from year-end prices. No per-chamber or post-midterm average is claimed: the sample cannot support one. Second-source validation — Nov 7, 2026: the Nov 6, 2026 prints in the Brief were confirmed against independent crypto close reports: BTC ≈ $75,747 (BusinessToday's Nov 6 close report — exact match) and ETH $2,235 (exact across Basis Desk / Coinbase-Kraken snapshots and CoinGecko-based boards). Disclosed honestly: crypto has no single official close — other reputable snapshots the same evening print BTC between $75,736 (UTC close) and $75,966 (5 PM ET), a ~0.3% spread by timestamp and venue. Capmark's figure sits inside that range; ETH agrees everywhere checked. |
| All Drivers map (31 drivers) | Episodes cited per card (Fed, BEA, BLS, IMF COFER, SEC, Census series) | Episodes 1970s–2024 | Medium | Episodes are real and sourced; the strong/mixed/weak/narrative grades are editorial judgment against the criteria at the top of this ledger. “Weak” and “mostly narrative” drivers are carried as unverified patterns by design. |
| Ask It Anything | Assembled only from the datasets above | — | Medium | Inherits each dataset's rating; every answer prints its source. Questions outside the page's data are declined rather than answered from general knowledge. |
| Watchlist — glance view (39 names) | Same calculated dataset as the Screener — Yahoo Finance daily closes; day change in $ derived exactly from close and Friday % change; all 39 of 39 names second-source checked Nov 7, 2026 (see Screener row) | Close of Fri Nov 6, 2026 · 39 names | HighTrust score · 96 / 100 | Sparklines join five sourced anchor closes (Dec 31, 2025 · 1 month ago · 1 week ago · previous close · Friday's close), each derived from the name's sourced % changes — the dots are data, the joins are not a daily price path, and the view says so in place. Snapshot, not live. Second-source validation — completed Nov 7, 2026: inherits the Screener cross-check: all 39 names' closes, day, week, 1-month and YTD changes are confirmed by Finnhub (16 in the first pass, the remaining 23 in the completion pass — per-name record in the Screener row). No name in this view is single-source. |
| Stock detail view — key stats & chart | Screener dataset fields only; previous close and 52-week high derived by formula from sourced % changes and labeled Derived | As of Fri Nov 6, 2026 · 39 names | MediumTrust score · 86 / 100 | Shown stats are sourced or exactly derived (rounding in the source % changes bounds derived prices to within ~0.1%). Second-source validation — completed Nov 7, 2026: Finnhub's previous closes now reproduce Capmark's derived previous closes for all 39 names (the first pass covered 8 — e.g. AMD $404.40, AAPL $344.17, MSFT $535.93, TSLA $343.69; the completion pass added the rest — e.g. CRM $222.41, COST $1,065.59, UNP $297.95, LLY $1,364.78), confirming the derivation method against a second vendor for the whole universe. Open, day range, share volume, average volume, market cap, P/E, dividend yield and EPS are not in this dataset — shown as Unavailable, never estimated — that incompleteness, not sourcing, is what holds this score at 86. Chart ranges beyond the sourced anchors (1D intraday, 3M, 6M, 1Y, 5Y, Max) show an explicit unavailable note instead of a chart. |
| Top News | Associated Press, CNBC, Reuters and Bloomberg reporting; BLS Employment Situation and BLS / Federal Reserve published schedules; CME FedWatch via dated reports; compiled in the Nov 7, 2026 Daily Brief | Nov 3–10, 2026 · 7 items | MediumTrust score · 80 / 100 | Every item names its source and date — now including CNBC on the market items and a CME FedWatch item added in the Nov 7 validation pass with its readings timestamped (72.8% pre-jobs, ~77–79% post-jobs, Nov 6). Exact election seat counts are not stated (AK & MS Senate races uncalled at compile time); only AP's call of chamber control is reported. No headline is published without a named source; tickers with no verified item state that in their detail view. Not a live news feed. |
What the November 2026 trust audit fixed: 1 · Conflict: the two six-combination datasets still rank unified governments oppositely — now RBC 1953–2025 is the designated primary (the only one the tools draw on) and 1926–2023 is a labeled alternative; they are never averaged together. The two presidential compilations still differ by up to ~7 points on a term total, and the AEA paper still disputes the chamber figures — both conflicts shown, not hidden. 2 · Removed, not left looking verified: one desk's S&P levels were removed and stay removed after the Nov 7 re-check; the December-hike odds and Cisco's earnings date were removed in this audit and restored on Nov 7, 2026 only after primary sources confirmed them (CME FedWatch readings on the record; Cisco Investor Relations' own release) — see the validation paragraph below; exact 2026 seat counts are no longer stated (two Senate races uncalled at compile time — only AP's chamber-control call is reported). 3 · Illustrative, excluded: the simulator's ±6-point band is labeled illustrative everywhere and excluded from the trust average. 4 · Thin samples: policy slices with n=1–2 (QT, tax hikes, tariffs, deregulation) no longer carry headline averages — episodes only, n stated. 5 · Still provisional: the Trump II era (in progress, excluded from completed-term rankings). 6 · Stale by design: the Daily Brief and Screener are frozen at the Nov 6, 2026 close — a live edition must be rebuilt each morning. Nothing in this ledger was filled in from memory: where a figure could not be tied to a named source, it is flagged above instead of asserted.
What the Nov 7, 2026 second-source validation pass did: 1 · Confirmed exactly: all index closes (CNBC/AP/FRED), all six RBC averages (multiple republications of the RBC/Bloomberg table), the Scenario Lab price-return series on every spot-checked year (second independent table), BTC/ETH Nov 6 prints — and, after the completion pass below, 39 of 39 Screener names (Finnhub, every field). 2 · Restored on primary-source confirmation only: Cisco's Nov 12 earnings date (Cisco IR) and the December-hike odds (CME FedWatch, quoted as a timestamped range). 3 · Corrected: the 2-year yield, 4.46% → 4.51%. 4 · Flagged, not averaged away: CNBC +0.31% vs AP +0.32% for Friday's S&P; SPY $735.29 vs $735.36 across vendors; one MarketBeat JPM outlier; crypto's $75,736–$75,966 timestamp range. 5 · Still single-source, said plainly: the USO oil print, and every dataset whose score did not move. No Screener or Watchlist name remains single-source. 6 · Still removed: desk S&P levels — no named desk source exists. 7 · Completion pass (Nov 7, 2026): the remaining 23 names were checked against Finnhub and matched on every field — all 39 of 39 are now verified, no price changed, and the Trust Center now splits the headline into Core Data Trust and Overall, both computed from this ledger. The scores rose because the data earned it; the formula inside each average did not change.
Method caution, stated plainly: these are historical averages over different windows, indexes and dividend treatments. They describe what happened; they do not establish that a party caused it, and they are not a forecast. The Trump second term is in progress and excluded from completed-term rankings.