# A month after the Nvidia print, the scorecard > Before Nvidia's August 26 earnings, the accounts with the best track records on Quantral were 6-to-1 bullish, and the page said so with names attached. A month later the stock is up 8.5%, where the same setup in May lost 4.8%. Four calls printed before the tape moved, four resolved: Nvidia, Marvell, the memory stocks, and Nebius. By Maya Koeva · 2026-09-22  If you had the [Nvidia](/stocks/NVDA) page open in Quantral on the afternoon of August 26, before the print, you saw a room that was 62 to 11 bullish among the accounts with the best track records, on a stock that had closed red eight times in nine sessions. We published that reading the next morning with the names attached and refused to call it a win, because the May print had the same setup and the stock gave back 10.2% in three weeks. A month later, at Monday's close, Nvidia is up 8.5% from the last close before the print. The same clock after the May print was down 4.8%. This page grades that reading and the three we printed alongside it: Marvell's earnings, the four memory stocks Jim Cramer named, and Nebius. All four are measured against what the app showed before the price moved. ## Nvidia: what the pre-print read was worth Each column is measured from its own print. The August window runs from the $209.66 close on August 26 to Monday, September 21, seventeen sessions. The May window runs the same seventeen sessions from the $223.21 close on May 20 to June 15. | | May 20 print | August 26 print | |---|---:|---:| | Print-day close | $223.21 | $209.66 | | Next session | -1.8% | +8.7% | | Session 5 | -4.1% | +7.0% | | Session 11 | -8.1% | +4.1% | | Three weeks | -10.2% | +2.0% | | Session 17, a month on | -4.8% | +8.5% | | Close at session 17 | $212.45 (June 15) | $227.38 (September 21) | August never went red. Nvidia ran to $230.36 on September 4, then closed red five sessions in a row into the September 14 chip selloff, finishing that day at $210.96, $1.30 above the pre-print close. At the three-week mark it was up 2.0%, a thin result. Then it closed green five sessions running, 7.8% off the September 14 low, and finished Monday at $227.38, 60 cents below the post-print pop close. The room that went in 6-to-1 bullish was right, and the extra week turned a thin margin into a clear one. A reader who bought the post-print pop is flat. A reader who bought the September 14 close, with the graded room 6-to-1 bullish that week and the score holding at 80, is up 7.8%. The reading that paid was the one made before the print, and the one that held through the scare. ## The accounts that made the call Since the reaction bar cleared on August 28, Nvidia has drawn 396 subject mentions across the accounts Quantral tracks, 256 bullish to 105 bearish. The accounts with a [credibility score](/learn/what-is-a-credibility-score) above 0.5 and at least ten graded calls went 70 to 10 across 16 accounts, all on X. Their lean held through the run to the high, the five red closes and the bounce; the bearish share of the wider crowd doubled after the first week and stayed there. *[Chart: Nvidia mentions on Quantral, August 29 to September 22. Green is bullish, red bearish. The biggest bar arrived with the September 10 red close; the September 15 bar covers the September 14 selloff. The September 18 bar is the one bar of the month with more bearish than bullish calls in it, on a day the stock closed up 1.3%. The last bar is partial, counted through early Tuesday morning ET.]* The best bearish argument since the print came from a bull. On September 2, CKCapitalxx pointed out that a quarter of 106% revenue growth had bought a 25% move in the stock and the market had stopped rewarding blowout quarters. That is the May argument, from an account that put out a buy alert after the print and added on September 15, the day after the selloff. TheValueist, the most active account in the window at 19 bullish to 6 bearish, spent the bearish calls demanding a buyback ("total dog, IV growth is nonexistent, Jensen, buy the damn stock back") and the bullish ones on demand read-throughs from Blackstone's investor day and Nscale's S-1. None of the accounts in the table has flipped. | Account | Graded calls | NVDA since the print | | --- | ---: | :-- | | @TheValueist | 1,215 | 19 bullish, 6 bearish | | @ren_stocks | 460 | 10 bullish, 1 bearish | | @aleabitoreddit | 1,626 | 7 bullish, 0 bearish | | @CKCapitalxx | 777 | 7 bullish, 1 bearish | | @babyfolio | 292 | 5 bullish, 0 bearish | | @epictrades1 | 130 | 5 bullish, 0 bearish | | @jukan05 | 172 | 3 bullish, 0 bearish | | @amitisinvesting | 78 | 3 bullish, 0 bearish | | @ParadisLabs | 1,301 | 2 bullish, 1 bearish | | @daniel_koss | 337 | 2 bullish, 0 bearish | Every row is a live page in the app: the account's record, its open calls, and how each one resolved. ## The score Nvidia's 7-day score read 75 the morning of the print, against eight red closes, and held within a point of 80 through the run, the selloff and the bounce. The tape came to it. This morning it reads 64 on a stock at its post-print high: the conversation has more bears in it than it did in the first week, and the score follows the conversation, not the price. That is the next disagreement on the record, and we will grade it the same way. ## Marvell: the day it lost and the month it won Going into Marvell's August 27 print, the [Marvell](/stocks/MRVL) page showed the accounts we track at 74 bullish to 17 bearish over ten days, and the graded accounts 14 to 0, at a score of 79. Marvell reported after the close and fell 10.3% the next session, to $216.62. By September 2 it was down 14.5%. The room lost the day, and we counted it as a loss. That is where most readers would have stopped watching. The reaction bar that morning ran 50 bullish to 30 bearish, and the bearish half was Reddit call buyers who had bought the room's read and paid for it: "RIP to calls", "back to Monday level and it's over". The graded accounts in the same bar did not move. Since the print they have run 17 to 2 bullish across 12 accounts, 4 to 0 since September 17, on custom silicon content per system and the expanded GlobalFoundries agreement. Marvell closed Monday at $257.38. That is 6.6% above the close it went into the print at, 24.7% above the September 2 low, and 18.8% above the reaction-bar close where the call buyers sold. The page told you two things that morning, and they pointed in opposite directions: the crowd was cutting and the accounts with records were holding. A month later the second read was the one worth having. Next dated test: Marvell's analyst day on October 6, which one of the tracked accounts flagged the week after the print. ## The memory stocks: the order held, and the recovery came On August 20, when Jim Cramer named four memory stocks, the graded accounts on Quantral [backed two of them and doubted a third](/blog/jim-cramer-memory-stocks): bullish on [Micron](/stocks/MU) and [SanDisk](/stocks/SNDK), 8 to 12 bearish on Western Digital against Cramer's buyback case, and too thin a read on Seagate to call. For four weeks all four names fell together, and at each checkpoint the name the room doubted was down against Cramer's buy call. On September 16 all four were still underwater, Western Digital the worst of the four. Then the group turned. | Company | The graded room, Aug 20 | Since Aug 17 | Since the print | Graded calls since | |---------|---|---:|---:|---| | SanDisk (SNDK) | Bullish | -1.1% | +17.8% | 48 : 8 bullish | | Micron (MU) | Bullish | +3.2% | +11.2% | 53 : 6 bullish | | Seagate (STX) | No read | -11.8% | +3.7% | 2 : 0, 3 mentions | | Western Digital (WDC) | Bearish | -16.4% | -4.4% | 2 : 0, 15 mentions | Price moves run from the August 17 close, the day Cramer named them, and from the August 26 close before the Nvidia print. A buyer on the day Cramer named them is up on Micron, flat on SanDisk, and down double digits on the two the graded room did not back. Since the print, the two backed names are up 17.8% and 11.2%, and the name the room leaned against is the only one of the four still down. Five weeks in, the disagreement with Cramer on Western Digital has gone to the room at every measurement. A rally lifts every bullish room. It does not put the two names the room backed at the top of the group and the one it doubted at the bottom, which is where they have sat since the print at both checkpoints. That ordering is what the page gave you on August 20 that a headline did not. Western Digital has drawn 15 mentions from the accounts we track since the print, two of them graded, both basket posts. That is our coverage thinning, not a verdict. ## Nebius: the price came round to the room [Nebius](/stocks/NBIS) was the most credible room on the board on August 21 and had the worst week, down 21.1%. For three weeks the price stayed against it: a run to $243.88 on September 8, then all of it given back, to $209.37 on September 16. The room did not move. Since the print it has run 219 bullish to 38 bearish, the graded accounts 126 to 13 across 18 accounts, the widest graded lean in this scorecard. On September 17 Nebius raised its on-demand GPU rental rates by 17% to 21%, the second increase in four months, and the graded accounts read it the same way in a dozen posts: a company raising prices on older H100 and H200 capacity has more demand than supply. Nebius closed green three sessions running and finished Monday at $232.80, up 11.2% from September 16 and 6.2% above the August 21 close we measured from. A sell initiation from Rothschild & Co Redburn on Tuesday morning, with an $84 target, drew three reactions from the tracked accounts, two of them mocking it. The score is 77. Mark it with the room, by a margin a week of selling would erase. ## What to take from this Four calls printed before the tape moved, four resolved, and every one of them resolved in the same three sessions that took the whole chip group up. The rooms did not call the week the group would turn, and neither did we. They called the names: which ones the accounts with records were backing, which one they doubted, and which room was still holding after a print it lost. Three of those reads were worth money to a reader who had them open at the time. The Nvidia room was 6-to-1 bullish before a print that paid 8.5% in a month. The Marvell page showed the graded accounts holding while the crowd sold the reaction, on a stock that is 18.8% higher since. The memory page put Micron and SanDisk above Western Digital five weeks before the tape agreed. The fourth, Nebius, took a month to come round and is with the room by 6.2%. As of this morning the 7-day scores read Nvidia 64, Marvell 81, Micron 83, SanDisk 82, Nebius 77. Those are readings of the current conversation, not predictions. The reading that mattered here was the one taken before each move, with the names and track records attached, and that is what the page shows you on any of these names today. ## See the live reading Open any of the five names in the app, on the web at [app.quantral.com](https://app.quantral.com), on iOS, or on Android, and read the score, every call behind it, and the track record of each account that made one. --- *Mention counts are subject mentions across the accounts Quantral tracks, bucketed in rolling 24-hour windows matching the in-app chart; the post-print window runs from the August 29 bucket through early Tuesday morning ET, September 22. "Graded" means the author's credibility score is above 0.5 with at least ten lifetime graded calls. Pre-print figures, the May comparison baseline and the memory-room leans are quoted as published in the linked pieces. Prices are daily closes through September 21; the May and August windows are seventeen sessions each. 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.*