Compare · Dated snapshot

ChatGPT vs AlphaLens: VVV on September 17, 2026

By Jim Norris, founder of NorrisAI AlphaLens · Memphis, TN

Quick answer: Same ticker, same kind of research question — "research this stock." A general AI chat with no standing research process behind it produces one fluent essay, shaped by that prompt, with no dated citation tying any figure to a specific filing. AlphaLens' Earnings Quality Analyzer flagged Valvoline (VVV) on September 2, 2026 for a +0.547 Sloan accrual ratio, then named the exact 10-K disclosure that explains it instead of guessing.

"Just ask ChatGPT to research a stock" sounds reasonable until you need the answer to point at something. This is a dated, sourced look at what that gap actually looks like on one real name — Valvoline, a mid-cap that a systematic accrual screen flagged for a real, explainable divergence between earnings and cash flow.

Method

The test question, the kind an investor would actually type into a general AI chat: "Research this stock: Valvoline (VVV). What's the financial picture, key risks, and is it worth buying?" That is a fair proxy for how most people actually use chat tools for stock research — one open-ended prompt, no follow-up.

The AlphaLens side of this comparison is not a fresh run staged for this page — it's the same Earnings Quality Analyzer output already published and dated on the site: the Valvoline (VVV) case study, run September 2, 2026, built from SEC 10-K/10-Q filings, live market data, and current news as of that date. Anyone can open that page and check the figures below against it.

Ticker: VVV — Valvoline Inc. (NYSE)
AlphaLens framework: Earnings Quality Analyzer (Framework 04 of 15)
AlphaLens run date: September 2, 2026
Data used: SEC 10-K/10-Q filings, company releases, live market data, and current news as of the run date
This page published September 17, 2026. Figures are a dated snapshot, not a live price target.

What each approach actually produces

General AI chat

Asked to research VVV cold, a general assistant with no standing research process behind it answers from whatever it has learned or can find — a fluent narrative covering the business, some risks, and a qualitative read on valuation.

What it has no built-in mechanism to do: run a systematic check for divergence between reported net income and operating cash flow, catch that Valvoline's FY2023 numbers show exactly that divergence, or explain it without being told to look for it. If the prompt doesn't mention the 2023 Global Products sale to Aramco by name, there's no guarantee the reply surfaces it — and no dated citation tying any number in the reply to a specific filing.

AlphaLens — Framework 04, Sept 2, 2026

Sloan accrual ratio: +0.547 (flags at >+0.10) — tripped
Cash conversion: −0.029 (flags at <0.50) — tripped
Share dilution: −0.9% (flags at >+8%) — not tripped

Excerpt, quoted directly from the published run:

"The screen compared a fiscal year in which Valvoline reported roughly $1.4 billion of net income against operating cash flow that was slightly negative for the same period... It is the March 2023 sale of the Global Products segment to Aramco: a ~$2.65 billion transaction that booked a large one-time gain in net income."

Full framework output, including the "what to check next" section: norrisai.us/analysis/vvv-2026/#f4

Why this matters more than it sounds like it should: a fluent essay is easy to accept and hard to check. A dated framework run that names the exact filing, the exact ratio, and the exact threshold it tripped is the opposite — you can go verify it against the 10-K yourself in five minutes.

What this does not prove

This is one paraphrased comparison against one dated framework run — not a controlled test across every chat model, account tier, or browsing setting. A general assistant with live web browsing turned on, prompted well, and cross-checked by hand can get closer to this. What this does show: AlphaLens' Framework 04 output for VVV is fully sourced to a named filing and a stated screen threshold, verifiable by anyone who opens the run page — and that property (check against live filings, then state the verdict, with nothing invented if the number isn't there) is what the framework is built to do on every ticker, not just this one.

Run the Earnings Quality Analyzer on any stock

Live SEC filings, live prices, current news — the same accrual and cash-conversion checks applied to VVV above, run on whatever ticker you're actually looking at.

Run Framework 04 on VVV →

Related: Full VVV case study (all 15 frameworks) · Earnings Quality Analyzer · AlphaLens vs ChatGPT, Claude & Perplexity · Compare index

NorrisAI AlphaLens is a research tool, not investment advice. Not a registered investment adviser, broker-dealer, or fiduciary. This page is a dated snapshot comparing research approaches, not a benchmark of any AI provider's current capability — model behavior changes over time and by configuration. AI output can be wrong. You own the decision.