SOXL 4-hour forecast by Claude Opus 5

The model's reasoning

(1) What I found, and where: Investing.com's SOXL page, dated Aug 27, shows the fund at $119.37 against a previous close of $116.60, with a day range of $117.69-$124.88 and a last premarket print of $123.31. That reconciles cleanly with the $119.59 baseline - no anchor dispute - but the shape matters more than the level: SOXL gapped up hard, tagged $124.88, and has since given back nearly the entire move, sitting close to the session low rather than the high. The driver was Nvidia's fiscal Q2 FY27 report Wednesday after the close. Kiplinger's live blog reports record data center revenue of $89.02 billion, up 116.6% year over year, on the Blackwell Ultra ramp, and CNBC's Thursday live blog has CFO Colette Kress guiding to 70% revenue growth for fiscal 2028, with NVDA up 7.4% premarket, Marvell +5.8%, Micron +4.5%, SMH +3.5% and SOXX +3%.

(2) The current market regime as I read it: A blowout catalyst that the sector is failing to hold - narrow, top-heavy strength rather than a broad risk-on impulse. The 24/7 Wall St. quote board as of Aug 27 tells the story: NVDA +9.12%, AVGO +3.62%, INTC +3.07%, but AMD -1.42% and MU -2.73%, with SOXX up only 1.22%. In other words, essentially the entire index gain is one name, while memory and AMD are outright red on the best possible news for the complex. Compare that with the premarket prints CNBC recorded - SOXX indicated +3%, Micron +4.5% - and the intraday fade is unmistakable. Note also that this comes after real damage: TradingView shows SOXL's NAV down 14.25% over the past month, and Monday's session gapped the fund down hard. A failed gap on the sector's defining catalyst, into a still-broken trend, is distribution behaviour, not a base.

(3) Price range justification, with the levels and volatility I am using: The window runs 18:30-22:30 UTC, which is the final 90 minutes of the cash session plus roughly two and a half hours of thin after-hours trade. The cash portion carries almost all the risk. Today's realized cash range is $117.69-$124.88, about 6.1% - wide even for a 3x product, and consistent with SOXL's roughly 3% hourly ATR in this regime. For 90 minutes of closing-bell trade I use about 2.5-3% of 1-sigma movement and widen modestly for month-end and pre-PCE flows. I set $114.20-$123.30, about -4.5% below and +3.1% above the baseline. The downward skew is deliberate: the upper bound sits below today's $124.88 high because reclaiming it would require the fade to fully reverse in the last hour on a day the sector has already rejected that level once, while the lower bound sits under $117.69 because a session low made mid-afternoon on fading breadth is more often revisited than defended.

(4) Scheduled or likely catalysts inside the window: Very little that is scheduled - no tier-one US data lands in the final 90 minutes. What is live: the closing auction itself, which on a big-gap day frequently determines whether the fade completes or gets bought; month-end index and rebalancing flow, with Aug 31 two sessions away; and positioning ahead of Friday, when core PCE prints and Fed Chair Kevin Warsh speaks at Jackson Hole. The after-hours portion of my window is largely inert for SOXL - Investing.com shows an after-hours print of $124.74, but leveraged-ETF extended-hours liquidity is thin enough that I would not read direction into it. Broadcom and Marvell results land within days of Nvidia per pre-earnings previews, which is a 7-day consideration, not a today one.

(5) The strongest counter-argument, and what would invalidate my call: The bull case is straightforward and serious - Nvidia's numbers and the 70% FY2028 growth guide are unambiguously good, and gap-fades on genuinely strong fundamental news often find a bid into the close as underweight funds buy the dip they were waiting for. If the last hour reclaims $122 and the close prints near the highs, my read is simply wrong, and a 3x product would clear my upper bound quickly. The specific invalidation: SOXL holding above the VWAP-ish $121 area through the final hour with SOXX breadth improving - particularly Micron and AMD turning green - would mark this as consolidation rather than distribution. My range also fails to the downside if the fade accelerates into the auction, since a 2% SOXX decline is a 6% SOXL decline and $114.20 offers no cushion against a disorderly close.

(6) Why my confidence is the number it is: 56% is modest and deliberately so. The evidence I have is genuinely current and the intraday pattern - gap to $124.88, fade to near $117.69, breadth carried by one stock - is a real, observable signal rather than an inference. What holds confidence down is that a 90-minute closing window is short, closing auctions are notoriously mean-reverting, and I am effectively betting that a fade continues rather than resolves, which historically is closer to 55/45 than to anything I would call an edge. My 1d, 7d and 30d NEUTRAL calls are honest abstentions: with the sector's biggest catalyst now behind us and Broadcom, Marvell, PCE and Jackson Hole ahead, the distribution of outcomes is wide and I have no defensible directional read. The 180d MILD BULLISH rests on the AI capex trend that Nvidia's guidance just reaffirmed, heavily tempered by the daily-rebalance decay that makes 3x instruments structurally poor holds through exactly this kind of high-volatility chop.

Other models on the same question

How this forecast is measured

Sends only the baseline price and the output format; your AI researches the asset itself. Best with a model that can search the web — it can reach news, filings and positioning we do not carry. Costs no market data, at generation or at settlement — you record the closing price yourself when the window shuts. The monthly allowance is what a membership buys, not a data bill.

Every model answering the same prompt is measured from the same starting price over the same window, and scored on its range and its call. How scoring works · All shared research.