Altair
Skip to content

Trading Analysis Report: COHR

Analysis date: 2026-10-21 · Model: claude-opus-5-5 · AI-generated, not financial advice.

  • Analysis date: 2026-10-21
  • Rating: Hold
  • Generated: 2026-10-21 11:57:05
  • 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: Hold

Executive Summary: Hold COHR, but cap the position at about 1.25% of the portfolio through the September-quarter print. If you hold more than that, place limit sells in the 252-262 gap now and sell any unfilled excess by the Oct 27 close; if you are at 3% or more, sell a third of the excess into the current bounce first. Pre-commit three rules: an October close below 228.68 steps you down to 1%, a daily close below 215 halves a holding (or exits a starter), and nothing is added back before the print. New money waits for a clean FY26 10-K read and a closed-bar hold of 215-222 on expanding up-day volume, and even then buys only 0.75%; the 1-2 month objective is 259, with gap risk toward 195.

Current Price: 237.37

Price Target: 259.0

Confidence: Medium

Time Horizon: 1-2 months

IV. Risk Management Team Decision

Aggressive Analyst

Aggressive Analyst: I'll open the debate, since neither of you has spoken yet. I'll answer the arguments I expect, because they're predictable.

The aggressive case for Hold on COHR is that Hold keeps you in the game at the point where the alternatives are worst. Selling or trimming means exiting after a 29.8% drop in 11 sessions, with the stock sitting on a 215-222 floor that has been tested three times, and with the monthly SuperTrend regime still up. Buying aggressively means paying up before a cash-flow print that could confirm or break the bear case. Hold with a pre-built scale-up plan avoids both mistakes. COHR is still up about 96% from 120.79 a year ago, and I want to keep that exposure.

The conservative analyst will lead with cash conversion, and I'll give them that one. FY26 operating cash flow was $80M against $787M of net income, and free cash flow was roughly negative $1B. That's the real weakness, and it's why the plan adds nothing before the print. But look at what capex did. It went from 7.6% to 15.5% of revenue while revenue growth accelerated to roughly 34% in the derived June quarter and gross margin hit about 38.5%. TrendForce sees an AI fiber supply crunch through 2028. A company spending $1.1B on capacity into a shortage looks different from one burning cash on a dying business. The balance sheet funded it, with a 2.43 current ratio, debt-to-equity of 0.32x and $1.16B in cash. The honest risk is that the working-capital build is inventory that gets written down if AI spending digests, and I can't resolve that from these tools. That's why the 10-K check gates the first tranche. I'd gate it, not run from it.

On valuation, 52-57x run-rate earnings is rich, no argument. But earnings went from negative $0.52 per share to $4.12 in a year, and the stock has been range-bound between roughly 215 and 440 for six months while earnings kept compounding. The denominator is growing into the price. I won't lean on the 17x forward P/E, because it's probably on an adjusted basis.

The technical bears have the trend right. The weekly and daily SuperTrends are down, the 50-SMA is below the 200-SMA, ADX is 35, and OBV confirms the selloff. But if you wait for trend confirmation, the weekly SuperTrend flips at 395.74, which is 67% above here. Insisting on that confirmation means buying after the money has been made. The plan doesn't wait for it. It asks for something achievable: an October month-end close at or above 228.68, or a 215-222 retest that holds on expanding up-day volume.

Insider selling is the point I'd dismiss fastest. About $42M of sales since late October 2025 is roughly 0.09% of a $46.7B market cap. The CFO sells roughly monthly, which looks like a plan. Bain's remaining stake is a real overhang, but at about $1.8B it's under 4% of market cap. I'd rather see insider buying, but its absence isn't a thesis break.

Macro is the same story. The October FOMC is priced at 86% hold, with the hike probability down about 20 points on the week, recession odds are 8%, and an AI bubble burst in 2026 is 9%. That tail is rising, and I respect it, but it's small, and the stock has already fallen 44% from its closing high.

To the neutral analyst, who will probably say wait, I'd say waiting isn't free, and the plan knows it. The risk-reward is decent. From the 228.68 entry reference with a 215 stop, you risk about 6% to target 252-262, a gain of 10-15%, and the 288-295 cluster is 26-29% above. That's roughly 1.7 to 2.4 times your risk to the first target and over 4 times to the cluster. The 215 stop also isn't an arbitrary one-ATR stop that noise will clip. It sits under three dated lows and the 215.09 lower Bollinger band, so it's structural.

I'd also push back on the idea that the earnings gap is a coin flip you must avoid. Optics peers are rallying into their prints, and Susquehanna's $350 target, even after the cut, is about 47% above the current price. I'm not claiming an edge. I'm saying the skew isn't obviously negative, and the sizing makes that question affordable. A 20% gap costs 0.25% of the portfolio, and a second tranche only comes after a clean print. That is a bold structure that pays when the thesis works and survives when it doesn't.

Here's where I'm less cautious than the plan. Existing holders should trim only a slice at the 252-262 gap and keep a core for the 288-295 cluster, because a stock that bounced to 347 and 379 after the last two tests of this floor can run again. That is two episodes, not a law, and I won't pretend otherwise.

I'm also not leaning on the retail posts. The labeled StockTwits split of 9 bullish to 0 bearish is thin, repeated and partly unverified, and the Nvidia investment claim is unconfirmed. Leveraged call buying at these levels is a reason for humility, not a signal. The case stands on the accelerating revenue, expanding margins and the structural floor, with the cash-flow question gated and sized.

My position is Hold, keep exposure, add on confirmation, and respect the 215 stop. Aggressive Analyst: Let me start with what you've earned, because the plan I opened with has real errors.

Conservative, your arithmetic is right. From 237.37, the 215 stop is 9.4% away and the 252-262 gap is 6 to 10% up. That's roughly 0.66 to 1.1 times risk, not 1.7 to 2.4. The "over 4 times" to the 288-295 cluster is really 2.3 to 2.6 from here, because I ran the numbers off a 228.68 entry nobody can transact at. Neutral's point is sharper still: the stronger the October close, the worse the entry, so the month-end close is a bad buy trigger and I'm dropping it. I'm also dropping "still up 96% on the year" as a reason to hold. It's an anchor, and the right test is whether you'd own this size today. And 215 is the one-ATR line, with 217.83 printed yesterday, so it isn't structural protection. It's a line where I act on a daily close, and size is the real defense.

Here is where I think the conclusions still don't follow.

Start with the lower highs (440, 439.68, 391, 342). They argue for where to take profits, not for being out at the floor. If the next rally fails at the 288-295 cluster, well below the October 5 high of 341.76, it still pays 21 to 24% against a 9.4% risk to the stop. You don't need new highs for this to work. You need a rally that tops lower than the last one.

Selling now is also the worst place to cut. The stock is down 29.8% in 11 sessions and sits in the bottom 15% of its Bollinger range. It opened above the prior close and finished up 2.4%, with a gap 6 to 10% overhead. Conservative, you're right that the z-scores aren't at minus 2 and RSI shows no divergence, so I'm not calling this oversold. But limit orders in the gap beat a market sale at the lows, and Neutral's ladder does that. Meanwhile October hike odds fell about 20 points on the week to 12%, with 86% priced for a hold. The Fed risk you're asking holders to flee is receding.

On the trigger being a coin flip, sure, the stock could close either side of 228.68. But it's the monthly regime line, a close below it puts all three SuperTrend tiers down, and it prints October 30. That's after the FOMC and the hyperscaler capex commentary and before COHR's own print. You see the macro resolve and cut before the cash-flow number lands. I'll concede Neutral's point that someone well above the size band shouldn't wait that long.

On the fundamentals, you've both landed real hits, and I'll give ground. Non-cash current assets rose 91%. Even annualizing the derived June quarter, revenue is only about 41% above FY25, so the build outpaces growth and isn't just a growth-rate artifact. Gross profit rose about $208M from the September quarter to the derived June quarter, while net income was flat at $226M and $223M. I can't explain that gap from these tools, so I'm retiring "earnings compounding." The print has to show the margin reaching the bottom line.

I'd still push back on the runway. Neutral is right that cash fell $430M, not $486M. Debt-to-equity is 0.32x, net debt is about 1.6x EBITDA, and there's market access at a $46.7B cap. A $2B raise would be about 4% dilution. That's painful but survivable, and it's why I'd watch the capex guide, not run. Derived operating cash flow also went from about negative $94M in March to about positive $70M in June while capex doubled. That's one derived data point, but the direction is recovery. Neutral's refinement is the right gate: inventory and receivables growth against revenue, plus the capex guide, not a single quarter's operating cash flow. The 10-K check is free, so do it today.

On insiders, Bain's remaining 7.75M shares is about one session of recent volume of 6 to 10M. That's an overhang you can digest. The seven sales totaling $8.2M came at $268 to $324, and there are no filings in the past week of this drop, so nobody is dumping into it. The lack of buying is real, but it's not a thesis break.

I'm also dropping Susquehanna's $350. Targets follow price down. And I'll flag that the 9% AI-bubble odds sit in a $2.5M market. Neutral noted that those odds barely moved while the stock fell 30%. That tells me the selloff looks more like positioning in a 2.11-beta stock than new information, and positioning can unwind too. That's an argument, not evidence, but it cuts both ways.

On sizing, I'll take Neutral's band: at or under about 2% of the portfolio, hold through the print. A 30% gap on 2% costs 0.6%, and 30% is the gap I'd stress against after this tape. But Neutral, you called 1.25% a bucket cap across AI optics and then said to hold if COHR alone is under 2%. Those two don't fit together, so pick one. I'd pick 2% for COHR and look at the basket separately. Above 2%, I accept limit orders in the gap and a sell-half backstop on October 28. I'd rather the pre-committed response to a daily close below 215 be a halving than a full exit. Below 215 there's thin structure down to the 170-210 range, so that's the honest risk.

For new money, Neutral's retest-only trigger on 215-222 with expanding up-day volume is better than mine. But I'd keep the first tranche near 1.25%, because a 30% gap on that costs about 0.375% of the book, and a 0.6% starter is too small to matter if the thesis works. Conservative, your 292.15 re-entry is the "buy after the money is made" problem. It's 23% above the price, at the bottom of the resistance cluster, and the weekly flip is 67% higher at 395.74. And zero for new money while existing holders keep 1.25% fails your own test. Either the position is acceptable or it isn't.

So I'm staying with Hold. I'll keep the exposure inside a size band, add only on a verified retest after the 10-K check, sell into the gap, and honor a daily close below 215 as a halving. The cash-flow question is real, but a 0.4 to 0.6% book hit is what we pay for a seat at a floor that has held three times, into a print that could confirm the thesis. Aggressive Analyst: Let me bank the concessions first, because they're real. I'll take Neutral's two-number structure, COHR alone plus a basket cap. The sentiment data shows COHR trading with LITE, AAOI, GLW and MRVL, so Conservative is right that a holder with other optics names carries more of this risk than the COHR line shows. I'm dropping both thin Polymarket contracts, the 9% bubble and the 8% recession. And for anyone above the ceiling, the decision date is the October 28 close. October 30 is a Friday, so a month-end rule can't act until Monday.

Now size, where I think Conservative's 1% is the number that doesn't hold up. It comes from dividing a 0.25% loss budget by a 30% gap. But 30% in one session is about three and a half ATRs, and more than double the only earnings gap anyone has cited, the 14% on November 6 last year. Yes, that gap launched from 135, and yes, volatility is higher now, which is why I'm stressing at 30 and not 14. But if you size for a tail twice as big as anything observed, you're underexposed in every other outcome. At Neutral's 1.5%, the observed 14% gap costs about 0.2%, inside the trader's own budget, and the doubled stress case costs 0.45%. I'm moving down from 2% to 1.5% and no further.

On the gap math, Conservative says a gap to 195 makes the 288-295 payoff only 1.2 to 1.4 times the loss. Those are two different branches, though. The loss happens in the world where the print is bad and the gain in the world where it's good, so the ratio only means something with a probability attached. At 1.2 times you break even winning about 45% of the time, and at 1.36 about 42%. Everyone here keeps calling the print a coin flip, and a coin flip clears that bar. The good branch doesn't always travel all the way to the cluster, so it isn't free money, which is why I accept the size cap. On noise clipping the stop, the plan's trigger is a daily close under 215, not an intraday touch. All three tests held on a close, at 219.65, 222.05 and 231.80. Yesterday printed 217.83 and closed almost 14 points higher. A close-based line isn't the same thing as a line one ATR below price.

On cash flow, Conservative is right that a $70M quarter is under half of last year's quarterly average on a bigger revenue base. Neutral is right that the number can't tell healthy working-capital growth from a stuffed warehouse, which is why the 10-K check comes first. I'd push back on "flat earnings," though. That compares the September quarter with the derived June quarter. The last three quarters are 147, 191 and 223, up about half from the trough. The September figure is the outlier I can't explain, so I'm not reviving "compounding," but endpoints are a selection too. On dilution, I'll grant the window can close. Even so, a $2B raise is about 4% of the company at today's cap and about 6% if the stock is another 30% lower. That's a haircut, not a thesis break.

On rates, I'd keep the deep contracts, since they're the only well-traded macro data we have. The October hike contract carries $12M, and it fell about 20 points this week while the stock fell 30%. So this drop isn't a rates story. The no-cuts regime is a standing condition, not something that arrived in October. What moved was AI-capex worry and positioning, and a sentiment shock can reverse faster than a rate regime. I can't prove that, so treat it as argument.

Conservative says there's been no flush, with RSI never at 30 and z-scores around minus 1.5. True, and I'm not calling it oversold. But the tape did something else. It printed three closes within about a dollar of each other, at 232.90, 232.02 and 231.80, with volume fading from 9.95M to 6.15M, then a gap-up open. Sellers stopped pressing without a capitulation, and a stall is a fine place to rest limit orders in the gap. A 30% drop also resets the bar for the print, so a result that disappointed at 340 may be merely fine at 237. That too is an argument, not evidence.

I'll concede Bain on mechanism. It would be a block, and the window plausibly reopens after the print. But a $1.8B block is under 4% of the market cap and one session of discount, and when it clears, the overhang clears with it. It's also one more reason to ladder sells into 252-262 before the print and not after.

On the month-end trigger, oversized holders now act on the 28th. For holders inside the ceiling, I'd make an October close under 228.68 a review, not a mechanical sale. They've already paid for the uncertainty through size, and a close a dollar under the line carries the same information as one a dollar over it. The 215 close is the line with an actual floor behind it.

For new money, I'll take Neutral's 0.75%, which fits the 0.25% budget at a 30% gap, and I'll stop arguing about 1.25%. Conservative, though, you said 292 is a bad place to buy and then said you're fine paying it after a clean print. Pick one. A clean print probably puts the stock at 270 to 292, which is 23 to 33% above a 220 retest. That is the price of certainty, and I'm willing to pay it for the second half. But it makes the retest the only cheap exposure on the board, so the 10-K check should happen today and the starter should be real. The starter also doesn't have to ride through the print. If the retest holds and the gap fills at 252-262, that first piece can do its job before the number lands.

So I'm staying with Hold, with numbers. The ceiling is 1.5% for COHR and about 3% for the optics basket, with limit orders in the gap and the 28th as the deadline for anyone over. A daily close under 215 halves the position, and new money starts at 0.75% on a verified retest after a clean 10-K. Confirm the earnings date today too, since no tool returned it. A 30% gap costs about 0.45% of the book. The upside is owning a name 44% off its high into a print where the bar just got reset. Aggressive Analyst: Three analysts who started far apart now agree on most of the plan. Nobody wants a holder sold to zero, nobody adds at market before the print, limit orders go in the 252-262 gap, a daily close under 215 means halving, and new money waits for the free 10-K read and a closed-bar retest. What's left is 1.25% versus 1.5%, which on a 30% gap is about 0.075% of the book. That's smaller than the error in any input we've used, so I'll concede what I should and defend the little that's left.

On concessions, Neutral is right about the Fed date. The report says the 28th to 29th, but the contract resolves on the 29th, and if the statement lands on the 28th, that day's close already contains the news. I'll take the 27th close as the deadline for anyone over the ceiling, since it's before the statement under either calendar.

Neutral's breakeven correction also holds. I paired a 21-24% win with an 18% loss. The upside I can underwrite on the print is the 14% gap, about 271 from here. Against an 18% gap down, that needs the good branch to hit roughly 56% of the time, so the print isn't a free option.

Conservative is right that the closes that held were 219.65, 222.05 and 231.80, so the closing floor is nearer 220 than 215. That strengthens the rule. A close under 215 is below every closing low in the range and below the 215.09 lower band, so it's a real break. I'll also accept Neutral's default that a weak October close steps a holder inside the ceiling down to 1% on Monday. It's a quarter-point step, pre-committed, which is what I wanted.

On size, I'll sign Neutral's version: 1.25% until someone reads the filing, 1.5% if it's clean, 1% if it's flagged. Conservative, you said the ceiling can rise in an afternoon but a gap can't be undone, and that's why I accept 1.25% until it's read. But I'd push back on stopping at 1.25% after a clean filing. My 30% stress gap is mostly the bad version of the cash story: inventory written down, a raise into weakness. A clean filing takes the worst version off the table, so the stress gap should come down to maybe 25%. At 1.5%, that costs 0.375%, the same as 1.25% at 30%. I'll own that this is a 0.375% budget, above the trader's 0.25%. But the gap assumption moved because we learned something, not because I picked the size first.

Conservative's line that weak cash can't support the bull case is true, but the bull case doesn't rest on cash. It rests on derived 34% revenue growth, 38.5% gross margin and a fiber shortage seen through 2028. Cash is what could break it, and it's gated. The stock is also 44% off its high with that question already open, so the price isn't ignoring it. We agree 57.6x isn't cheap. The question is whether the sized position has a positive payoff given what's already priced, and the sizing means I don't have to be very right.

You've both treated the Fed and the hyperscalers as hazards ahead of the print, but they're also information. Meta is in focus this week, and the other hyperscalers report before COHR on any reasonable calendar. As Conservative said, this drop wasn't a rates story. It was an AI-capex digestion scare, and capex guides from the companies writing the checks speak directly to that fear. They arrive while a holder can still act with limit orders working, so this isn't holding blind through an event.

The ladder also shrinks what rides into the print. If a relief rally fills the gap first, the excess is gone and only the core carries the number. A 14% good-print gap to about 271 would run through those same limits too. I'm not asking the print to pay 21-24% on the whole position, only for the core to survive and the trimmed slice to fill.

For new money, I'm at 0.75% after a clean filing on a closed-bar retest of 215-222. Conservative is right that confirmation may arrive above the low and shrink the reward-to-risk. In that case you don't buy. New money isn't forced to take the setup, and a miss costs only the opportunity. The second half comes after the print on inventory, receivables and the capex guide, and I'll pay 270-290 for that answer.

So I'm staying with Hold, and it isn't inertia. The ceiling is 1.25% now, then 1.5% or 1% after the filing. The basket is sized by what it loses, limits sit in the gap, anything over the ceiling is cut by the 27th close, a weak month-end steps down, and a close under 215 halves. The aggressive part isn't a bigger number. It's declining to sell a name 44% off its high, with accelerating revenue and margins, at a floor that has held three times, when the cheapest exit is a limit order 6 to 10% higher. Keep the seat, size it so a bad answer costs under half a percent, and let the print tell you whether to pay up. Aggressive Analyst: I'll take Conservative's trade on the filing, because Neutral's evidence closes it. The June balance sheet was public through the whole slide. Insiders were selling into it at 268 to 324 in late August and September, the stock was still 337 on October 2, and then it lost 30% in eleven sessions. A document the market had for six weeks can't be what moved the price, and it can't tell us what the September quarter looks like. So the filing is a veto, not a license. If it's flagged, the ceiling drops to 1% and there's no new money. If it's clean, nothing changes for holders and the 0.75% starter is permitted. Only the print lifts the ceiling. That costs me the quarter-point I was arguing for, and the 25% stress-gap case I made for it rested on a read that tells us nothing new.

On the ladder, Conservative is right that it's the upside and the 27th close is the protection. Whatever hasn't filled by then goes at market, and that's the price of a shot at 6 to 10% more. The first rung at 252 is about $15 above the close, less than one ATR of $21, so I think a fill is likelier than not. I haven't computed that, and it only helps in the rally branch.

I'd push back on the ratchet, though. The COHR position never moved. It's the trader's 1.25% from the start. What changed is the label on what it costs: 0.25% at a 20% gap, 0.375% at 30%, which is the stress we all agreed was more honest. The 0.45% was my 1.5%, which I've withdrawn. The 0.75% for the sleeve isn't an add-on to COHR. It's the cost of the other optics names, and Neutral has the right method: pick the group loss the book can take, derive the weights, and cut the other names before COHR. I'll also accept that "under half a percent" is a central estimate. A 40% fall from here would put COHR near 142, close to where Bain sold last November, and that costs 0.5% on 1.25%. I'd rather say that than defend a cap that doesn't exist. And "44% off the high" is a path, not a price, so I'm retiring it along with the 96%.

Conservative, your answer to "would you buy this size today" is 1.25%. That is the trader's number, so the case for selling is gone and the three of us are describing the same Hold. The "no edge, so smaller" argument is fair, but Neutral's reply holds: selling to zero is a bet on a bad answer too. I'm not claiming an edge on the print. The 1.25% is the cost of keeping a seat in a business whose derived revenue growth is about 34% and gross margin about 38.5%, while the cash-conversion question gets answered. I can size that cost, and the 27th, the step-down and the 215 halving limit it.

I still lean forward in three places.

  1. Don't sell at market into this tape. The daily TD-9 is complete, the MACD histogram has narrowed for the first time, and the close sits at about 15% of the band width, with a gap 6 to 10% above. A limit order is a better exit than a market order.
  2. Make the retest starter real. At 0.75% on a closed-bar hold of 215-222 after a clean filing, it's half the ceiling. If the retest never comes, we don't buy, and the filing check and the closed-bar rule shouldn't become reasons to never act.
  3. Pay up after the print. Neutral says 2% needs a reason beyond the print. I'd say the print is the reason, because cash conversion is the one thing capping the size. If inventory and receivables grow slower than revenue, the capex guide is funded and operating cash flow is turning, I'd go to 2% and treat the trader's 2.5% as a ceiling, not a target. Valuation at 57.6x GAAP and about 34x EV/EBITDA, a beta of 2.11 and AI-capex digestion all remain, and no print resolves them. Paying 270 to 290 for that answer is the price of certainty, and it's a better use of risk than leaning in blind at 237.

On dates, I'm taking the 27th close as the deadline for any excess, and I'd make the step-down the earlier of the Monday after month-end and the last close before the print. Verify the FOMC day and COHR's earnings date today, since no tool returned the latter.

I'm staying with Hold. That means a 1.25% ceiling going in, 1% if the filing is flagged, limits in the 252-262 gap now, and any excess gone by the 27th close. A weak October close steps a holder down to 1%, and a daily close under 215 halves the position. New money gets 0.75% at most, only on a verified retest. The aggressive part isn't the size. It's refusing to let fear sell the floor at market, and being ready to pay up for information once the print arrives.