Trading Analysis Report: COHR

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 15:24:55
• 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: Trim COHR, but do not exit or short. Confirm the earnings date and pull the implied event move first. Then sell down to about 50% of a standard allocation over the next two sessions, half each day (nothing if you are already there), and finish at 30% of standard by the print (35% at most, lower if the implied move is 13-15% or more), with a GTC stop-market under 252 resting now on every share above that final size. Work the remaining 15-20 points at 280-288 with a market-order deadline five sessions before the FQ1 FY27 print, put a second stop-market near 217 on a third to half of the core, and add no new money, re-adds or call spreads until the print can be read over the next 1-3 months.

Current Price: 263.32

Price Target: 250.0

Confidence: Medium

Time Horizon: 1-3 months

IV. Risk Management Team Decision

Aggressive Analyst

Aggressive Analyst: I'll open, since nobody else has spoken yet, and I'll answer the two objections I expect to hear.

My case is that the trim is the aggressive move. Holding a full position through an unconfirmed print isn't boldness. It's taking risk you aren't paid for. The trader's math puts a clean pass at about 18% and the expected value of sitting tight at roughly -3% to -3.8%. The stock has an 8.5% ATR and beta above 2, and it just fell 22% in six sessions. A stop can't help, because the weekly SuperTrend stop at 255.29 is only about 8 points away, less than half a daily ATR. That level will get tested by ordinary noise, and an earnings gap will jump straight over it. Cutting to 40-50% of a standard allocation means a 15% gap costs about 6-7.5% of a full position's value instead of 15%. That sizing is the stop.

Aggressive investors also need ammunition. The trim frees capital to buy at 241, or at 230-235 near the monthly stop at 228.67 and the July low around 221, if the selloff turns out to be sector-driven. That is a much better entry than 263 into an unconfirmed print.

Now the conservative view, which I expect to be "just get out entirely." I disagree. The business is real. Revenue growth accelerated from 17% to roughly 34% year over year in the derived Q4, and gross margin is up from 30.9% to about 38.5%. The weekly and monthly SuperTrends are both still up. The sector headlines say capacity is sold out. And the tape has already shown how violent the bounces are: the 264.41 close on September 3 was followed by 281.86 the next day, and the 266.50 close on September 14 was followed by 289.93 two days later. Going to zero at 263, on a day that closed at the lower Bollinger band with a daily z-score of -2.05, means selling the low of a six-day flush and giving up the right tail. The two-step plan sells some now, then sells the rest into 280-295, where the 10 EMA, 200 SMA, 50 SMA and Bollinger midline are stacked. That captures the likely bounce without needing it.

The neutral view will probably be "hold, it's oversold, the selloff is sentiment-driven, and 18.7x forward earnings is cheap." I'd push back on each part.

On oversold, the evidence is thinner than it sounds. RSI is 39, not below 30. Volume of 6.7M was below the ten-day average of about 7.5M, so there was no capitulation. The MACD histogram is still deepening, ADX is rising from 8 to 16.6, and the daily TD count is only at 5 of 9. Only the daily z-score is stretched, while the weekly is -1.33 and the monthly is positive. That is a flush that could easily run further, not a washout.

On the selloff being sentiment-driven, I agree that it is. But the trigger is AI funding fears, and Coherent is itself a heavy consumer of funding. Free cash flow was about -$1.0B last year: operating cash flow of $80M against $1.1B of capex. The company also raised roughly $1.9B of equity in the March quarter. Quarterly capex went from $104M to about $556M, and cash fell $431M in the June quarter with no capital raised. If funding conditions tighten, this isn't a self-funding story, and the rates backdrop is unfriendly. Polymarket puts no 2026 cuts at 97% and another hike at 86%.

On 18.7x forward earnings, that is $14.05 of forward EPS against $4.12 trailing, a 3.4x jump. The basis is unverified and probably non-GAAP. Trailing GAAP earnings may also be inflated, since operating income is nearly equal to net income despite about $3.4B of debt, and the March quarter showed a $297M investing inflow that looks like a one-off. On enterprise value, the trader's run-rate figure is 28-31x EBITDA, and the trailing figure is closer to 38x. Neither is cheap for a name where insiders sold about $8.5M in the last three months and the only open-market buyer, the CEO, last bought nearly two years ago. Bain's remaining stake of about 7.75M shares, worth around $2B, is also a possible overhang.

The neutral analyst will say I'm ignoring that gaps run both ways, and they do. The stock gapped from 288 to 326 at the August open. That is why the plan is a trim and not a short, and why new money stays flat instead of shorting. I'm not claiming a bad print is certain. I'm saying that with a 28x-plus multiple, negative free cash flow, and a thin cushion, the downside tail is fatter than the upside tail. At 40-50% sizing I can still take the upside, and I keep the cash to buy lower.

There are real gaps, and I'm not going to hide them. The earnings date isn't confirmed. Today's volume figure looks non-final, so the bar might change. Implied volatility isn't available, which is why the call spread is optional and capped at 10-15% of a standard position's risk budget. The plan says to verify those items, and I agree.

If you want to outperform, size your risk where you have an edge and keep powder dry for the entries that pay. Trim now, sell the rest into 280-295, finish several sessions before the print, and re-add into weakness. I support the Sell as a trim. Aggressive Analyst: I'll give back what I should first, because several of those hits landed.

You're both right that I mislabeled the sell zone. The 10 EMA, 200 SMA, 50 SMA and Bollinger midline run from 294 to 306, not 280 to 295. You're right that the re-add bids at 241 and 230-235 sit below the 255 line where the plan says to finish selling, so they contradict the plan. The weekly-close-plus-ADX trigger is too slow, and a daily close below 255 is better. The call spread is the least important piece, and I won't defend it blind. You're also right that "sizing is the stop" is only true once the second tranche is done. If the bounce doesn't come, the deadline sale is what delivers the target size, and I should have said so instead of claiming the plan doesn't need the bounce. I'll take a hard deadline with a market order behind the limits, a confirmed earnings date as step one, and re-adds only after the print.

I'll keep the dry powder. A trim with no plan for the cash is just de-risking. After the print you can read guidance, capex and financing together and buy with information, instead of leaving bids that fill when the news is worst. That's where the aggressive edge is.

Where I disagree is on the bounce history, and I think the neutral analyst has it wrong. You said both bounces "stopped short" of the moving-average stack at 281.86 and 289.93. Those are the closes one and two days after the lows, not the tops. The fundamentals data has insiders selling at about $306 to $309 on September 9 to 11, less than a week after the 264.41 close on September 3, which is above the whole cluster. After the 266.50 close on September 14, the stock closed at 319 on October 1 and 338 on October 6. So the last two tests of this floor returned roughly 17% and 27% within weeks, not 7%. The lower highs are real, since 427, 379, 338 is a bad sequence, and a third test may fail. But that history is a reason not to dump the bulk at the lowest print of a six-day flush, and not to park the exit order at 280 either. I'd work the second tranche across 280 to 295.

On the 25% gap, I'll accept the neutral analyst's 20% as a fair stress case. But the conservative's "fell 9% today on sector headlines" isn't gap evidence. The stock opened at 281.13 against a 289.98 close, about a 3% gap, and the rest of the drop happened during the session. That's the kind of risk a daily-close trigger can act on. An earnings gap is the kind it can't, which is why the two tools do different jobs. The trigger handles the grind and sizing handles the gap. The largest overnight gap I can see in the supplied data is August's 13% up. I can't see opens for every date, so 20% is a sensible stress case, not 25%.

On expected value, I'll borrow the neutral analyst's point. If -3% to -3.8% is the full-position number, the first 55 points of trim capture most of the benefit. Going from 45% to 30% saves about half a point of a full position, and those inputs aren't even reproducible from this data. The argument for going lower is tail loss. At a 20% gap, 30-35% costs about 6-7% of a full position, and 40% costs 8%. We're arguing over about two points either way.

I'll also drop the valuation fight. The $14.05 forward EPS is unverified, and the 38x trailing EBITDA is my estimate stacked on a derived EV. Even the trader's run-rate figure of 28-31x isn't cheap with free cash flow near -$1B, so I'm not setting size off it. And I overplayed Bain, since that 7.75M-share row is unlabeled.

On funding, I raised it, so I'm not minimizing it. But the conservative's best line, that a strong print with higher capex guidance isn't a clean pass, is what the 18% already says. The stock is down 38% from its peak and 22% in six sessions, so some of that is priced. And the cash math ignores the $1.9B raise in March, which shows the company can access capital. That's a dilution risk, not a solvency risk, and the neutral analyst is right to say so.

Here's where I land. Confirm the date first. Take 30 to 35 points off over today and tomorrow, since the bar looks non-final. Work the rest at 280 to 295, and sell at market three or four sessions before the print if it hasn't filled. Finish the trim immediately on a daily close below 255. End at 35 to 40% of standard if the implied move is comfortably inside 20%, and 30 to 35% if it's near or above 20%. No new money, no pre-print re-adds, and no call spread unless I can see the chain.

The extra five points at the print are small. What I really care about is not selling the bulk at the low, just before a violent bounce in a stock that has shown three of them. That's what the staged trim protects, and I still back the Sell as a trim. Aggressive Analyst: We've converged enough that the useful thing is to say exactly where I've moved and where I haven't, because what's left is worth a point or two of a full position.

Conservative, you made the best point of the round, and Neutral built on it. I sized for the print and not for the interim. Carrying 65-70% of a beta-2 stock through weeks of 8-9% days, protected only by a rule that reads closes, doesn't hold up. A close rule on a stock that opened at 281 and finished at 263 sees the damage after it happens. So I'm taking Neutral's version. An intraday break below about 252 sends you to the final number. It clears the 255.25 wick from September 3 and should fill near 250, not in the 240s. I'm also taking Neutral's step of going to roughly 55% of standard over today and tomorrow, half each day, since today's bar may not be final.

Here is where I hold.

You said I can't call this a flush that might run further and also refuse to sell the bulk at the low. I can, and that's the argument for staging. I don't know whether the low is in, so I don't want to be fully committed either way. Selling 45 points over two days is a good chunk, as you put it. We're arguing about whether the next 15 points go at 263 or get worked.

On the second tranche getting its worst price in the paths that matter, I'd put numbers on it. The ladder is 15-20 points. If the stock grinds to 252, that's about 4% on 17 points, roughly 0.7 of a full position. If it gaps through to the 240s, it's about 1.5. If it bounces to 285, the same 17 points sell about 8% higher, which is about 1.4. That's roughly symmetric, and on the loss side the 252 trigger is the protection, not the ladder's fill price.

The "three more sessions like the last three" scenario doesn't run the way you described. The first down day of that sequence goes through 252 and trips the trigger. The extra shares are gone at a few percent below today, not 25%. The core 30-35 points carries that interim risk in every plan, including yours at 25-30.

On final size, I'll land at 35% as the base, with 40% only if the chain shows an implied move clearly inside 20%. I'd treat the implied move as a dial between 30 and 40, not a trigger to go lower. After a 22% drop in six sessions, the chain will be priced for fear, and implied earnings moves usually run above realized ones anyway. Giving up the last five points isn't free either. A third of a position participates in a 13% gap like August's for about 4.5 points. I don't know whether that August gap was earnings, but the upside tail exists. Selling more because the premium is richest after a selloff means sizing off the fear itself.

On the ladder, I'll drop 295 as the main target. You're right that 294-306 is a stacked supply cluster, and 280-290 is where the next-day closes landed. But this ladder lives for weeks, not sessions, and the same tape went from 264 to roughly 309 in four sessions in early September. That's inferred from the insider sale prices, not confirmed closes. So most of the ladder goes at 280-290, with a small last slice at 295-300 as the upside kicker.

I'll grant the March money looks spent and another raise is plausible. But the $1.9B raise and the preferred conversion are both inferred from balance-sheet arithmetic, so I'd carry that as a risk, not a forecast. It's one reason I'm not arguing for the high end, and a reason nothing gets re-added before the print.

The deadline should be five sessions before the print, or ahead of a peer print if that's earlier, but only once someone verifies those dates. A market order sits behind the limits. If the print is inside about seven sessions, skip the ladder and go straight to the final number.

My plan is to confirm the date and pull the implied move first. Sell to about 55% of standard over today and tomorrow, then work 15-20 points mostly at 280-290 with a small slice at 295-300. An intraday break under 252 takes you to the final number immediately. The final number is 35%, with 40% if the chain is comfortable. There's no new money, no pre-print re-adds, and no call spread until someone can see the chain. Scale the final number to what you can absorb at a 20% gap, since I can't see your starting size or loss budget.

The trim only pays if you use the cash. After the print you'll have guidance, capex and financing in hand, and that's where the aggressive edge is. Standing bids at 241 would have filled on the worst news with none of it. I still back the Sell as a trim. Aggressive Analyst: Most of this is settled, so I'll say what I'm taking and then spend my time on the one place I think the plan still leaves money on the table.

Neutral, you're right on step one. Ten points between 55 and 45 is a near-symmetric coin flip with a small payoff, and the things that would settle it (whether today's bar is final, tomorrow's open, the date, the chain) all arrive within two days. That's what staging is for. I'll go to about 50, half today and half tomorrow, and zero if the caller is already there. I'm also dropping 40 entirely. I borrowed the "inside 20%" test last round and it was wrong for the reason you gave: 20% is the stress gap, not the implied move. I'll take the stop as a resting order sized only for the shares between where you are and final, resized every time a ladder rung fills. Selling the same shares twice is an easy mistake to make by hand on a stock moving 8% a day.

Conservative, I'll take your formula of loss budget over roughly twice the implied move, and I'll defend it. A straddle prices about 0.8 of a standard deviation, so twice the implied move is about 1.6 standard deviations. That's roughly a one-in-twenty downside if returns were normal, and earnings gaps have fatter tails than that. It's already a real tail, so I don't want a second haircut stacked on top. Your plan has a conservative budget, a conservative multiplier, and then "default to the low end" and "err low on the end state." Three conservative choices in a row is a bias, not a decision rule.

What you're protecting with that bias is five points. A 35-point position shrinks to about 32 on a day like today, and a 30-point position grows to about 33 on a day like October 1. The 30-versus-35 argument is smaller than one session of drift. It also only matters if nobody can read the chain, and that's a one-minute lookup for anyone with a brokerage login. If it really isn't available, the proxy is how the stock moved on its last few reports. None of us has that either, but it's also a lookup. Use 30 as the placeholder for a day, capped at 35, and move on. Neutral already took apart the "bet against your own thesis" line. My thesis was to hold less than the trader's 40-50% because of the tail, not to hold nothing.

Now the part I think we've undercooked. The stop sits about 11 points below the close. The first ladder rung is about 17 points above, and the 285 area is about 22 above. In a driftless walk, the stop trips first roughly 60 to 65 percent of the time. That's rough, since it ignores how fast this tape's bounces have been, but it says that in most paths this plan simply ends at final size around 250 and never sells a bounce. That's fine, and it's why I won't fight over 288 versus 295. The ladder is the minority path, and the expected cost is about the same either way. A couple of points at 295-300 is a free option and nothing more. The plan's money isn't made by the ladder or the stop. It's made on the other side of the print.

That's where the aggressive edge is, and so far the plan says only "rebuy deliberately with guidance, capex and financing in hand." That's a mood, not a rule. I'd write the rebuild criteria now, while nobody's emotional, from what this data says matters:

• Does operating cash flow recover against capex? The derived June quarter was about plus $70M against roughly $556M of capex.
• How is the build being financed? That includes whatever the $741M jump in current liabilities turns out to be, and whether there's an equity raise.
• Does revenue guidance extend the acceleration from 17% to about 34%?
• Does gross margin hold near 38%?

If those come back clean and the stock sells off anyway on sector funding fear, that's the entry the trim was for. It beats any standing bid at 241 because you'd know what you're buying. Rebuild in steps toward the trader's 60-70% instead of all at once. If the print includes a raise, wait to see whether it prices and gets absorbed. If the print is clean and the stock gaps up 15%, you hold a third, take about five points, and don't chase. That's the cost of the insurance, and it's worth paying.

So here's where I land, in terms the caller can scale to their own book.

1. First, confirm the earnings date, pull the implied move, and check whether today's bar is final.
2. Sell to about 50% of standard over today and tomorrow, half each day, or nothing if you're already there.
3. Rest 15 to 20 points at 280-288, with a couple of points at 295-300.
4. Rest a stop under 252 for whatever gets you to final, and resize it as rungs fill.
5. Final size is your loss budget divided by about twice the implied move, capped at 35, with 30 as the placeholder until the chain is visible.
6. Set the deadline at five sessions before the print, or before any verified peer print if that's earlier, with a market order behind the limits. If the print is inside about seven sessions, skip the ladder and go straight to final.
7. Put no new money in, make no re-adds before the print, and skip the call spread unless you can see the chain.
8. After the print, rebuild by the written rules above.

I still back the Sell as a trim. Aggressive Analyst: We've landed on the same plan, so I'll only go after the places where the last two rounds added a rule that works against the plan's own logic.

Conservative, you're right that the stop is a regular-session order and tomorrow can open through it. But look at what it touches. It only governs the 15 to 20 points between 50 and final. If tomorrow opens at 245 and fills there instead of 252, that's about 3% on seventeen points, under half a point of a full position. The core is already sized for the gap, which is the job of the formula. So I don't read the overnight risk as a reason to shave final size again. Thirty as the placeholder and 35 as the ceiling is the deal, and I'll keep it.

You're also right that the quarter is over and management already knows the funding answer. To me that is the best argument for buying after the print, not before it. The uncertainty collapses in one session and we all read it together. Your 0.7B cash figure only repeats the June pace, and the $741M jump in current liabilities could cut either way, so I wouldn't weight dilution on feel. But I take your runway test over my operating-cash-flow test, since no clean quarter closes $70M of operating cash against $556M of capex. I'm also dropping 60-70% as a rebuild target. Neutral is right that after the print the cap comes from stop distance, which gives the high 30s to about 50. That's the first rebuild ceiling anyone derived instead of borrowing from the trader.

One caution on the implied-move formula, because it's easy to misuse. If someone reads a straddle on an expiry three weeks out, it includes the whole interim, which is the 5-6% days the stop and the sizing already cover. Doubling that number would push final into the mid-teens on a figure that's mostly not about the print. Use the expiry closest after the print, or the event move the platform backs out.

Neutral, your rally-branch catch is right, but the rule as written doesn't work. A clean print on a stock in the 260s or 270s that gaps 15% opens at roughly 300 to 320, inside or above the 294-306 stack. "Reclaims the stack" is already met at the open, so it tells you nothing. The usable test is a hold: take the first step only after the stock stays above the print-day low for a second session, or retests the cluster from above and it acts as support.

Where I disagree is rule six. We dropped the close-below-255 trigger because a close is only observed after the fact. A close below 229 has the same flaw, and it's worse there. By then the stock has been falling 8% a day, and the fill is the next open, probably in the 215-220 range. Also, a daily close under 228.67 doesn't by itself flip the monthly SuperTrend. That tier turns on a monthly bar, and the report doesn't say how the tool treats a partial month. So "all three tiers are down" is an assumption.

The only precedent we have for this zone is the July low. The close was 222.05 on July 29, and 288.14 on August 3, which is about 30% in three sessions if I'm counting right. That's one data point, not a law, but the rule would have sold the core at that bottom. I'd keep the second line, but build it like the 252 stop: a resting stop-market just under the July intraday low of 220.68, call it 217, cutting a third of the core. It sits below the wick and executes at the broker, with no one deciding at the lows. If the stock trades through the cycle low, the structure is broken and I'm fine being out of a third of what's left.

The book-level check is right too. If trim cash goes into LITE, CIEN or AAOI, it's the same funding trade in a different ticker, so judge the rebuild against the whole book.

I still back the Sell as a trim. It's below the trader's 40-50% at the print because the trader sized the event and left the interim bare, and this plan fixes that. It stays a trim, not an exit, because about a third of a position still gets paid if sold-out capacity shows up in the numbers. And the buying happens after the print, with runway, financing and guidance in hand, when the risk is something we can read and not something we have to guess.