What We Learned
WHAT THE SYSTEM CAN LEARN
Which sources and repeated patterns are actually useful?
Evidence scores measure how complete the saved research is. They do not claim that a source or research area predicts markets.
How complete each publisher’s saved records are
A complete saved record is not a truth or forecasting score. It only says whether the required information was preserved.
Open every research area’s saved checks
Results by prediction type
Only individual saved results with the required proof count here. This breakdown is separate from the official overall score.
Repeated patterns with the most saved events
Different story links do not prove unique information. How much a source added, whether opposing evidence improved a decision and whether a market relationship helped a forecast remain unmeasured until their outcomes are linked.
How varied are today’s sources?
Publisher variety: 95.14 out of 100. Agência Brasil supplies 2 of 24 retained stories (8.33%), the largest publisher share in this edition.
Source locations · variety not measured
Languages · variety 94.79 out of 100
- ko: 2 stories · 8.33%
- pt: 2 stories · 8.33%
- ur: 2 stories · 8.33%
- ar: 1 stories · 4.17%
- de: 1 stories · 4.17%
- en: 1 stories · 4.17%
- es: 1 stories · 4.17%
- fa: 1 stories · 4.17%
- fr: 1 stories · 4.17%
- hi: 1 stories · 4.17%
- id: 1 stories · 4.17%
- it: 1 stories · 4.17%
- ja: 1 stories · 4.17%
- ms: 1 stories · 4.17%
- nl: 1 stories · 4.17%
- pl: 1 stories · 4.17%
- ru: 1 stories · 4.17%
- sw: 1 stories · 4.17%
- th: 1 stories · 4.17%
- uk: 1 stories · 4.17%
- vi: 1 stories · 4.17%
Research areas · variety 91.67 out of 100
- Energy & Power: 4 stories · 16.67%
- Agriculture & Food Supply: 3 stories · 12.50%
- Climate & Environment: 2 stories · 8.33%
- Extreme Weather, Meteorology & Natural Disasters: 2 stories · 8.33%
- Finance: 2 stories · 8.33%
- Banking & Credit: 1 stories · 4.17%
- Education & Skills: 1 stories · 4.17%
- Law & Regulation: 1 stories · 4.17%
- Manufacturing & Automation: 1 stories · 4.17%
- New Tools / AI Tools: 1 stories · 4.17%
- Price Hike Detector: 1 stories · 4.17%
- Semiconductors & Hardware: 1 stories · 4.17%
- Singapore News: 1 stories · 4.17%
- Social Trends: 1 stories · 4.17%
- Space & Satellites: 1 stories · 4.17%
- Telecom & Connectivity: 1 stories · 4.17%
Source locations describe the publisher or organisation, not necessarily the country of the reporter or the event. An unrecorded wire origin remains unknown.
Next improvement: do not score whether a source helped a prediction until enough released predictions have properly checked results.
The completed 18:00 UTC Bitcoin observations showed negative four-hour and twenty-hour changes. At 19:00 UTC, the same unchanged rule found both positive: approximately 0.507% over four hours and 0.190% over twenty hours. Recording that reversal is a useful lesson about instability, not evidence that either direction is reliably profitable.
The earlier fixed diagnostic remains six matching directions in 19 directional cases from 28 scheduled samples. It is a short sign-only diagnostic sampled every 20 hours; shared history does not make the cases independent. It did not test the exact one-basis-point target of the new prospective experiment and does not supply a calibrated probability.
The Graph retains 172 complete historical measurement windows and now displays 53 instruments, including six new story markers attached to four current source stories. New weekend events can fall beyond the last available cash-market bar. In that case there is no measured post-event return yet; a marker must not imply that a later move was observed.
The 29 saved learning events include evidence-method findings, the nine current Long Term reviews and this observed rule reversal. They are not 29 successful predictions. Learning here means retaining what the evidence changed, what it failed to establish, and which original terms must remain fixed.
See all 24 headlines and their sources →
What changed in the learning record
AAPL original August research review reconciled
The original August research receipt records an unverifiable result because its non-crypto endpoint method was not declared.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordBTC original August research review reconciled
The original 96-hour August research receipt records a missed direction and retains its eligible Coinbase candle.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordCL=F original August research review reconciled
The original August research receipt records an unverifiable result because its non-crypto endpoint method was not declared.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordCNY original August research review reconciled
The original August research receipt records an unverifiable result because its non-crypto endpoint method was not declared.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordETH original August research review reconciled
The original 96-hour August research receipt records a missed direction and retains its eligible Coinbase candle.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordEUR original August research review reconciled
The original August research receipt records an unverifiable result because its non-crypto endpoint method was not declared.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordGC=F original August research review reconciled
The original August research receipt records an unverifiable result because its non-crypto endpoint method was not declared.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordGOOGL original August research review reconciled
The original August research receipt records an unverifiable result because its non-crypto endpoint method was not declared.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordMETA original August research review reconciled
The original August research receipt records an unverifiable result because its non-crypto endpoint method was not declared.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordMSFT original August research review reconciled
The original August research receipt records an unverifiable result because its non-crypto endpoint method was not declared.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordNQ=F original August research review reconciled
The original August research receipt records an unverifiable result because its non-crypto endpoint method was not declared.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordNVDA original August research review reconciled
The original August research receipt records an unverifiable result because its non-crypto endpoint method was not declared.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordSOL original August research review reconciled
The original 96-hour August research receipt records a missed direction and retains its eligible Coinbase candle.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordTLT original August research review reconciled
The original August research receipt records an unverifiable result because its non-crypto endpoint method was not declared.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordUSD original August research review reconciled
The original August research receipt records an unverifiable result because its non-crypto endpoint method was not declared.
Next test: Use the list of issued decisions that do not already have a completed review and the exact original cohort method. Do not queue this completed receipt again.
This is an original completed review receipt reconciliation, not a new prediction result or historical founder score change.
See the underlying recordA missing L1 quote cannot be replaced with a candle
The saved price recovery Ethereum decision remains unverifiable because the required in-window L1 receipt was not captured.
Next test: Schedule observations under each issued contract; the new current publication terms declare a retrievable completed minute candle before issuance.
No old entry, method, outcome or founder total changes.
See the underlying recordBTC rule direction reversed before issuance
The same four-hour confirmation of twenty-hour direction changed from bearish at18:00Z to bullish at19:00Z. No prediction had been issued, so this is neither a miss nor a win. It is evidence that the short-horizon observation can change before an entry is frozen.
Next test: Freeze any actual new call only at its real issue time and retain the original future outcome rule. Compare later outcomes after they become due; do not change confidence automatically.
Observed preissue market change only. Original6/19 sign diagnostic, founder91 aggregate and all original outcome contracts remain unchanged. ETH and SOL also aligned positively but were not independently validated.
See the underlying recordBTC earlier fixed-rule diagnostic reviewed
Earlier observation at 2026-09-06T17:00:00.000Z: BEARISH. The 20-hour-spaced, sign-only diagnostic matched 6 of 19 directional samples. The later pre-issuance rule review at 2026-09-06T19:00:00.000Z supersedes this earlier Bitcoin direction.
Next test: Use the later pre-issuance review for the current decision. Only a real prospective issuance can create a future scored observation under its frozen terms.
Twenty-hour-spaced samples compare return signs only. They do not test the actual one-basis-point target, establish calibrated probability or causal evidence, or represent previously issued predictions.
See the underlying recordETH earlier fixed-rule diagnostic reviewed
Earlier observation at 2026-09-06T17:00:00.000Z: HELD, with no evaluated direction at that observation. The 20-hour-spaced, sign-only diagnostic matched 7 of 21 directional samples.
Next test: Use the later pre-issuance review for the current decision. Only a real prospective issuance can create a future scored observation under its frozen terms.
Twenty-hour-spaced samples compare return signs only. They do not test the actual one-basis-point target, establish calibrated probability or causal evidence, or represent previously issued predictions.
See the underlying recordSOL earlier fixed-rule diagnostic reviewed
Earlier observation at 2026-09-06T17:00:00.000Z: HELD, with no evaluated direction at that observation. The 20-hour-spaced, sign-only diagnostic matched 10 of 20 directional samples.
Next test: Use the later pre-issuance review for the current decision. Only a real prospective issuance can create a future scored observation under its frozen terms.
Twenty-hour-spaced samples compare return signs only. They do not test the actual one-basis-point target, establish calibrated probability or causal evidence, or represent previously issued predictions.
See the underlying recordCloudflare source evidence refreshed
Revenue rose from $639.8m in Q1 to $696.1m in Q2; annual growth increased from 34% to 36%. Current RPO growth increased from 34% to 35%. Q2 operating cash was $117.6m and free cash flow $56.4m. Q2 GAAP gross margin was 71.8%, below 74.9% a year earlier. GAAP operating loss widened to $205.7m from $67.3m. Cash and securities of $4,162.8m are a balance-sheet fact, not proof of recession survival or fair valuation.
Next test: Review remaining subtraits and valuation evidence before a complete score or ranking change.
Current partial evidence review; historical whole-rubric score remains separately dated.
See the underlying recordRocket Lab source evidence refreshed
Q1 revenue was $200.3m and Q2 $234.1m; backlog rose from about $2.2bn to $2.36bn. Q2 GAAP operating loss narrowed to $57.514m from $59.639m a year earlier. First-half operating cash use increased to $134.407m from $77.467m. Financing supplied $1,523.403m. Ordinary shares outstanding rose from 543.575m at December 31 to 598.180m at June 30. Large cash balances therefore do not establish self-funding growth. Neutron’s planned launch is not completed execution.
Next test: Review remaining subtraits and valuation evidence before a complete score or ranking change.
Current partial evidence review; historical whole-rubric score remains separately dated.
See the underlying recordTempus AI source evidence refreshed
Q2 revenue was $382.5m, up 22%. The issuer reports oncology volume growth of 31%, versus 28% last quarter, and MRD tests of 9,000 versus 6,500. It reports FDA approval of tumor-only xT CDx and a delivered AstraZeneca model. Net income of $5.6m included $98.5m of unrealized securities gains. GAAP operating loss was $75.913m; first-half operating cash use was $80.802m. Cash and securities were $820.7m, while convertible senior notes net were $1,172.781m. The Personalis transaction was an agreement, not proof of a completed acquisition.
Next test: Review remaining subtraits and valuation evidence before a complete score or ranking change.
Current partial evidence review; historical whole-rubric score remains separately dated.
See the underlying recordSolana source evidence refreshed
The official live status response reported all systems operational at the captured retrieval bracket. A point-in-time operational status does not measure long-run reliability, security, users, fees, token-holder value capture, concentration or valuation.
Next test: Review remaining subtraits and valuation evidence before a complete score or ranking change.
Current partial evidence review; historical whole-rubric score remains separately dated.
See the underlying recordMTG Final Fantasy Serialized Golden Traveling Chocobo sources rechecked
Wizards still identifies the golden serialized card as a 77-copy English Collector Booster variant. That distinguishes it from other Traveling Chocobo printings; it does not establish a current seller, completed-sale value or remaining sealed supply.
Next test: A verified physical copy, its condition, current seller terms and completed sales are still needed before any affordability or availability judgment.
Fresh source attempts and ten trait judgments do not prove current executable inventory, sale price or condition.
See the underlying recordPokemon Mimikyu #075 Paldean Fates Pokemon Center promo sources rechecked
The official product URL returned an access-challenge page despite HTTP 200, and PSA returned HTTP 403. This refresh therefore cannot reconfirm the exact stamped promo, current graded population or completed auction prices.
Next test: Accessible official variant details and a dated, condition-matched completed-sale record are missing. The blocked attempts are saved with their actual retrieval times.
Fresh source attempts and ten trait judgments do not prove current executable inventory, sale price or condition.
See the underlying recordPOP MART THE MONSTERS Big into Energy ROCK THE UNIVERSE sources rechecked
The official page still specifies the exact ROCK THE UNIVERSE doll, including its size and August 2025 release. The re-opened company report describes broad interest in the Big into Energy series, but those series-wide claims do not prove scarcity or resale demand for this particular doll.
Next test: An exact production limit, present purchasable inventory, condition evidence and completed sales for this doll remain missing.
Fresh source attempts and ten trait judgments do not prove current executable inventory, sale price or condition.
See the underlying recordLEGO Inside Tour Anniversary Collection 4000046 sources rechecked
LEGO still describes the annual participant-only gift. The freshly opened BrickLink entry for 4000046-1 displays 5 sale lots and 30 wanted lists, but seller price and condition rows remain placeholders. Brickset is blocked.
Next test: The exact production count, item completeness, seller-level stock and comparable completed prices remain unverified. Marketplace interest counts are not completed purchases.
Fresh source attempts and ten trait judgments do not prove current executable inventory, sale price or condition.
See the underlying recordWicked Soundtrack 2025 RSD numbered glitter 2LP sources rechecked
The current retrieval of the official 2025 Record Store Day list identifies Wicked on two green/pink glitter LPs. It does not confirm the previously described numbered 2,000-copy US pressing; the direct Rough Trade request returned HTTP 403.
Next test: The exact region, numbering, pressing quantity and current seller evidence must be established before treating the UK list as proof of the US numbered edition.
Fresh source attempts and ten trait judgments do not prove current executable inventory, sale price or condition.
See the underlying recordOpen the complete outcome and learning record