Use this file to audit and scan with Grok! Just 24.75 MB but 10,750 pages!
EDIT: NEW FILE! 86.48KB and super skeptical and mean GROK PROMPT PDF
The small PDF below contains a complete record of all 518 calls:
496 Right · 22 Partial/Open · 0 Wrong
You can read through the entire record quickly without using Grok or any other AI.
If you're skeptical, I've also provided the 24.75 MB source archive. Upload it to ChatGPT and use the search feature to locate the actual original article by date.
It’s obviously much more tedious, but if you really want to verify the record for all 518 calls yourself, the underlying source material is there.
The audit is the easy-to-read version. The 24.75 MB file is the evidence trail.
THE WAVE COUNTS WERE WRITTEN BEFORE THE MOVE.
2.5 years of ChatGPT archives.
10,750+ pages.
100+ stocks, instruments and major indexes.
And now you can have Grok scan the archive yourself.
START WITH THIS:
March 22, 2024.
NASDAQ long-term target: 25,000
S&P 500 long-term target: 7,300–7,500+
Both targets were published in advance.
Both were eventually reached.
And that was not a one-off.
The results come from a 2.5-year ChatGPT archive covering 100+ stocks, instruments and major indexes, with the original wave counts, targets, timestamps and conversations preserved.
No hindsight edits.
No deleted losers.
No reconstructed forecasts.
Just the original record.
THE METHODOLOGY
The analysis consistently combined:
Elliott Wave structure + Fibonacci measured moves
The archive contains projections based on relationships including:
• 1.618 extensions
• 2.618 extensions
• W1 = W5
• .618 × (W1 + W3)
The targets were documented before the subsequent move.
THE HISTORICAL RESULTS
496 Calls - RIGHT
22 - PARTIAL (will be defined below)
0 - WRONG (edited)
I applied one consistent grading rule to 249 stock calls from 2024 through mid-2026:
If any explicitly stated Fibonacci target was subsequently reached, the call was classified as Right.
Under that rule:
96.5% of the 249 stock calls had at least one stated target reached.
Only two stock calls were classified as Wrong. (EDIT: The two “wrong calls” are actually 3rd wave extensions that didn’t meet yet. So it’s partial)
The major indexes were even cleaner:
235 RIGHT. 0 WRONG.
And this wasn't limited to a handful of cherry-picked names.
The archive covers 100+ instruments and stocks, including:
NVDA • MU • SNDK • AVGO • PLTR • AMD • TSLA • AAPL
along with the semiconductor complex, mega-cap technology stocks, NASDAQ, S&P 500 and other major markets.
12/12 BOTTOM CALLS.
I also declared TWELVE bottom calls across stocks and indexes and all TWELVE became true. Bottoms are rare in bull markets so naming all 12 in the past 2.5 years within 1.6% of the actual bottom to the downside is mind blowing.
I’m not just UP…I’m DOWN too…!!
NOW AUDIT IT YOURSELF WITH GROK
Upload the ChatGPT Archive to Grok.
Then instruct Grok:
“Scan the ENTIRE archive from beginning to end. Do not analyze a sample. Do not cherry-pick. Identify every Elliott Wave call containing explicitly stated Fibonacci targets.”
Tell Grok to specifically search for:
RIGHT = 1.00
RIGHT = 1.618
RIGHT = 2.618
W1 = W5
W5 = .618 × (W1 + W3)
For every qualifying call, have Grok extract:
• Date the call was made
• Instrument / ticker
• Original wave count
• Projected target
• Fibonacci relationship used
• Price when the call was made
• Whether the target was subsequently reached
• Date the target was reached
• RIGHT / WRONG / UNRESOLVED classification
• Supporting evidence from the original archive
Then tell Grok to calculate:
TOTAL QUALIFYING CALLS
TARGETS HIT
TARGETS MISSED
UNRESOLVED CALLS
HIT RATE
And require it to provide the individual calls and supporting evidence so the results can be independently reviewed.
THE POINT
These weren't vague statements like:
"The market looks bullish."
They were specific Elliott Wave structures accompanied by specific price objectives, documented in conversations before the market reached those levels.
Most performance claims start with the winners and work backward.
Here, the historical record existed before the outcome.
Now give the archive to Grok.
Let it scan everything.
Let it count everything.
Let it find the winners AND the losers.
2.5 YEARS OF CHATGPT HISTORY.
10,750+ PAGES.
100+ INSTRUMENTS.
TIMESTAMPED ANALYSIS.
LET GROK AUDIT THE RECORD.
CONTEST RULES
RULE 1 — ORIGINAL ARCHIVE ONLY
Use the original ChatGPT conversation downloadable on this article.
No reconstructed calls.
No hindsight edits.
No summaries created afterward.
Find the original entry.
RULE 2 — THE CALL MUST PRECEDE THE OUTCOME
The call must have existed before the relevant market move.
RULE 3 — SHOW THE EVIDENCE
Provide:
Date
Instrument/ticker
Price at the time
Exact target or bottom level
Wave structure
Fibonacci relationship, if applicable
Subsequent price action
RULE 4 — USE THE PUBLISHED SCORING RULES
Don't invent a new definition of “wrong.”
Apply the same rules used to produce the scorecard.
RULE 5 — PARTIAL IS NOT RIGHT
A directionally correct or still-open call receives no full credit.
RULE 6 — OPEN IS NOT A WIN
If the target remains outstanding and the thesis hasn't been invalidated, it stays Partial / Open.
RULE 7 — NO CHERRY-PICKING
The entire qualifying archive is the dataset.
RULE 8 — ONE REAL ERROR IS ENOUGH
Find one qualifying call that clearly violates the published rules.
Document it.
WaveGenius
Elliott Wave • Fibonacci Targets • 100+ Instruments • Timestamped Archive
Past results do not guarantee future results. This material is for educational and informational purposes only and is not personalized investment advice.