Altair
Skip to content

Trading Analysis Report: AMAT

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 15:29:22
  • 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 AMAT. Holders above target should sell half the excess over 55% of a standard allocation into the 10/14 close near $447, which replaces the stale 464.24 entry, and rest the remainder as limit sells across $464-490. Sell any leftover into the close if the stock is still below $436.79 late in the session after a first close under it, and finish by the 10/27 close, landing at 50% of standard (55-60% only if a pre-written 10-year/CPI test passes) or lower if your stress budget at $360/$335 requires it; no short and no adds before the FOMC. A daily close below $387.79, executed into the close, caps every holder at 40% of standard if the Q3 10-Q gate passes and 25% if it fails or is unresolved (25% regardless on a 10/30 monthly close below it), while a daily SuperTrend flip up (now $513.10) held for two closes means the bearish view is wrong; after 10/30 rebuild through one gated rung only until the November print, with a $412.50 target over 1-3 months.

CONFIDENCE: Medium. The direction is unanimous among the three analysts and the trader. But the 10/14 bar may be partial; the 10-year, CPI and 10-Q detail are missing; and the $18.47 EPS basis, the FOMC date and the earnings date are unconfirmed. Recheck 436.79, 411.51, 387.79 and 490/513 at the final close.

Current Price: 446.63

Price Target: 412.5

Confidence: Medium

Time Horizon: 1-3 months

IV. Risk Management Team Decision

Aggressive Analyst

Aggressive Analyst: Neither of you has spoken yet, so I'll answer the arguments I expect and make the case for selling AMAT.

The conservative will say you don't sell a record-quarter company after a 38% drawdown, because that's selling the lows. But the plan doesn't sell the lows. The daily TD-9 is only at +6, RSI is 36.4, not 30, and the daily z-score is -1.31, so nothing here says capitulation. The plan is to trim into a bounce toward 464-490 and hold the current weight if price falls into 411-417. That's a trader who knows where he wants to sell and where he wants to buy. And the 17.6% drop in 7 sessions is only a bad moment to sell if you were already planning to sell. The way to stay ahead of the move is to sell the bounce.

The conservative will also point to the business. Revenue was $9.1B, up 24.8%, and operating margin was 33.7%. I agree it's an excellent business, which is why this is a trim to 50-60% of standard weight and not an exit. But the stock isn't priced on Q3. It's priced on forward EPS of $18.47, which needs about $4.62 a quarter against $3.17 delivered, roughly 46% above the current run-rate. Free cash flow conversion is 49% against 72% a year ago, capex is up 35%, and the free cash flow yield is under 1%. Insiders sold about $180M, including $169M in June near the high, with no open-market buying. Then Investor Day delivered 2030 margin targets that RBC called conservative, and the stock fell about 6-7% anyway because there was no WFE forecast. When good news gets sold, that tells you who owns the stock and what they'll do next.

The neutral analyst will say it's balanced: strong fundamentals against a weak tape, so hold standard weight. That's a position too, and it's the one I'd least want. Holding means carrying full exposure, with a 1.59 beta and a 4.5% ATR, into an FOMC on 10/29 where Polymarket has a hike at 30%, up 15 points in a week, and 97% odds of no cuts all year. The market has moved from a hawkish hold toward a live hike, and a long-duration, long-payback equipment name is the wrong thing to be fully loaded in. Weekly SuperTrend is down, the MACD histogram has widened from +12.09 to -5.49 in two weeks, and the 10 EMA and 50 SMA are both overhead at 487-491. The tape is telling you which way the pressure runs.

Then there's the math. From 446.63, the first leg to 490 is about +9.7%. A drop to the monthly SuperTrend at 387.79 is about -13.2%. That's roughly 0.7:1, and 1.1:1 even if you stretch to the 513 stop. At 411-417 that flips to nearly 3:1, with about 6% of downside against about 19% to 490. Selling around 464 and buying around 413 exploits that gap, and nobody is paid to stay in the middle of it.

The sentiment data makes the same point. Retail's dip-buyers keep saying "under 400" and "dry powder ready." Tagged sentiment reads bullish at 8 to 1, but the content shows people waiting for lower prices. That's not capitulation. The sell-side is the same story: Cantor at $850 is 83% upside on paper, but targets chase price. And SOXX is on a seventh straight losing session, with put flow showing up in the semis.

I'll own the data gaps. The 10/14 bar may be partial, the 10-Q isn't in, and we have no 10-year yield or CPI number. But the plan already builds that in. It says to recheck levels at the final close and to hold off rebuilding until the 10-Q passes on DSO, DIO and contract liabilities. Missing information is a reason to size each tranche, so a 7% gap through 387.79 costs no more than about 25bp of the book. It isn't a reason to freeze.

The downside of being wrong is bounded. If AMAT closes above 513.10, the bearish thesis is dead and we've given up part of a position. If it breaks 387.79 without the gate passing, we're down to 25% of standard and still have powder. The upside is that we hold dry powder at 380-417 for a company still compounding revenue at 25% a year, after surviving the macro event. That's the higher-return path, and I'd take it over being fully long into a hike risk. Aggressive Analyst: Let me start with what's settled, because it's most of the plan. All three of us want AMAT trimmed and nobody wants an exit, so the argument is over mechanics. I'm giving ground on some of them and holding it on others.

Conservative, you're right about the things I'd have been embarrassed to defend. The 25bp cap measures slippage past 387.79, not the trip from 446.63, and the trip is what costs money. Restate it as loss from here to a stress price. I'd use roughly 360, about 7% under the monthly SuperTrend, which is a 19% fall, or about 19bp per 1% of book. Say standard weight is 3% of the book. Then 60% of that is 1.8%, and a trip to 360 costs about 35bp. That's illustrative, since we don't know the real book. Pick the loss you can live with and back out the weight.

You're also right that 464.24 is a stale anchor and that 513.10 tells you the trim was wrong, not that capital is protected. I'd relabel 464 as the next tranche level and 513.10 as the thesis-is-dead line, with an earlier check: a daily close back above the 487-491 cluster means stop trimming. I'll also drop the three arguments you flagged. The forward EPS gap isn't like-for-like, we don't know whether the insider sales were 10b5-1 plans, and the WFE explanation comes from StockTwits. The case stands on valuation, the tape, the rate backdrop and cash conversion. On that last one, Q3 operating cash flow was $3.04B against $2.54B of net income. I'd treat weak conversion as a cap on the multiple you'll pay, not a crack in the business.

Where I disagree is making a close under 436.79 the deadline. It's 9.84 points away, about half an ATR, on a stock that just ranged 32.82 points in a single session. Finishing the trim there means selling within a day or two, after a roughly 10% two-day drop, with the daily z-score at -1.31 and the daily TD-9 three down closes from a 9. That's selling the second leg of a flush. Neutral's two-consecutive-closes filter keeps your urgency without the noise, so I'll take that.

You're also right about the 10-Q. The 40-day deadline for a July 26 quarter-end passed in early September, and the fundamentals report already carries balance sheet and cash flow figures dated July 26. Pulling it is a data task, not a market event. What we can see is suggestive but not conclusive. Current assets rose about $2.5B last quarter, current liabilities about $1.4B, and cash only $0.7B, so roughly $1.8B went into receivables, inventory or short-term investments. But some of the liability growth could be customer deposits, which would be a demand signal. Pull DSO, DIO and contract liabilities today. Once the gate is resolved, the 380-395 bids and the 25% cut stop colliding, because they belong to different states of the world. Also, the monthly SuperTrend flips on a monthly close. A daily close under 387.79 is a warning, so the gate should decide whether it means 25%.

Neutral, you handed me the best fact in the debate and then used it against me. Yes, AMAT ran 30% from 417.40 to 542.28. It then gave back 95.65 of those 124.88 points in seven sessions, about 77%, and the high it made was far below June's 722.23 close. That rally got sold, and it makes a retest of 411-417 very live. The weight should follow the skew. At 447 the skew is 0.7:1 to 1.1:1 against you. At 413 it's roughly 3:1 for you. Holding a reduced weight at 413 and holding full weight at 447 are different stances, so I'm not being inconsistent about "hold."

I'll take your thirds structure, including the first third once the 10/14 bar is final. Conservative is right that protection that only works if the market cooperates isn't protection. Selling at 447 instead of 464 costs about 4% on that third, and I'll pay that to avoid depending on a fill. I'd add "excess over target" to the wording, as you said.

I'd go further than you on the rebuild. The plan says add nothing before 10/29 but also posts bids at 410-417, and those contradict each other. If the gate passes and the stock reaches 411-417 first, another 7% of weakness means a good part of the hike risk is already priced. That's the best skew on the map, so let the stress budget set the size, not a calendar rule. Retail's "under 400" posts cut both ways. Bids may be thin at 417, but a crowd stacking up under 400 sits right next to my 380-395 zone and the monthly stop, which is a reason not to sell down there.

One correction to your scenarios. In a V-bounce, the second and third tranches also fill into 464-491, so most of the trim is done by 490 and the regret isn't just one tranche. I'm fine with that. 490 is where the 10 EMA and 50 SMA sit and where the 10/13 gap-down opened. But call it what it is.

So the Sell stands. The direction is underweight, the first third goes after the final bar, and the rest goes on a bounce or two closes under 436.79. 513.10 is the thesis-dead line, the gate gets resolved today, and the rebuild is sized by stress loss. That's a bolder plan than carrying full weight into a 30% hike probability and 97% odds of no cuts. It's also a less fragile one than waiting for a 464 fill that may never come. Aggressive Analyst: We've converged more than any of us expected, so I'll say where I move and where I don't.

Conservative is right that my 3:1 was a best case. It used 387.79 as the floor, but in the world where I'd actually be buying, with the gate passed, the plan doesn't cut at 387.79, so the honest floor is the stress price. From 413, 490 is about 18.6% up and 360 is about 12.8% down, so roughly 1.5:1. I also accept that 490 isn't a value target. It's where the plan sells. If I want a rebuild, it has to rest on the trader's own max-price ranges of $378-$447, which put 413 inside the range, and not on a round trip to the 10 EMA.

I'm dropping "another 7% down means the hike is priced in," since nothing in our data supports it. I'm also dropping "a retest of 411-417 is very live." Neutral is right that my claim and Conservative's "retests fail" are the same kind of unsupported assertion. I also used RSI 36 and the z-score both ways, and I'll take oscillators out of the timing argument. Dated, price-based rules do the work.

So I accept selling half the excess once the 10/14 bar is final, the two-close trigger under 436.79, the path budget at both 360 and 335, and the 10/28 backstop. The backstop may sell the last piece right at 411-417, which I said I didn't want to sell into. By then it's a small piece, and Conservative is right that hold-at-411-417 only makes sense for a book already at target.

I'd verify the backstop date, though. Polymarket says the October market resolves on 10/29, but my recollection is that the Fed's October meeting ends on the 28th, with the statement at 2pm that day. I can't check that here, and a resolution date can trail the meeting. If the announcement is on the 28th, a backstop at the 10/28 close lands after the event. I'd set it at the 10/27 close unless someone confirms the date. The monthly close is Friday 10/30, so the 387.79 regime check and the rate decision land within a couple of days of each other. That favors being at target before the meeting, not after.

On 50% versus 55%, I'll stop spending credibility there. Five points of standard weight on the illustrative 3% book is about 15bp of book, and a 10% move on that is about 1.5bp. Let the stress budget set the number. To make that usable, take the loss you can tolerate on the gap case and divide it by 0.194 for the 360 case and by 0.25 for the 335 case, then use the smaller weight. Thirty basis points gives about 1.55% of book at 360 and 1.2% at 335. If the budget binds at 50%, go to 50%.

I'm also fine with 387.79 in the stop field on a closing basis. It's the number that limits damage on what's left, and 513.10 only says the bearish view was wrong. I'd write both in plain words so nobody reads 387.79 as the line where the sell thesis dies.

Here's where I still push. The first is the rebuild. No standing bids before the FOMC is fine, and I'll drop the starter add too, since zero is defensible. But a plan that only says how to sell is half a plan, and dry powder only has value if its use is written down now. So I'd keep the original ladder, 410-417 toward about 80% and 380-395 toward full, and start the clock after the FOMC and the 10/30 monthly close. That also resolves the contradiction in the original plan, where bids at 410-417 sat next to "add nothing before 10/29." Two conditions apply: the gate must have passed, and there must be no close below 387.79. I wouldn't wait for a weekly SuperTrend flip. That stop is at 599.51, about 34% above the close, and waiting for it means rebuilding far higher than we sold. I also wouldn't buy back above where we sold without new information. The V-bounce is the scenario where this costs us, and I'd accept that cost.

The second is the gate. Defaulting to failed is a sensible fallback, but it shouldn't be how we operate. The July-quarter 10-Q has been public for weeks, so pull DSO, DIO and contract liabilities today, along with the 10-year, CPI and China share. Conservative is also right that I over-read the $1.8B working-capital residual. Investing outflows were $3.15B against $1.99B of capex, so about $1.2B went to something other than capex, plausibly short-term investments. The balance sheet can't settle it, but the filing can. This matters more than the 50-versus-55 argument, because the gate decides whether the cut at 387.79 is to 25% or to nothing.

The third is what this plan is. Selling half the excess immediately, after a 17.6% drop in seven sessions, while retail posts "dry powder ready" under 400, is the uncomfortable part, and I'd rather own it than dress it up with a limit order at 464 that may never fill. The plan costs something in every scenario, but each cost is bounded. The risk-taking is in cutting before a binary event, holding the powder, and committing in advance to when to use it. That's a better risk-adjusted bet than carrying full weight, a 1.59 beta and a 4.5% daily range into a meeting where the market has 30% on a hike.

The plan is to sell half the excess over target at the final 10/14 bar. Sell the rest on a bounce into 464-490 or a second close under 436.79, and finish by the 10/27 close, or 10/28 if the date is confirmed. Go into the FOMC at 50-55% of standard, with the stress budget setting the weight and no bids. Cut on a close under 387.79, with the gate setting the size. Rebuild only after 10/30, on the written ladder, if the gate passed. And 513.10 is where I'm wrong. Aggressive Analyst: We're close enough now that the useful thing is to be exact about what I'm giving up, what I'm keeping, and where our own rules still collide.

First, what I'm giving up. I contradicted myself on weight. I said to take the smaller of the two budget numbers, then said to go to 50% if the budget binds at 50%. On my own illustration, 30bp divided by 0.25 is 1.2% of the book, which is 40% of a 3% standard weight. So the rule is the lesser of the thesis weight and the budget weight. The view supports 50 to 60%, and the budget is a ceiling on it. Neutral is right that a budget landing under 50% means the position was too big before AMAT fell, not that the thesis wants 40%. The plan should say that so nobody reads it as a new call on the stock.

I also overstated the hike. Polymarket has 68% on a hold and isn't a rate-futures market. I'm not saying the hike is coming. Going in at 50 to 55% is insurance with a premium, and it's partial because the hold branch is the likeliest one and I still want to own it.

Conservative, you're right on the gate. A balance sheet dated July 26 can't cancel what a close under 387.79 says about October. A passed gate should still mean a step down to about 40% on a daily close under the line, unless you're already there. Failed or unresolved means 25%, and a monthly close under it on 10/30 means 25% regardless.

Neutral caught the 380-395 rung, and the arithmetic is worse than it sounded. A buyer at 395 with a closing-basis cut at 387.79 has a stop 1.8% away on a stock whose average daily range is 4.5%. That's a coin flip with the stop built in, so I'm dropping it. The first rung is 405 to 417, capped at 80% of standard or the stress budget, whichever is less. Nothing beyond that until the November print, which is unconfirmed.

One of my own rules also collides with another. I said a close back above the 487-491 cluster means stop trimming, and I also agreed to finish by the 10/27 close regardless. They can't both govern. The bounce tranche should rest as limit sells across 464 to 490, so a gap through the cluster fills at the open and there's nothing left for a stop-trimming rule to cancel. If anything is left, the backstop wins, because it exists for the meeting and a rally doesn't change the meeting. I'm dropping my rule.

Now where I'm holding.

The first is the gate test. Everyone wants the filing pulled, and Neutral said to write the pass rule first, but nobody has. Here's mine, fixed before I've seen a number. A pass means year-over-year growth in receivables and in inventory each comes in no more than about ten points above the 24.8% revenue growth, or the excess is covered by growth in contract liabilities. Anything that isn't a clear pass counts as a fail. Change the thresholds if you like, but before anyone looks. It doesn't need to hold up the sale at today's close, since it only decides the size of the cut at 387.79 and whether any rebuild is allowed.

The second is the zones. Conservative, I know you prefer conditions without a ladder, but Neutral's answer works. A ladder written now isn't live until the conditions are checked after 10/30, and a limit order that never fills costs nothing. It's the same reason to write the gate test first. A rule that says "rebuild when conditions look right" gets interpreted by whoever is watching the tape at 350 or at 500, which is the failure this structure is built to prevent. Write the zone, switch it on after 10/30, and size it off the same budget.

Since Neutral said the plan costs something in every scenario, let me put numbers on that. On the illustrative 3% book, a V-bounce to 540 costs about 20 to 25bp against doing nothing, and a slide to 360 saves about the same. The tails are roughly symmetric, so the edge comes from where the probability sits. I think it sits with selling: a weekly downtrend, a price at the top of every max-price range the trader built, and a binary event two weeks out with no rate data in hand. If I'm wrong, the cost is bounded, and we still own half the position plus the right to rebuild.

So the plan in plain terms. It applies only to holders above target. Someone at or below 50% has nothing to sell, and nothing here argues for shorting. Sell half the excess into today's close. Rest the remainder as limit sells at 464 to 490, or sell into the close after a second close under 436.79, and be finished by the 10/27 close either way. Go into the meeting at 50 to 55% of standard, or the budget weight if lower, with no bids and no adds. A daily close under 387.79 cuts to 40% if the gate passed and 25% if not. 513.10 is where I'm wrong. After 10/30, rung one goes live at 405 to 417, but only if the gate passed, the meeting is known, the monthly close held above 387.79, and the 10-year and CPI are in hand.

What's left is lookups, not debate: the final 10/14 close, the filing figures, the 10-year and CPI, China revenue share, the meeting date, and the earnings date. Do those today, and the rest follows from the rules we've written. Aggressive Analyst: I'll take most of what's on the table, because the last round improved the plan, and then push in two places.

Neutral, you caught me on the edge claim. I called the tails roughly symmetric and then said the probability sits with selling, and I can't put a number on that. Take it out of the written plan. I'll keep your framing instead: half size is a sizing answer to a lean you can see but can't price. What's behind the lean is in the data, not in a forecast. Price is at or above the top of every max-price range the trader built, the weekly SuperTrend is down, and hike odds went from 15% to 30% in a week. The monthly regime being up is real, but its stop sits 15% below the price. That says the multi-year trend is intact, not what the next three weeks hold, which is why we keep half. On the 68% hold, I'll push once. That number was 83% a week ago, so a hold the market has been marking down is the likeliest branch but not an obvious rally trigger. It's still a good reason not to go below half.

The mirror rung is the most aggressive idea in the plan, and I'm taking it. A plan that only buys weakness is biased, and the way we're wrong here is a V. Your own twelve-session run from 417 to 542 shows how fast that can happen. If the meeting is a hold, the gate passes and the stock closes above 513.10, the daily trend has flipped, and that's the new information I said I'd need before buying above where we sold. On my illustrative book, paying about 10% more on roughly 0.45% of the book costs a handful of basis points. I'd write it as one cap instead of "never both." Until the November print, the book doesn't go above about 70% of standard, whichever rung fills. That also covers the case where we buy at 410-417, the stock rips, and the mirror rung finds nothing left to do.

On the gate, I'm moving off my ten points. Conservative was right about cost of goods: 4,529 against 3,740 is about 21% growth. Neutral is right that inventory leads shipments in a ramp. So the pass rule is receivables no more than five points above the 24.8% revenue growth, inventory no more than eight points above roughly 21% cost growth, and contract liabilities counted only as a dollar-for-dollar offset. Fix those numbers before anyone opens the filing. At the threshold, days sales outstanding are up about 4% and days of inventory up about 6-7%. That's a real test and neither a free pass nor a trap. If the filing isn't pulled by tomorrow's close, the gate counts as failed. That's a clock on us, not information about the company.

The same discipline has to apply to "adverse rate data," which Neutral wants to use to pick the spot in the 50-60% band. Define it now. If the 10-year is at or below its level a week earlier and the latest CPI year-over-year isn't higher than the prior print, aim for 55-60%. Otherwise aim for 50%. Without that, whoever is watching the tape will pick the number that matches their mood. Given that the headlines already have yields rising on hot inflation, I expect the low end, and I'm fine with that.

Everything else I accept as written, and the 10/27 backstop is a good example of why. I can't confirm the meeting date from here, so I'd rather lose a day than miss the event.

  • First sale: Sell half the excess over 55% of standard weight into today's close, or the next close if today's has passed. Treat about 447 as the entry reference, since 464.24 is stale and retired.
  • Second sale: Rest the remainder as limit sells across 464 to 490. If the stock closes under 436.79 and is still under late the next session, sell into that close. Whatever is left goes at the 10/27 close.
  • Before the meeting: No bids and no adds. The weight is the lesser of the thesis weight and the caller's stress budget at both 360 and 335. A budget under 50% is a sizing correction and not a new view on AMAT.
  • Stop field: A closing-basis cut under 387.79, executed into the close if the stock is under it in the last half hour. It cuts to 40% if the gate passed and 25% if it failed or is unresolved. A monthly close under it on 10/30 means 25% regardless. This is a damage limiter. The line where I'm wrong is 513.10.
  • Rebuild: After 10/30, one capped rung at 410-417, or the mirror rung above 513.10, up to 70%. It needs the gate passed, the meeting known, the monthly close above 387.79 and the rate data in hand.
  • Scope: This applies only to holders above target. Someone at or below 50% has nothing to sell, and nothing here argues for shorting.

The lookups left are the final 10/14 close, the filing figures, the 10-year, CPI, China revenue share, backlog or customer concentration, the meeting date and the earnings date.

I still back the Sell. We sell into the bar we can see instead of waiting on a 464 fill that may never come, we stay half-exposed for a business growing 25% a year, and we've pre-committed to buying on both the dip and the breakout. The conservative alternative is to wait for the tape to cooperate. The neutral alternative is to sit at standard weight and hope the 68% branch arrives. Our plan costs a few basis points in a V-bounce and avoids a gap-down at full size into a binary meeting. Being willing to act before the event, with a written way back in on either side, is the better use of risk.