Altair
Skip to content

Trading Analysis Report: ORCL

Analysis date: 2026-10-14 · Model: claude-opus-5-5 · AI-generated, not financial advice.

  • Analysis date: 2026-10-14
  • Rating: Underweight
  • Generated: 2026-10-14 17:33:48
  • TradingAgents 0.6.0: anthropic, deep claude-opus-5-5, quick claude-sonnet-5-5
  • Analysts: market, sentiment, news, fundamentals; research debate rounds 5, risk debate rounds 5
  • Data vendors: core_stock_apis yfinance, technical_indicators yfinance, fundamental_data sec_edgar,yfinance, news_data yfinance, macro_data fred, prediction_markets polymarket

V. Portfolio Manager Decision

Rating: Underweight

Executive Summary: Underweight ORCL. Add no new money (stay out if flat). Cut existing exposure to one-third of a standard allocation by default before the Oct 29 FOMC, or to one-half only if a read of the already-filed Aug-31 10-Q clears the funding test. Sell half the distance to target on Oct 15-16 (not in the opening auction), rest limits near $135 and $138.75-140.79, then sell half of anything unfilled by Oct 21 and the rest by Oct 27; finish the cut on a close below $125.70, step a half position down to a third on a close below $120.93 or Oct-hike odds above ~40%, and exit the remainder below $114.57 expecting slippage. Rebuild one-third of what was sold on a volume-backed close above $145.12 with a clean filing, and the rest only on funding evidence; horizon 2-3 months through the December print, price target $116.

Current Price: 130.55

Price Target: 116.0

Confidence: Medium

Time Horizon: 2-3 months

IV. Risk Management Team Decision

Aggressive Analyst

Aggressive Analyst: Neither of you has spoken yet, so I'll open the debate and answer the objections I expect to hear.

The Sell is the bold call here, and the data supports it. The stock fell 8.8% on about 2x volume and closed in the bottom 11% of its range. It sits below the 10-EMA, the 50-SMA and the 200-SMA, and SuperTrend is down on all three timeframes. The macro backdrop got worse the same day: Polymarket puts a 25 bp hike at the October 29 FOMC at 28%, up 13 points in a week, and 86% on at least one more hike this year. ORCL is a 1.77 beta, debt-funded, long-duration AI infrastructure name. That is the asset class a hawkish Fed hurts most.

The conservative analyst will say we're selling after an 8.8% flush, with the daily z-score at -1.89 and price at the lower Bollinger band, so this is the worst time to sell. But the plan doesn't sell the low. It trims into a bounce toward the 10-EMA at 138.75 or the 140 to 143.20 gap zone. RSI is only 40.57, not oversold, and the TD counts are 1 on the daily and 4 on the weekly and monthly, nowhere near exhaustion. Nothing has said the selling is done. Selling at 138.75 with a reassessment trigger at 145.12 risks about 6.4 points to target the July low around 114.57, which is 24 points. That's roughly 3.8 to 1, and most "balanced" holds can't offer that.

The neutral analyst will point to 30% revenue growth, expanding margins, a 11.9x forward P/E, a PEG of 0.8, and a director buying at 139.35. The growth is real, and the plan is a trim, not a full exit, for exactly that reason. But the forward P/E rests on about $11 of forward EPS, 72% above trailing EPS, and trailing EPS itself includes a one-off. The 11.5x market cap to EBITDA isn't EV/EBITDA. With debt that the report infers at $100B or more against $36B of cash, the true multiple is closer to 13 or 14x, though that's my inference. Cheap-on-forward-numbers is a bet that a $45.9B trailing free cash burn turns around on schedule, while capex ran above 1x revenue and equity jumped about $24B in one quarter against $4.8B of net income. That looks like issuance, and issuing at $130 dilutes more than issuing at $160. Diluted shares are already up about 5%.

On the insider buy: $3.5M from one director against roughly $113M of priced sales over twelve months is a data point, not a thesis. He is also already about 6% underwater.

Two honest caveats. The Aug-26 operating cash flow of $23.1B looks anomalous and needs checking against the 10-Q. The 8.8% drop might also have been partly a closing-bar artifact if the data was pulled before the session ended. Still, I'd rather act before the 10-Q and the FOMC than wait for confirmation that arrives after the move. The cost of being wrong on the trim is a missed bounce on half the position. The cost of waiting is holding a full allocation through a hike, a possible equity raise, and a retest of the July lows.

That's also why the 0.5x sizing matters more than the stop. With ADX at 7.4 and ATR near 5.5%, a stop 6 points above entry will get whipsawed. Sizing is what bounds the loss. I'd also push back on anyone who wants to buy this tape: the sentiment data is low-confidence and skewed to bearish in the actual post text, even though the labeled tags look bullish. The conditions to flip are specific and testable: a volume-backed close above 145.12, a clean 10-Q, or a cooler inflation print. Until one of those happens, the bold move is to cut to about half a standard allocation into strength, add nothing, and exit the rest if the July low close breaks. Aggressive Analyst: Let me give ground where it's been earned, because it makes clearer what I'm holding onto.

The conservative is right that 3.8 to 1 depends on the fill. From the 130.55 close it's about 16 points down to the July low close and 14.6 up to the 50-day, so a market sell here is roughly a 1.1 to 1 trade. The neutral framed it better than I did. At a 138.75 fill the downside only has to win about one time in five, and at today's close it has to win nearly half the time. But nobody proposed selling here, including me. The plan is to sell into strength, and this tape rewards that. ADX is 7.4, and closes have swung between roughly 132 and 150 since mid-September. In a chop like that, a resting sell in the upper half of the range is the right tool, and a market order at the lower band is the wrong one. I'll also stop calling 145.12 a stop. It's 11% above the close, and the daily SuperTrend doesn't flip until 154. It's a rebuild line, and size is the real control.

The conservative's attack lands on the no-bounce case. If PPI is hot and the stock goes to 120 without touching 138, a pure wait-for-strength plan sells nothing. The neutral's fix is the best one on the table. It has resting limits near 135 and between the 10-EMA and the midline, a floor on October 27, and a floor pulled forward on a close near 125.7. If the after-hours chatter about a 135 open is true, those orders fill. If it isn't, I haven't paid for being wrong. I'll take all of that, plus the checkpoints at 120.93 and hike odds above 40%. The conservative was right that 114.57 as the only exit leaves the half I kept riding a 12% slide.

I'd quibble only with the floor size. Selling a quarter leaves you at 0.75 of current size, which at a 1.77 beta is still about 1.3 of market beta going into the FOMC. That prevents a disaster, but it doesn't deliver the Sell, so I'd make it a third.

I won't take the hard deadline or the 129.23 trigger. A close under today's low is 1.3 points away, about a fifth of an ATR. That's far tighter than the six-point stop I already admitted would whipsaw, and it would likely fire near the bottom of the range and sell a second tranche at the worst price. Forced selling by calendar date is how a trim becomes capitulation. As for sizing, a third isn't better anchored than a half. Mine is anchored to what I can afford to be wrong about, which is a missed bounce on half. I'd step down to a third when evidence arrives: a close under 120.93, hike odds past 40%, or a filing that shows funding stress.

On fundamentals, I'll concede one thing and defend another. The neutral's share count math is better evidence for dilution than my equity-jump argument. Net income over EPS gives about 2.93B shares in February and 3.05B in August, roughly 4% in two quarters after about 1% over the three before. I'll add a wrinkle against myself. The quarter's financing line is only $13.1B, which doesn't fit $18B or more of fresh equity unless debt was paid down or part of the equity rise was non-cash. So the dilution is real, but how it was funded isn't pinned down. The neutral is also right that if shares went out at higher prices over the summer, a raise at today's price is less imminent than the retail posts imagine.

That doesn't dissolve the thesis, though. Capex ran $28.5B against $19.3B of revenue, and trailing free cash flow is still negative $45.9B. Even if the $23.1B operating cash flow is a real single-quarter number, which the roll-forward mildly supports, that quarter still took $13.1B of financing. For the burn to look benign, that operating cash flow has to repeat. Calling it customer prepayments is a guess, mine no less than anyone's. The trader's $20B-a-quarter burn is the weakest number in the plan, which is why this is a trim and not an exit. If tonight's check shows the inflow is recurring, the trim shrinks.

I also concede the 10-Q point. I was treating it as a future catalyst, but with a 40-day deadline from August 31 it was likely due around October 10. Reading it comes before placing any order. I'll drop the insider evidence both ways, since $63.7M of the $113M is one programmatic exercise. I'll drop the closing-bar caveat too, since the posts are stamped after the close.

The neutral's correlation point is the best new idea here. ORCL fell 8.8% on a day CoreWeave fell 10% and Nebius 6%. A $396B company moving with a neocloud whose bonds touched 15% tells me the market is pricing the funding channel, not the company. That's the exposure to cut, and parking proceeds in those other names undoes the trade.

So I keep the direction and take the neutral's mechanics. The floor is a third, I refuse the 129.23 trigger, and a rebuild needs two of three: a volume-backed close above 145.12, a clean filing, and a cooler inflation print. Aggressive Analyst: Let me give up what I should first, because the fight that's left is smaller than it looks.

The conservative is right that a floor is a date, and I can't accept one while calling dates capitulation. What separates a floor from a fire sale is size. The neutral is also right that the close-based ladder is thin. 125.7, 120.93 and 114.57 sit five or six points apart, today moved 12.65, and a gap walks through all of them. I shouldn't have sold that ladder as protection. Nothing bounds a gap except the size you carry when it happens, and that goes for every plan on the table, a third included.

I also have to correct my own pitch. The 3.8 to 1 is a payoff ratio, not an edge. The neutral's random-walk numbers show this. From a 138.75 fill, the stock reaches the July low close before the 50-day about 21% of the time, which is exactly break-even. From 135 it's about a third, again break-even. A good fill doesn't create an edge, it just stops you from overpaying. Any edge has to come from the thesis: a funding need that isn't verified, hike odds that went from roughly 15% to 28% in a week, and a stock that is trading as a rate and credit instrument. With ADX at 7.4 I can't prove that tilts the odds. It's why this is a trim and not an exit or a short.

Here is where I still disagree. The conservative calls the 140 limit a coin flip, and at 140 it nearly is, about 55/45 against 119. At 135 it's closer to 70/30, and the neutral's plan puts a bit more than half the trim there. The no-bounce path doesn't leave the holder at full size either. It ends at two-thirds at worst by October 27, sooner if 125.7 closes. That's a bounded outcome. It isn't the "I haven't paid for being wrong" outcome the conservative described.

On PPI, the neutral's timing point settles it. The release comes before the open, and the sentiment report dates it October 15. That's the standard morning schedule, which is outside our data. So no regular-session order sells ahead of it, and "hold full size into PPI" versus "sell before it" is mostly a debate about a print that will already be in the price. The only way to act first is extended hours. The "ripping after hours" and "135 open tomorrow" posts are the only evidence we have there, and they're unverified. If a holder can see an extended-hours quote near 135 tonight, a resting limit there is the best fill on the table. If not, it's a pre-open limit. This matches what the conservative asked for on the first tranche, and I agree with it.

I'll take the conservative's filing rule, because it settles the third-versus-half fight with information instead of temperament. Read the cash flow statement tonight. If you can name the source of the $23.1B, end at up to half a standard allocation. If you can't, end at a third. I would stop there, though. Right now the plan has a filing check, a spread check, a floor, a pulled-forward floor, three step-down triggers and a two-of-three rebuild. In a tape that gaps, that many conditionals get executed badly or not at all. So I'd collapse the step-down to four: the cash flow doesn't reconcile, ORCL's spreads visibly widened, a close under 120.93, or hike odds above about 40%. The spread check is useful for anyone with a terminal, but our tools can't produce it. A gate nobody can satisfy would quietly pull everyone to a third.

On fundamentals, the neutral's reconstruction helps the bear case more than the neutral intended. Burn was about $0.4B in the August 2025 quarter and averaged about $10.7B across the November and February quarters. It then jumped to roughly $19B in the missing May quarter, so it has been rising every period. The conservative is right that the uncertainty is skewed. An operating cash flow of $23.1B on $19.3B of revenue can't be earnings. It's working capital, and a prepayment is a deferred-revenue obligation that has to be delivered with more capex. The neutral is right that customers funding part of the build is a real positive. But it also ties the stock more tightly to a customer base whose infrastructure lead is leaving, though the OpenAI link is inference. Nothing here argues for an exit, and I'm not making one.

The neutral also noted that the filing is more than a week old, so the market has had those numbers. I agree, and it cuts my way. If the fundamentals were already known and the stock still fell 8.8% on a rates print, the funding channel is what's being repriced. A filing is unlikely to rescue it, and a hike would make it worse.

One more point from the neutral is worth keeping. Half of a standard allocation and half of what you hold are different things. So state the target as a fraction of standard allocation. If the caller is already at or below half, there's nothing to sell, and they should keep the same exits and rebuild rules. If they hold nothing, stay out, and don't park proceeds in CoreWeave, Nebius or Nvidia.

My plan is the neutral's mechanics with the conservative's filing rule. Add nothing. Rest the sells tonight, a bit more than half the trim near 135 and the rest between 138.75 and 140.79. Sell at least a third of the position by October 27, and pull that forward on a close under 125.7. End at half a standard allocation if the cash flow ties out, and at a third if it doesn't or a step-down trigger fires. Exit under 114.57 and expect slippage. Rebuild on two of three: a volume-backed close above 145.12, a reconciled filing, and a cooler inflation print.

That is the bold version of the Sell. You trim a company you like, with 30% growth and expanding margins, because you can't verify what funds it. You make the market come to you with limit orders instead of selling into a flush. Aggressive Analyst: I'll start with the conservative's fill math, because it's the cleanest point anyone made. 0.72 times 135 plus 0.28 times 119 is about 130.5, today's close. In a driftless walk a resting limit doesn't improve your average price, it just reshapes the outcomes, and one of them is still holding. I said a good fill isn't an edge, and this is the proof.

It follows that a market order earns nothing on average either, so the only reason to wait for a bounce is a belief in mean reversion. Mine is thin. The daily z-score of -1.89, the pierce of the lower band and the mild OBV divergence lean toward a bounce. The weekly z-score of -0.82, an RSI of 40 and early TD counts say the washout isn't finished. That's a coin flip with a small tilt, which isn't worth nine sessions of variance. So I'll take the early first slice, spread across tomorrow and Friday as the neutral suggested. If someone can see a real quote near 135 on their own screen tonight, take it, because that's the bounce arriving early. A post saying "ripping after hours" is not a quote. With the first slice sold early and the October 27 backstop in place, the no-bounce path ends at target before the FOMC, not at two-thirds.

I'll also take the flipped default of a third unless the filing earns a half. That's where I'd already landed when I said to size to the smaller number if the cash flow can't be reconciled. But I won't go below a third on what we have. Revenue growth went from 12% to 30%, operating margin from 28.7% to 34.8%, and the forward P/E is 11.9. What the data shows is unverified and heavily financed, not broken. Going lower turns a hedge against the unknown into a verdict, and nobody here has evidence for one.

The conservative's runway math also uses the wrong frame. $36.4B of cash against a $19B quarterly burn sounds like two quarters, but this is an investment-grade issuer that pulled roughly $59B of financing in the periods the feed covers. Oracle isn't going to run out of cash. The risk is the price of the money and the dilution that comes with it, which is exactly what a hawkish Fed and a 15% neocloud bond are repricing. That supports a trim and argues against an exit.

On the filing test, the neutral's three questions are good, and I'd fold the capex gap into the second. Don't just ask whether twelve months of maturities are covered by cash plus committed facilities. Ask whether they're covered after the capex the company guided. Maturities covered while capex isn't is the exact situation we're worried about.

The neutral says I can't claim the filing won't rescue the stock and then use it to size. Those are different jobs. The market has had those numbers for a month, so I don't expect the filing to move the price. It tells me how fat my left tail is, and that's a sizing question.

On triggers, the conservative caught me. Four becomes three: a failed filing test, a close under 120.93, or hike odds above about 40%. I'll drop the spread check. CoreWeave and Nebius become a reason to be patient or to sell faster, not a gate. CoreWeave has its own Gartner problem, so a bounce there doesn't prove Oracle's funding is fine.

The neutral's beta arithmetic is right but cuts less than advertised. An 8.8% drop at a 1.77 beta implies roughly a 5% index fall. If the index was down 2%, about 60% of ORCL's move was something other than beta, and at 3% it's about 40%. Either way it's a large piece. The conservative's observation that about three quarters of the drop came after the open, on 52 million shares, once the CPI gap was already in, is the best evidence we have that the funding channel is being repriced. The conservative handed me the best fact for the Sell in this entire debate.

Where I still disagree is the rebuild. Both of you want one door, a volume-backed close above 145.12. I've argued all along that levels whipsaw with ADX at 7.4 and a 5.5% ATR, and that applies to buying back too. Closes have swung between 132 and 150 for a month, so one high-volume close above 145 can easily be a head fake. Selling around 131 to 139 and rebuying at 146 costs roughly 5% to 11% on that slice. Scale in the way you scaled out. Rebuy half the trimmed slice on that first close above 145.12, once the filing test has passed. Rebuy the rest when the daily SuperTrend flips at 154.01 or hike odds retreat. Treat the round-trip cost as an insurance premium and size the trim so you can afford it.

This is also where the reward is. ORCL fell 53% in under eight weeks, rose 41% in the next six, and has since given back about two-thirds of that rebound. In a tape like that, whoever holds cash can act on either leg. Whoever is at full size in a levered name heading into a 28% hike tail can't. Trimming isn't a retreat from a company growing 30%. It buys the ability to be aggressive about it later, when the funding question is answered.

So my plan is to add nothing and sell the first third across tomorrow and Friday. Rest limits go near 135 and between 138.75 and 140.79, with everything unfilled sold on October 27. End at a third, or at a half if the filing test passes. Step down on the three triggers, exit under 114.57 and expect slippage, and rebuild in two steps. That's still a Sell. Aggressive Analyst: We've converged enough that the useful thing is to separate what I'm giving up from what I'm keeping. There's less of the second than there was an hour ago.

To the neutral first. You asked me to drop the bold framing, and on sizing you're right. A third as the default is a hedge against numbers none of us can verify, and I shouldn't dress it up as conviction. I gave up the edge claim when the fill math showed a good price doesn't create one. What I'll keep is narrower. The size is a hedge, but the timing is a decision. Selling a slice this week, after an 8.8% flush, with the daily z-score at -1.89, no company-specific news and nothing confirmed, is the uncomfortable version of this trade. The comfortable version is to wait for the filing read, PPI and the Fed, and each of those waits is cheap until it isn't. I'd rather be early at a size I can afford than confirmed and late. That's all "bold" means here, and it's small.

You're also right about 154.01. Finishing the rebuy at the daily SuperTrend flip means buying back about 14% above where I sold, which goes well past an insurance premium, so I withdraw it. I'd change your second rebuy leg too. You wrote that the rest comes back after October 29 if there's no hike and the stock is still above 145. The no-hike part is information, so it's a good trigger. The above-145 part only lets you rebuy at a worse price than you sold at. If the Fed holds, the filing is clean and the stock is at 128, the thing I was hedging has resolved and the stock is cheaper. That's when I'd want the second leg, not after a 10 to 12% rally. Tie that leg to the event and the filing, and let price be whatever it is. I'd also settle the filing questions before any order goes in, as you said, so the target is known before tomorrow's open. Then nobody sells down to a third over something an hour of reading would have cleared, and pays to buy it back.

I accept your split between hard and soft evidence. A rating action, a negative outlook, a covenant problem, or maturities that exceed cash plus committed facilities after guided capex earn a position below a third. A line I can't reconcile in ten minutes earns a third and nothing lower. That settles most of my disagreement with the conservative. I'll also take total interest incurred, expensed plus capitalized, since capitalized construction interest could hide the real burden.

I'll give you the after-the-open point as well. I called it the best fact for the Sell, the conservative walked it back, and you're right that the part of the move that isn't index beta is the AI-credit factor. A levered, long-duration builder falls with CoreWeave whatever its own spreads did. I don't need it to be Oracle-specific. The trim exists because ORCL carries that exposure alongside -$45.9B of trailing free cash flow, so reducing it is reasonable either way.

On the covered call, it's worth a look, but I'd be clear about what it does. After an 8.8% day, implied volatility is probably high, which favors the call seller and makes the put spread expensive. But a call near 145 has modest delta, so it barely reduces exposure if the stock goes to 118. It monetizes the bounce scenario and doesn't protect against the one we're worried about. I'd use it on the portion you plan to keep, or as a tax-sensitive holder's version of the upper limit tranche. It doesn't replace the trim or the October 27 backstop. That's reasoning, not a quote, since I have no options data. And I agree on not sitting at 145 hoping. The highest limit in the plan is the 140.79 midline.

To the conservative. I agree a third is the price of not knowing, and hard bad evidence would justify going lower. The $87B funding need is the trader's figure with no derivation shown, so I'd treat it as a scenario, not a forecast. Even so, a quarter of it as equity at today's price is about 165 million shares, roughly 5% on top of the 4% in the last two quarters. That's the shape of the risk I care about even if the number is off. Your Nasdaq-close homework is the best cheap check on the table. If the index fell 2 to 3%, a large part of ORCL's 8.8% wasn't beta, and I'd go at the fast end of the schedule. If it fell 4 to 5%, ORCL did what its beta says, and I'd let the limits work longer. The October 27 backstop stays either way, because the point of a backstop is that nobody can tell in advance which day the bounce fails.

You also caught a real inconsistency on rebuying. I said to sell the upper half of the range, then proposed buying half back on the first close above 145 while arguing that close could be a head fake. A third of what was sold, on a clean filing, is the right size for a signal I don't yet trust.

Here is where I land. Add nothing. Before the first order, answer three questions. Is the $23.1B explained by named working-capital lines? Are the next twelve months of maturities plus guided capex covered by cash and committed facilities? Is total interest incurred growing more slowly than operating income? If all three are clean, the target is half a standard allocation. If anything is unclear, it's a third. Go below that only on hard evidence.

Sell half the distance to target tomorrow and Friday, not in the opening auction. Rest limits near 135 and between 138.75 and 140.79, with nothing above. Sell whatever is left on October 27. A close under 125.7 finishes the cut next session. A close under 120.93, or hike odds above about 40%, takes a half down to a third. Exit the remainder under 114.57 and expect it to fill lower.

Rebuild a third of what was sold on a volume-backed close above 145.12 with a clean filing. Bring back the rest on a Fed hold on the 29th with the filing still clean, whatever the stock is doing. If you hold nothing, stay out. If you're already at or below target, there's nothing to sell, and the proceeds shouldn't go into CoreWeave, Nebius or Nvidia.

It's still a Sell, but a modest one. It trims a company growing 30% with widening margins to a size where being wrong in either direction costs little, on a schedule that doesn't depend on the tape being kind.