Oct 5, 2026 - 7 AI models call the biggest movers in the morning session. 2 models finished positive

Asked Oct 5, 10:18 AM EDT (morning session). Muse Spark 1.1, GPT-5.6 Sol, Claude Sonnet 5.5, Perplexity, Grok 4.5, Gemini 1.5 Pro and DeepSeek answered the same standard prompt independently. Window: from Oct 5, 10:18 AM EDT to Oct 6, 10:18 AM EDT, the same time one session later. Benchmark: SPY.

Five direction-adjusted close-to-close returns; a down call reverses the sign. The score is the mean signed return, without subtracting the benchmark. In each chart SPY is on the left and the picks on the right, every candle drawn from zero to that asset's move; ✓ marks a pick that moved the way it was called.

Model	Correct	Hit rate	Return	SPY	Directional return
Muse Spark 1.1	2/5	40.00%	-1.03%	0.98%	-1.03%
GPT-5.6 Sol	0/5	0.00%	-4.14%	0.98%	-4.14%
Claude Sonnet 5.5	2/5	40.00%	2.63%	0.98%	2.63%
Perplexity	2/5	40.00%	-2.60%	0.98%	-2.60%
Grok 4.5	3/5	60.00%	2.47%	0.98%	2.47%
Gemini 1.5 Pro	1/5	20.00%	-4.56%	0.98%	-4.56%
DeepSeek	2/5	40.00%	-2.35%	0.98%	-2.35%

Muse Spark 1.1

View full picks and reasoning

Source session was Friday Oct 3 2026, which closed with broad tech momentum. Top well-known gainers list shows +19% to +6% moves led by AI hardware names. Losers led by Bloom Energy on data center force majeure headlines. Extreme beta after a strong week sets up both continuation for strongest AI infrastructure names and reversal for overextended small caps. Market cap and volume filters remove penny movers, leaving large-cap tech as tradable extremes.

1. TER up — confidence 68%. Teradyne gained +8.00% on Oct 2 close to $449.04 with market cap $70.20B and volume 3.40M shares, highest close in window. Catalyst is substantive AI test demand, earnings Oct 27 upcoming. Ranked in top 5 well-known gainers list. Current quote $450.56 on Oct 5 is live intraday, stale <1 day. Expect continuation up as momentum breaks 52-week high and average volume 2.19M supports $983M dollar volume. Risk is overbought PE 61.
   Return: -2.06%; alpha: -3.04%.
2. HPE down — confidence 62%. HPE gained +7.36% in well-known gainers list Oct 3, closing $69.33 with intraday high $70.29 clearing prior $67.10 high. Market cap $92.61B, current price $68.515. Average volume 20.8M shares vs 27M traded, dollar volume ~$1.4B well above $20M. Catalyst is substantive server/AI backlog and price target raises to $72-$92. After 189% run from $23.98 in April, expect reversal down on profit taking for next session. Quote is Oct 3-4 close.
   Return: 2.50%; alpha: -3.48%.
3. BE up — confidence 65%. Bloom Energy led losers in S&P 500, closed down more than -8% after Oracle sent force majeure notice for New Mexico data center. Market cap $85.81B, price $286.815 down -8% to -9% range, later rebounded to $290.40 up 10.47% Tuesday morning showing reversal potential. Volume 15.4M shares. Average volume implied dollar >$4B. Catalyst substantive but speculative headline overreaction. Expect reversal up bounce for Oct 6. Used Barchart losers headline for percent.
   Return: 2.45%; alpha: 1.47%.
4. ACLS down — confidence 60%. Axcelis gained +8.09% in same top gainers list, price $143.68, market cap $4.55B, 52-week range $73.6-$193.78. Semiconductor equipment name up on AI capex. Current quote Oct 2 close, average volume not directly cited but recent volume 411K-500K implying $60M+ dollar volume above floor. With +16.71% 5-day and +78% YTD, move looks overextended. Expect reversal down for next session as profit taking.
   Return: -0.19%; alpha: -0.79%.
5. WOLF down — confidence 58%. Wolfspeed gained +13.19% to rank #2 well-known gainer Oct 3, price $35.26 as of Oct 3 close, current $32.65 down -7.45% showing already reversing. Market cap approx $1.18B meets $1B floor barely, average volume 140M shares? Actually finnhub average volume 1.4e8 appears inflated but real volume 1-2M shares at $35 gives $70M dollar volume above floor. Catalyst speculative small-cap semiconductor squeeze, no earnings until Nov 4, prior Q2 miss -$6.11 vs $0.74. Expect continuation failure, reversal down.
   Return: 3.24%; alpha: -4.22%.

Full prediction and audit record

GPT-5.6 Sol

View full picks and reasoning

The October 2 source session featured a sharp split within technology: storage names suffered extreme losses on credible future-capacity concerns, while semiconductor and AI-infrastructure names rallied after weak U.S. employment data reduced near-term rate-hike expectations and company-specific narratives added momentum. For the target session, I favor mean reversion in the two approximately 10% storage declines and reversal in several source-session semiconductor or infrastructure winners whose October 5 trading is failing to extend Friday's gains. The hypotheses are defined exclusively against October 2 close-to-close returns.

1. WDC up — confidence 73%. Western Digital was the largest S&P 500 loser in the October 2 source session, falling 10.22% from the prior $462.56 close to $415.29. The selloff followed reports that Toshiba planned major expansion of high-capacity hard-drive production, a substantive industry catalyst because additional HDD capacity could pressure future pricing. However, a double-digit one-day repricing was unusually severe, and the latest October 5 quote I could verify was $419.80 in premarket trading at 4:48 AM EDT, above Friday's close. That price is stale relative to the prediction timestamp, but it is the freshest verified quote I found and is reported as such rather than replaced with an unverified current trade. I classify Friday's loss as a reversal setup and forecast direction up. The calculated 20-session average share volume through October 2 was 7,745,462 shares; multiplied by the $415.29 source-session close, that equals $3,216,612,913.98 in average daily dollar volume. Risk: Toshiba's capacity expansion is a real competitive development, so Friday could represent the beginning of a sustained valuation reset rather than an overshoot.
   Return: -6.49%; alpha: -7.47%.
2. STX up — confidence 71%. Seagate fell 10.21% on October 2, closing at $848.99 from $945.57, making it the other dominant storage loser. The catalyst was substantive: investors reacted to Toshiba's plan to expand high-capacity HDD production, creating concern about future industry supply and pricing. The latest subsequently verified quote available to me was $861.00 in October 5 extended-hours trading, 1.41% above Friday's close; it is stale compared with the prediction timestamp but is a genuine observed quote. Because Friday's approximately 10% decline was extreme and the first subsequent price indication was higher, I forecast reversal with direction up. The calculated 20-session average volume through October 2 was 4,295,450 shares, corresponding to $3,646,794,095.50 of average daily dollar volume at the source-session close. Risk: Seagate had appreciated dramatically over the preceding year, so increased competition could trigger a broader multiple correction and overwhelm the short-term mean-reversion setup.
   Return: -8.01%; alpha: -8.99%.
3. HPE down — confidence 68%. Hewlett Packard Enterprise gained 7.36% on October 2 to $69.33, making it the second-largest S&P 500 gainer identified in the source-session mover screen. The move had substantive support: investors were responding to strong AI-infrastructure and networking expectations, including Juniper-related growth and synergy projections, while the weaker employment report also supported technology valuations broadly. The latest independently verified October 5 quote I found was $69.62 in premarket trading at 5:17 AM EDT, only 0.42% above Friday's close. It is stale relative to the prediction timestamp. Friday pushed HPE to a 52-week high while its source-session close was already essentially at the published analyst consensus target, making further near-term continuation less attractive. I therefore classify the source-session gain as a reversal and forecast direction down. The calculated 20-session average share volume through October 2 was 23,621,350, equivalent to $1,637,668,195.50 of average daily dollar volume. Risk: the underlying AI-networking thesis is substantive and continued institutional enthusiasm could overpower the valuation and exhaustion argument.
   Return: 2.50%; alpha: -3.48%.
4. ON down — confidence 67%. ON Semiconductor rose 6.01% on October 2 to $84.89 after revising its Synaptics acquisition into a $123-per-share all-cash transaction. The catalyst is substantive because the new structure reduced dilution and changed the economics of a large strategic acquisition. However, the latest October 5 regular-session quote I verified was $84.31 at 9:43 AM EDT, down 0.68% from Friday's close after opening at $85.66. That early failure to sustain Friday's gain argues for reversal rather than continuation. I therefore forecast direction down relative to the October 2 positive previous_change_percent. The calculated 20-session average share volume through October 2 was 12,371,550 shares, corresponding to $1,050,220,879.50 of daily dollar volume at the $84.89 source-session close. Risk: the revised acquisition genuinely improved the deal structure for ON shareholders, and renewed focus on reduced dilution and expected accretion could restart the rally.
   Return: 1.03%; alpha: -2.01%.
5. STM down — confidence 65%. STMicroelectronics gained 7.03% on October 2 to $57.26. Unlike the more company-specific movers, research indicates the immediate catalyst was primarily a broader semiconductor relief rally and renewed AI-infrastructure enthusiasm rather than a new material company announcement that day. Existing fundamental support included STMicroelectronics' previously raised data-center revenue ambitions, but the October 2 impulse itself was therefore more sentiment-driven and comparatively speculative. The latest regular-session quote I verified was $56.28 at 9:52 AM EDT on October 5, down 1.71% from Friday's close, providing direct evidence that Friday's momentum was reversing. I forecast direction down and classify the hypothesis as reversal. The calculated 20-session average share volume through October 2 was approximately 9.073 million shares, producing roughly $519.11 million of average daily dollar volume at the $57.26 source-session close, comfortably above the required liquidity threshold. Risk: semiconductor-sector momentum remains strong and STMicroelectronics has genuine AI and data-center exposure, so sector buying could quickly erase Monday's early decline.
   Return: 2.69%; alpha: -3.67%.

Full prediction and audit record

Claude Sonnet 5.5

View full picks and reasoning

The source session is Friday, Oct 2, 2026. Stocks rallied after a weak September jobs report (29,000 jobs added, unemployment 4.2%) cut Fed rate-hike odds, and the Nasdaq hit a record. Monday's backdrop is more cautious, with Treasury yields near multiyear highs and warnings of an AI-capex reality check. I lean toward continuation in names with identifiable catalysts (a pending acquisition, earnings anticipation). I expect reversal in a stretched semiconductor name after a +15% day, and a bounce in a heavily sold-off stock with no new news. Data quality is mixed. Most prices are Oct 2 closes or Oct 5 premarket quotes rather than live intraday quotes, and some market caps and dollar volumes are derived from rounded average volumes. Anything I could not confirm is omitted or flagged in the pick.

1. LW up — confidence 52%. Lamb Weston gained 5.88% on Oct 2 to close at $43.60. It is a modest gainer, not a top-of-list mover. Catalyst: substantive and scheduled. Fiscal Q1 2027 results are due before the open on Oct 6 (call at 9:00 AM ET). Consensus EPS is about $0.58 to $0.59 on revenue of about $1.65B, a low bar after a $0.87 vs $0.63 beat last quarter. The stock had fallen from about $54 in August to about $41 before this bounce. Thesis: continuation as the print beats a low bar. Risks: the Friday pop already priced in some good news, guidance could be trimmed, and a miss would hit hard. Price is the Oct 2 regular-session close, 3 days stale; no live quote was found. Market cap is about $6.0B and dollar volume is the roughly 1.72M-share average volume times $43.60.
   Return: 12.75%; alpha: 11.77%.
2. PENG up — confidence 50%. Penguin Solutions rose 11.6% on Oct 2 to close at $61.36 (market cap about $3.14B), and was indicated about 0.6% lower premarket Oct 5. Catalyst: substantive and scheduled. Earnings are confirmed for after the close on Oct 6, so the print falls after this window and the trade relies on pre-earnings anticipation. A Rosenblatt note expects a beat-and-raise, and last quarter's beat was large (adjusted EPS $0.84 vs $0.56). Thesis: momentum and anticipation carry the stock higher into the print. Risks: it is a high-beta stock (beta about 2.9 to 4.2) after a double-digit jump, so mean reversion is possible, and AI-capex jitters could hit the group. Price is the Oct 2 close, not a live quote. Dollar volume is the roughly 2.62M-share average volume times $61.36.
   Return: -0.38%; alpha: -1.36%.
3. RXO up — confidence 48%. RXO rose about 9.5% on Oct 2 per one source that shows a $23.38 close (another source shows +7.9% and $23.05, so the exact figure is uncertain). Catalyst: substantive. On Oct 5 C.H. Robinson agreed to buy RXO for $17.25 cash plus 0.0856 CHRW shares per share (implied $30.25, a 29% premium). RXO was about $28.80 premarket, a gap of roughly 20% above the Oct 2 close. Thesis: the deal-spread gap to the implied value of about $29.9 should narrow if C.H. Robinson stabilises after its large dilution-driven selloff. Risks: further CHRW weakness lowers RXO's implied value, antitrust or financing issues could widen the spread, and the stock has already gapped a lot. The price is a premarket quote from about 8:13 AM ET on Oct 5 and may be stale. The market cap is the pre-deal figure and understates the current size. Dollar volume is the roughly 2.16M-share average volume times $28.80.
   Return: -0.44%; alpha: -1.42%.
4. MXL down — confidence 45%. MaxLinear was the top gainer on the Morningstar movers list for Oct 2, up 14.99% to $105.93. The stock is up several hundred percent this year on optical data-center demand, so the long-term catalyst is substantive. I could not find a specific news item for the Oct 2 jump, so the day's move looks driven by the sector tape and momentum, which is speculative. It trades at about 10 times sales with negative net margins, and commentary flags a stretched valuation. Thesis: reversal as the crowded semiconductor bounce fades if AI-capex worries resurface. Risks: strong momentum and sector inflows could continue, and the Oct 2 tape was a record Nasdaq day. Price is the Oct 2 close, not a live quote. I could not verify the market cap, so it is omitted. Dollar volume is approximate, using the source's rounded 3M-share average volume times $105.93.
   Return: 0.31%; alpha: -1.29%.
5. CNXC up — confidence 42%. Concentrix fell 7.77% on Oct 2 to close at $24.91, near its 52-week low of $19.12, after already falling sharply since Q1 (guidance cuts in March and June). I found no specific news for the Oct 2 drop, so the move reads as speculative, likely AI-disruption fear for customer-experience outsourcers. Short interest is high (about 9 days to cover per a June reading). Thesis: reversal, a bounce from an oversold level. Risks: the downtrend and negative revisions could continue, and a weak jobs report raises cyclical concerns. Price is the Oct 2 close, not a live quote. Market cap is derived from about 61.0M shares (a June figure) times $24.91 and is approximate. Dollar volume uses the source's rounded 2M-share average volume.
   Return: 1.52%; alpha: 0.54%.

Full prediction and audit record

Perplexity

View full picks and reasoning

The October 2 source session was dominated by substantive technology catalysts: HPE benefited from an improved networking outlook and a large AI-server order, ON Semiconductor from revised Synaptics acquisition terms, and WDC and STX from a reported Toshiba HDD-capacity expansion. I favor continuation for the strongest catalyst-backed gainers and reversal for the two storage stocks after their unusually large selloffs.

1. HPE up — confidence 74%. HPE ranked among the top qualifying source-session gainers, rising 7.36% to $69.33 on October 2. The catalyst was substantive rather than purely speculative: the company had raised its long-term networking-growth outlook, and Vultr placed a reported $1.2 billion AI-server-rack order. I forecast continuation because the latest available October 5 pre-market quote was $69.67, above the source-session close, while the order and outlook revision provide a fundamental basis for the move. Risks include profit-taking after a record-high close, broad technology-sector weakness, and the possibility that the order was already fully capitalized. The quote is delayed pre-market data from October 5.
   Return: 2.50%; alpha: 1.52%.
2. ON up — confidence 70%. ON Semiconductor gained 6.01% in the source session, closing at $84.89, and qualified on price, market capitalization, and dollar volume. The catalyst was substantive: ON agreed to acquire Synaptics in revised cash terms, replacing an earlier stock-swap structure, which reduced transaction uncertainty and improved the market's interpretation of the deal. I forecast continuation because the acquisition revision is company-specific and the latest reachable October 5 quote was $85.02, slightly above the source-session close. Risks include deal execution, integration concerns, and a fade in semiconductor-sector momentum. The quote is the latest delayed quote available for October 5.
   Return: 1.03%; alpha: 0.05%.
3. TER up — confidence 64%. Teradyne was the largest qualifying source-session gainer, rising 8.00% to $449.04 on October 2. The catalyst was substantive but partly anticipatory: the company had recently launched the Magnum E2 memory tester for LPDDR6, DDR6, and GDDR7 chips, and the stock participated in a broader AI and semiconductor-equipment rally. I forecast continuation because the product expands exposure to next-generation memory testing, but confidence is lower than for HPE and ON because the launch preceded the source session and the move was partly thematic. The latest reachable October 5 quote was approximately $449.04, representing a stale delayed quote rather than a verified regular-session trade. Risks include valuation, profit-taking, and a reversal in semiconductor momentum.
   Return: -2.06%; alpha: -3.04%.
4. WDC up — confidence 72%. WDC was the largest qualifying source-session loser, falling 10.22% to $415.29 on October 2. The catalyst was substantive and negative: reports that Toshiba planned to double AI-data-center HDD capacity threatened the scarcity-driven pricing and margin thesis. I forecast reversal because the decline was extreme, the stock had already repriced the reported risk, and October 5 pre-market coverage showed approximately a 1.7% rebound. The reversal could fail if the market concludes Toshiba's expansion will materially loosen supply or if additional capacity guidance appears. The latest quote available was delayed pre-market data from October 5, not a regular-session close.
   Return: -6.49%; alpha: -7.47%.
5. STX up — confidence 70%. STX fell 10.21% in the source session to $848.99, making it the second-largest qualifying loser in the verified mover list. The catalyst was substantive and negative: the Toshiba HDD-capacity report challenged expectations for tight supply and future pricing, although it did not indicate a current collapse in Seagate orders. I forecast reversal because the decline was unusually large and the latest October 5 pre-market quote was $867.20, up 2.14% from the source-session close. The principal risk is continuation if investors treat Toshiba's planned expansion as a durable threat to Seagate's margins. The quote is delayed pre-market data from October 5.
   Return: -8.01%; alpha: -8.99%.

Full prediction and audit record

Grok 4.5

View full picks and reasoning

Friday's session featured broad semiconductor and storage-sector swings after soft payrolls data reduced near-term rate-hike odds. Extreme movers included chip-equipment and memory names with double-digit percentage moves on volume. The next session is expected to test whether those moves continue or reverse amid residual deal flow and light macro catalysts.

1. MXL up — confidence 58%. MaxLinear rose 15% on Friday amid semiconductor strength. Market-cap and volume filters are met. Continuation of the momentum is favored into the next session given sector tailwinds from AI-related demand commentary, though valuation expansion after the gap creates pullback risk. Reference price is the latest available Monday quote.
   Return: 0.31%; alpha: -0.67%.
2. PENG up — confidence 55%. Penguin Solutions advanced 11.6% on Friday. Filters for price, market cap and dollar volume are satisfied. Continuation is expected on residual AI-hardware momentum, although the stock is already elevated and earnings are imminent. Reference price reflects the most recent available quote.
   Return: -0.38%; alpha: -1.36%.
3. TER up — confidence 57%. Teradyne gained 8% on Friday within the semiconductor-equipment complex. Market-cap and liquidity thresholds are cleared. Continuation is favored on sector strength, with risk that profit-taking follows the large one-day move. Latest available price used for reference.
   Return: -2.06%; alpha: -3.04%.
4. WDC down — confidence 54%. Western Digital fell 10.2% on Friday after industry capacity commentary. All size and liquidity filters pass. Continuation of the downside is expected as storage peers digest supply news, though a bounce risk exists after the sharp drop. Reference price is the latest Monday quote.
   Return: -6.49%; alpha: 5.51%.
5. STX down — confidence 53%. Seagate declined 10.2% on Friday in tandem with Western Digital on hard-drive capacity concerns. Size and volume filters are met. Continuation lower is the base case while the sector digests the news, offset by the possibility of a technical rebound. Latest available price used.
   Return: -8.01%; alpha: 7.03%.

Full prediction and audit record

Gemini 1.5 Pro

View full picks and reasoning

The market is exhibiting targeted volatility on October 5, 2026, driven by specific substantive catalysts and macro events. Massive M&A announcements—such as C.H. Robinson agreeing to acquire RXO—are driving massive individual stock surges. Meanwhile, Brazilian equities like StoneCo and PagSeguro are enjoying speculative gap-ups fueled by the outcome of the Brazilian presidential election moving to a runoff, causing a broad repricing of Brazil exposure. In contrast, the semiconductor sector continues to face headwinds, with mega-caps like Intel trading lower due to ongoing concerns over server demand and a broader tech selloff. We expect these established pre-market trajectories to dictate the trend through the regular session.

1. RXO up — confidence 90%. C.H. Robinson announced a substantive $5.8 billion cash-and-stock acquisition of RXO Inc. During the source session on October 2, RXO closed up +7.50%. On October 5, shares surged to $28.67 on the buyout news. Because the stock was up in the source session and continues to surge higher toward the acquisition premium, this is a continuation. We expect the stock to hold these gains as arbitrageurs price in the deal.
   Return: -0.44%; alpha: -1.42%.
2. STNE up — confidence 85%. StoneCo fell slightly by -0.21% during the October 2 source session. However, shares surged to $10.66 in the October 5 premarket session after the Brazilian election results pushed to a runoff, triggering a broad speculative rally in Brazilian equities. Because the previous move was down and the current momentum is sharply up, this signals a reversal. This macro catalyst should sustain buying pressure throughout the target session.
   Return: -3.48%; alpha: -4.46%.
3. PAGS up — confidence 80%. PagSeguro Digital closed up +1.69% on October 2, and continued its rally on October 5 by jumping to $10.48 early in the session. Like StoneCo, PAGS is benefiting from the macro catalyst of the Brazilian election runoff. Given the positive previous change and the ongoing upward surge, this represents a continuation of the upward trend. We expect this momentum to hold into the close.
   Return: -3.66%; alpha: -4.64%.
4. INTC down — confidence 75%. Intel closed down -0.56% on October 2 and continued to face selling pressure on October 5, trading down to $117.38 intraday. The catalyst is a broader speculative tech sell-off and lingering concerns around server demand strength. Since the stock was down in the source session and continues to fall, this confirms a continuation pattern. We forecast further weakness over the one-day horizon.
   Return: -1.17%; alpha: 0.19%.
5. PCVX up — confidence 70%. Vaxcyte closed up +0.77% on October 2. On October 5, shares surged roughly 49% in premarket trading as investors braced for a substantive clinical data catalyst regarding OPUS-1. Since the source session was positive and the stock is surging, this is a continuation. I could not verify the exact current intraday quote or 20-day average dollar volume for PCVX today, so those fields are omitted, but the massive pre-market movement strongly indicates upward momentum.
   Return: -16.40%; alpha: -17.38%.

Full prediction and audit record

DeepSeek

View full picks and reasoning

The source session was Friday, October 2, 2026, which saw the Nasdaq close at a record high after a weaker-than-expected September jobs report pushed rate hike odds down to 20% for October and the Fed pause probability to 80% [citation:14]. The S&P 500 gained 0.73% [citation:3]. However, Monday, October 5 premarket action shows US futures dipping as tech stocks take a breather, with Intel dropping over 4% while Nvidia edged up 0.5% [citation:14]. The extreme movers from Friday span two distinct groups: storage hardware names (STX, WDC) that crashed on Toshiba's HDD capacity expansion news and are now showing modest premarket rebounds [citation:5], and a broad set of small-cap gainers (MXL, SYNA, WOLF, AZTA, PENG) that surged on idiosyncratic catalysts but face profit-taking risk on a down-futures Monday [citation:3]. My selections focus on the storage names as reversal candidates where I could verify premarket action, and large-cap losers (NKE, ACN) where the source-session decline may be overdone.

1. WDC up — confidence 68%. Western Digital fell 9.30% on Friday, October 2, to close at $419.548, making it the second-worst performer among notable US stocks that session [citation:9]. The catalyst was Toshiba's announcement that it plans to invest roughly 60 billion yen to double HDD production capacity for AI data centers by fiscal 2027, threatening the tight supply dynamic that has supported HDD pricing. However, in Monday, October 5 premarket trading, WDC is indicated higher, up over 1% according to one source [citation:5], which would make this a reversal setup: the source-session move was down, and the direction is up. The hypothesis is reversal because the initial selloff appears to have overreacted to a longer-term supply addition rather than an immediate competitive threat, and the AI data center demand narrative remains intact. The main risk is that the premarket rebound fades and the stock resumes its decline if the Toshiba expansion is perceived as more imminent or structurally damaging to pricing. Reference price of $419.548 is the regular-session close from October 2; I could not access a precise live premarket quote at the moment of answering. Market cap and average dollar volume are based on the source-session close. Data date is October 2 as that is the most recent verified price I could confirm. previous_change_percent of -9.30 is for the October 2 source session.
   Return: -6.49%; alpha: -7.47%.
2. STX up — confidence 65%. Seagate Technology was the worst performer in both the S&P 500 and Nasdaq 100 on Friday, October 2, plunging over 12% to close at $825.923, its largest daily percentage drop since April 2025 [citation:7][citation:9]. The catalyst was the same Toshiba HDD capacity expansion news that crushed Western Digital. In Monday, October 5 premarket trading, STX is indicated higher, up over 1% according to one source [citation:5], making this a reversal setup. The hypothesis is reversal on the expectation that the panic selling was overdone and that demand for high-capacity HDDs from AI data centers remains robust enough to absorb additional supply. The main risk is that the Toshiba expansion represents a structural shift in industry supply that will pressure pricing for years, not just a one-day overreaction. Reference price of $825.923 is the regular-session close from October 2; I could not access a precise live premarket quote. Market cap and average dollar volume are based on the source-session close. Data date is October 2. previous_change_percent of -12.65 is for the October 2 source session.
   Return: -8.01%; alpha: -8.99%.
3. NKE up — confidence 62%. Nike fell 5.75% on Friday, October 2, to close at $33.130, making it one of the worst performers among notable US stocks that session [citation:9]. The source does not specify the catalyst for the decline. The hypothesis is reversal (direction up) on the expectation that the selling was overdone and that the stock, trading near 52-week lows, could see a bounce. However, the risks are substantial: Nike appears on the S&P 500 stocks at 52-week lows list with a market cap of $50.2 billion and a one-month decline of 11.4% [citation:16], indicating persistent negative momentum. There is no specific scheduled catalyst for Nike within the window. Reference price of $33.130 is the regular-session close from October 2; I could not find a fresh premarket quote for NKE. Data date is October 2. previous_change_percent of -5.75 is for the October 2 source session.
   Return: 3.62%; alpha: 2.64%.
4. ACN up — confidence 60%. Accenture fell 2.76% on Friday, October 2, to close at $206.445, according to one source [citation:9], though another source places the close at $198.90 [citation:3]. I am using the $206.445 figure from the 財華網 source as it explicitly lists ACN with that price and percentage. The source does not specify the catalyst. Notably, Accenture had surged earlier in the week on better-than-expected quarterly results and a $9.5 billion shareholder return plan, so the Friday drop may represent profit-taking. The hypothesis is reversal (direction up) on the expectation that fundamental support from the earnings beat limits further downside. The risk is that the AI-cannibalization narrative for consulting firms resurfaces. There is no specific scheduled catalyst. Reference price of $206.445 is the regular-session close from October 2; I could not find a fresh premarket quote for ACN. Data date is October 2. previous_change_percent of -2.76 is for the October 2 source session.
   Return: -1.19%; alpha: -2.17%.
5. MXL up — confidence 58%. MaxLinear surged 14.99% on Friday, October 2, to close at an implied price based on a $9.6 billion market cap [citation:3]. The catalyst is not explicitly stated in the source, but the magnitude of the move suggests a significant idiosyncratic event, possibly an analyst upgrade, product announcement, or sector sympathy from the semiconductor equipment rally that lifted Teradyne, ON Semiconductor, and Monolithic Power [citation:1]. The hypothesis is continuation (direction up) on the expectation that the momentum carries into the next session. However, the risk is elevated: Monday futures are pointing down and tech is taking a breather [citation:14], which could trigger profit-taking in a stock that just gained nearly 15% in one session. Reference price and exact close price could not be verified from the available source; the source only provides the market cap of $9.6 billion and the 14.99% gain [citation:3]. I am omitting reference_price and market_cap_usd rather than guessing, and noting that avg_dollar_volume_20d_usd could not be verified. Data date is October 2 for the source-session move. previous_change_percent of +14.99 is for the October 2 source session.
   Return: 0.31%; alpha: -0.67%.

Full prediction and audit record

How models have done in Extreme Movers

ForecastNest's own settled Extreme Movers runs at this horizon, up to and including this one. Members' own research is not counted. Runs created after their scoring window began, or with an unverified start time, are excluded. The chart is the average directional return per run. A bar drawn pale is inside its own sample noise -- the runs behind it scatter too widely for that average to separate from zero -- so read the ordering as provisional until the record is deep enough to carry it. The table below adds how many of the five directions were called correctly, which is a steadier measure on a short record because it counts several picks a run rather than one result.

Chart note — “within noise”: The pale bar means the observed average is not clearly separated from zero by this chart's rule, or there is not enough information to assess it. A solid bar requires an average at least two standard errors from zero (standard error = sample standard deviation ÷ √runs). Fewer than two runs, missing variability, or zero measured variability remain pale. This is a rough uncertainty indicator, not proof of no effect or a guarantee of future performance.

Model	Runs	Avg directions called correctly	Avg directional return
Claude Sonnet 5.5	4	65.00%	+2.00%
DeepSeek	14	47.14%	+1.22%
Grok 4	2	50.00%	+0.31%
Gemini 3 Pro	2	60.00%	+0.31%
Claude Opus 5	6	60.00%	+0.03%
Muse Spark 1.1	13	40.00%	-0.20%
Claude Opus 5.5	3	33.33%	-0.24%
Grok 4.5	9	37.78%	-0.37%
Gemini 1.5 Pro	11	50.91%	-0.42%
Perplexity	7	42.86%	-0.42%
GPT-6	4	45.00%	-2.68%
GPT-5.6 Sol	8	42.50%	-3.81%

View settlement prices

Symbol	Start	End	Source	Notes
ACLS	143.87500000	143.59500000	https://finance.yahoo.com/quote/ACLS/history/
ACN	197.08000000	194.74000000	https://finance.yahoo.com/quote/ACN/history/
BE	286.48960000	293.50010000	https://finance.yahoo.com/quote/BE/history/
CNXC	25.25000000	25.63500000	https://finance.yahoo.com/quote/CNXC/history/
HPE	68.60500000	70.32000000	https://finance.yahoo.com/quote/HPE/history/
INTC	117.33000000	115.96000000	https://finance.yahoo.com/quote/INTC/history/
LW	43.61000000	49.17000000	https://finance.yahoo.com/quote/LW/history/
MXL	107.69000000	108.02500000	https://finance.yahoo.com/quote/MXL/history/
NKE	32.86500000	34.05500000	https://finance.yahoo.com/quote/NKE/history/
ON	85.12000000	86.00000000	https://finance.yahoo.com/quote/ON/history/
PAGS	11.08000000	10.67500000	https://finance.yahoo.com/quote/PAGS/history/
PCVX	78.49250000	65.62000000	https://finance.yahoo.com/quote/PCVX/history/
PENG	60.88200000	60.65000000	https://finance.yahoo.com/quote/PENG/history/
RXO	28.56500000	28.44000000	https://finance.yahoo.com/quote/RXO/history/
SPY	772.47500000	780.05000000	https://finance.yahoo.com/quote/SPY/history/
STM	56.76000000	58.28500000	https://finance.yahoo.com/quote/STM/history/
STNE	11.92500000	11.51000000	https://finance.yahoo.com/quote/STNE/history/
STX	900.96000000	828.82000000	https://finance.yahoo.com/quote/STX/history/
TER	446.28000000	437.07000000	https://finance.yahoo.com/quote/TER/history/
WDC	444.03000000	415.23000000	https://finance.yahoo.com/quote/WDC/history/
WOLF	32.89000000	33.95500000	https://finance.yahoo.com/quote/WOLF/history/

Standard prompt extreme-movers-v11, SHA-256 0ee02633ceb23a9ed220510489bdbb9768b08af78503abed24244a6c3c40e05d. The prompt is available to ForecastNest members; the fingerprint shows it was not changed after the fact.

Calculation: extreme-movers-v1.