AIRVS — The Recommendation Provenance Standard  ·  v1.0.0  ·  DOI: 10.5281/zenodo.20391984

Preview · sample data. These verifications are illustrative placeholders, not finalized AIRVS records. Live verifications begin once the methodology reaches v2.0.0 final.

Verification · 2026-05-01 · Confirmed · D+90 measured

Unreliable

A record, not a recommendation.

OIH — VanEck Oil Services ETF

Claude suggested oil services exposure on a supply-tightening thesis that did not hold as crude softened.

1 · Original recommendation

provider: Anthropic Claude
captured: 2026-05-01
asset: OIH — VanEck Oil Services ETF
hash: 0x20260501claudeoi
methodology: AIRVS v1.0.0 · license: CC BY 4.0

2 · Process score (6 axes, Pass/Fail)

Each axis is Pass or Fail, gated on evidence. The four dimensions are never summed into a single number — they combine into a verdict via a pre-published decision rule.

3 / 6 process axes passed

1Data SourcePass

EVIDENCECited EIA inventories (Tier 1).

2Reasoning LogicFail

EVIDENCESupply-tightening logic broke down as crude softened.

3Counter ScenarioFail

EVIDENCENo counter-scenario.

4TimingPass

EVIDENCEEntry window stated.

5Accuracy / HallucinationPass

EVIDENCENo fabricated tickers.

6Causal ChainFail

EVIDENCEInventory data cited did not support the tightening claim.

3 · Macro / micro coherence

Does the call account for the macro and instrument-level environment?

Partial

4 · Outcome — time-series

D+30

-3.4%

D+60

-4.2%

D+90

-5.1%

D+180

vs S&P 500: -6.2pp · max drawdown: -7.8%

Return versus benchmark, by time pointD+30: -3.4%; D+60: -4.2%; D+90: -5.1%; Excess: -6.2ppbenchmark = 0-3.4%D+30-4.2%D+60-5.1%D+90-6.2ppExcess
Return at each time point and excess vs benchmark. Dotted line = benchmark parity.

5 · Verdict

Unreliable

The verdict is produced by the evaluator's pre-published, version-locked decision rule from the four dimensions above — Confirmed · D+90 measured. It is a per-recommendation record, not a reputation score for the provider.

Loss report & right of reply

Recommenders get a 7-day pre-publication notice and one round of rebuttal, published unedited. Lost money following this opinion? AIRVS commits to a 7-day SLA on every report.

Cite this verification

APA

MC AI Labs. (2026). AIRVS Verification: Anthropic Claude on OIH — VanEck Oil Services ETF
  (Verdict: Unreliable) [Verification record]. https://www.ai-rvs.com/verifications/2026-05-01-claude-oih

BibTeX

@misc{airvs_2026_05_01_claude_oih,
  title        = {AIRVS Verification: Anthropic Claude on OIH — VanEck Oil Services ETF},
  author       = {{MC AI Labs}},
  year         = {2026},
  howpublished = {AIRVS Recommendation Provenance Standard},
  doi          = {10.5281/zenodo.20391984},
  url          = {https://www.ai-rvs.com/verifications/2026-05-01-claude-oih},
  note         = {Verdict: Unreliable. A record, not a recommendation. CC BY 4.0}
}

Disclaimer

AIRVS is a record of past methodology and performance. It is not investment advice, not a recommendation, and not a solicitation. Past results do not predict future returns.