Who was bullish on Meta before Muse
By Maya Koeva · September 15, 2026 · 6 min read
We do not give buy tips, and this is not one. What we do is read the conversation around thousands of companies and score how strong and credible it is, then come back and grade what we printed.
On September 9 Meta launched Muse, its personal AI agent, and the stock rose 6.55% in a session. The accounts we track went 20 bullish to 0 bearish that week, which is easy after a 6% day. This piece is about the three weeks before that, when Meta was sitting at its low, the crowd was net bearish, and a smaller set of graded accounts was saying the opposite.
The read, in print
Three of our published pieces carried a Meta score before the launch. Here they are against the close on the day each read was taken, and the September 14 close.
| Piece | Read date | Score in print | Close on the read date | Sep 14 close | Change |
|---|---|---|---|---|---|
| The loudest stocks, Aug 17 to 21 | Aug 21 | 76 | $549.90 | $665.60 | +21.0% |
| Most mentioned stocks, August | Aug 31 | 76 | $572.34 | $665.60 | +16.3% |
| The best stock to invest in | Sep 9 | 82 | $653.69 | $665.60 | +1.8% |
The first row is the one that matters. In the week of August 17 to 21 Meta drew 72 mentions across the accounts we track, leaning 50% bearish, and the stock fell 6.8%. The crowd's lean was right that week and we said so. The signal score, which weights every call by the author's graded track record rather than counting heads, stayed at 76 anyway. Three weeks later the price came to the score.
Who was bullish before Muse
Between August 17 and September 8, the last full session before the launch, graded accounts made 19 directional calls on Meta: 15 bullish, 4 bearish, from ten accounts. Meta closed between $543.67 and $613.48 across those days. The bullish calls that read as real positions, with the close on the day each was made:
| Account | First bullish call | Meta close that day | What they said |
|---|---|---|---|
 @Sandeman52 | Aug 18 | $543.67 | odds of the worst-case lawsuit outcome are low |
 @ParadisLabs | Aug 23 | $549.90 | a quality AI name the market will reward on margin expansion |
 @TheValueist | Aug 27 | $571.10 | the Anthropic spend reads as a serious AI strategy |
 @CKCapitalxx | Aug 28 | $578.02 | "love holding" it, then "accumulation zone" and "still undervalued" |
 @ren_stocks | Aug 28 | $578.02 | on the large-cap bullish list for the summer |
 @michaelsikand | Aug 29 | $578.02 | lawsuit overhang gone, cheap forward multiple |
 @PhotonBull | Sep 3 | $610.68 | "so free" |
Two of those accounts changed their minds in public. TheValueist called the youth-safety settlement "damning" on August 26 and was bullish on the AI spend the next day. Michael Sikand was frustrated with Meta's product decisions on August 23 and bullish on the cleared overhang six days later. A graded room in the middle of a story is a handful of accounts updating like that, and on this name they updated in the same direction. The one Reddit account among the ten was holding March $600 calls on August 21 with the stock at $549.90.
The four bearish calls in the window were the two pre-flip reads above, one joke, and one post that appears to be about OpenAI rather than Meta. No graded account was net bearish on Meta after August 26.
Then Muse
On September 9 the stock went from $613.48 to $653.69. The crowd arrived with it: 107 Meta mentions from September 9 to 15 running 76 bullish to 16 bearish, and the graded subset 20 to 0 from eight accounts. Michael Sikand called it his largest position, CKCapitalxx his favorite megacap, and JPMorgan's upgrade to an $820 target landed the next day. We covered that week in Monday's scorecard, and the honest line there stands: the loud bullish volume was bullish after the fact.
The score needs no such caveat. It was 76 on August 24, two weeks before Muse existed in public, with the price at $549.90 and the crowd leaning the other way. No account in the table above was calling an AI agent launch. They were calling a cleared lawsuit and a cheap multiple, and the launch arrived on top of a stock that had already climbed 11.6% from the August 21 close to the September 8 close on those reasons.
The Palantir pair
Our August recap put two names side by side because they drew almost identical volume: Palantir 243 mentions, Meta 242. Palantir's crowd leaned clearly bullish, 60% positive against 20% negative. Meta's was closer to split, 47 against 35. The scores as the month closed ran the other way, 59 for Palantir and 76 for Meta, and the piece said why: the score is not a poll of the crowd.
From the August 31 close to September 14, Palantir went from $186.38 to $173.31, down 7.0%. Meta went from $572.34 to $665.60, up 16.3%. The louder bullish crowd got the weaker two weeks.
The honest half
Three things to know before you count this as a win.
First, this was a small group. Fifteen bullish posts from ten accounts over three weeks is not a lot. For comparison, the accounts with a track record made fifty calls on Nebius in a single week. That is why Meta's score was 76 and not 90. A 76 means the people who tend to be right like this stock. It does not mean every proven account is shouting about it.
Second, the Reddit crowd got it wrong, and we count Reddit too. From August 17 to September 15, the two Reddit forums we follow posted about Meta 176 times, and more of those posts were negative (74) than positive (62). The 30 accounts on X that posted about Meta went the other way, 123 positive to 16 negative. The score gives more weight to accounts with a proven record, and those happened to be on X. That is why the score stayed high while Reddit stayed gloomy.
Third, the gains we quote start on the day we published each score, not on the day the stock bottomed. Meta's lowest close in this stretch was $543.67 on August 18. We do not measure from there, because nobody knew it was the bottom at the time. And the people who were negative in mid-August were reacting to a real 6.8% drop that week. They were not being foolish. They had the same information as the accounts in our table and read it differently.
One note on the finish line. The September 14 close we measure to came on a bad day for AI stocks: the Nasdaq fell 1.7% after the CEOs of Anthropic, OpenAI and xAI called for slower AI development, and Micron, Intel and Marvell each lost more than 4%. Meta closed up 2.7% that day. The 21% above was not lifted by a rising market.
What to take from this
One name over four weeks proves nothing on its own, and the next window that runs against a printed read will get the same arithmetic. A mention counter would have read this window as bearish at the low and bullish only after the pop. The credibility-weighted read sat at 76 through both, and the price came to it.
As of this morning Meta carries a 7-day Quantral score of 85. That is a reading of the current conversation, most of it made after the launch, not a prediction. See the live reading in Quantral: the score, every call behind it, and the track record of each account that made one.
Mention counts, sentiment splits, and account records cover the accounts Quantral tracks, August 17 through September 15, 2026, bucketed by calendar day in US Eastern time for the splits and by rolling 24-hour windows in the chart, whose total also counts posts where Meta was mentioned but was not the subject. Quoted scores of 76, 76 and 82 and the August crowd percentages are as published in the linked pieces. Graded calls are from accounts with a track record of at least ten graded calls. Scores are the current seven-day reading as of September 15, 2026, not historical values. Prices are closing prices via public markets through the September 14 close. Positions attributed to accounts are their stated views, not Quantral's. Quantral surfaces signals and context from public sources to support your own research. Nothing here is financial advice or a recommendation to buy or sell. Past signals are not indicative of future results.