Altair
Skip to content

Trading Analysis Report: CRDO

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

  • Analysis date: 2026-10-21
  • Rating: Underweight
  • Generated: 2026-10-21 11:23:37
  • 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 CRDO. Cut exposure to about 1% of the portfolio, which is a 0.3% loss budget divided by a 30% gap shock, or roughly a quarter to a third of a standard 3.5-4% weight. Sell about five-sixths of the excess into today's session with no price floor. Sell the last sixth on a single order: a $196 limit, with a market-on-close at Friday 10/23 as the backstop, or at the next open if the filing read fails sooner. If you are already at or below about 1%, hold. If you are flat, do not buy this gap. Halve the residual on a daily close below $159.83 (or the higher current weekly SuperTrend stop), and exit it on a weekly close below that level. Gap risk is managed by size, not stops. The horizon runs through NVDA on 11/18, the CRDO fiscal Q2 print (about early December, unconfirmed) and the 12/9 FOMC.

Price target: $170. - Primary method (valuation): base-case NTM EPS of about $3.40, which is about 21% above TTM $2.82 per the research plan, times about 50x. That multiple is the midpoint between the about 53-54x currently paid and the 46x the bear case would pay, giving $3.40 × 50 = $170. - Cross-check 1: if the unverified $525-535M outlook is the Q2 guide, Conservative's arithmetic gives about $0.74 Q2 EPS and TTM EPS of about $3.10. At 55x, that is about $170. - Cross-check 2: the technical failure scenario's first target is the $169.70-171.51 closes. - Bounds: 46x gives about $156; 54x gives about $184, the current price. - The target implies about -7.8% from $184.31 over 2-3 months.

Current Price: 184.31

Price Target: 170.0

Confidence: Medium

Time Horizon: 2-3 months

IV. Risk Management Team Decision

Aggressive Analyst

Aggressive Analyst: Let me open with what I think is the real misconception in this debate: that Sell at 184 is the timid call. It isn't. Trimming CRDO to about a third of a standard weight is the most aggressive move available, because it is the only one that leaves capital free to act on the next big dislocation.

Look at what this stock does. It went from 150 on September 14 to 220 by October 6, up 47% in 16 sessions. Then it fell 23% in 8 sessions to 169.70, and today it gapped back up 7.5%. Beta is 3.25 and ATR is about 8% of price. The profits in a name like this come from the swings, and you can only trade swings with dry powder. Whoever sat at full weight through the 302 peak had no cash at 150, and that is how a great business turns into a mediocre trade.

I expect the neutral analyst to say Hold. ADX is 11, so it's range-bound, the weekly and monthly trends are up, and the daily trend is down, so just wait. But Hold isn't neutral. It is a decision to carry a full 3.5-4% weight in a 3.25-beta stock into NVDA on 11/18, the CRDO Q2 print in early December, and an FOMC on 12/9 where Polymarket puts a hike at 78%. A repeat of September's 27% two-session gap would cost about 1% of the portfolio at standard size. At the trimmed 1.2% it costs about 0.3%. The neutral view treats that as caution, but it's an unpriced bet on the calendar going well.

Neutral will also point out that the gap hasn't filled, and that's fair. Today's low of 178.59 is above yesterday's high of 177.19. But that bar has 4.1M shares against a 6-12.5M norm. The 10-EMA at 185.35 is overhead, the 50 SMA at 201.84 sits above that, and the daily SuperTrend stop is 213.28. In a market with ADX at 11, you want to sell the upper part of the range, not wait for a trend that isn't there. The plan scales out at about 184 and again into 197.6-201.8, so it doesn't dump anything.

The conservative analyst will push the other way: exit entirely. Margins fell 400 bps, opex jumped to about 39% of revenue, EPS fell sequentially despite record revenue, OCF/NI is 70%, there is an unexplained $500-650M balance sheet item, insiders sold about $42M in 30 days, and Polymarket's AI-bubble odds rose 3.5 points in a week. Most of that is right, and it's why I'm selling. But it doesn't justify a full exit. Revenue is up 115% year over year with no debt and a 7.4x current ratio. Insider selling is largely programmatic. The CTO sells on a regular cadence, and the identical $101.45 prints in April look like sell-to-cover. The stock is also up roughly 5x since April 2025, so I wouldn't read that selling as a verdict on the business. The opex step might be partly stock comp from the May grants, and the balance sheet item might be an acquisition. Those are open questions the 10-Q can answer. A full exit bets that the answers are bad, and a Hold bets that they're good. The trim is the only position that doesn't need me to guess.

There's also real upside in the third we keep. A clean filing plus a Q2 beat is a plausible path to 205-220, which is 11-19% above here. That fits the plan's own 30-40% estimate of that path.

On valuation, the bulls will cite 19x forward earnings. That rests on a $9.71 forward EPS that nothing I have can verify, against trailing EPS of $2.82 and a Q1 run rate of about $2.68. The plan's 53-54x implies a base case near $3.45. Paying up for a tripling that hasn't been confirmed, right after margins moved the wrong way, is the crowded trade.

Sentiment supports that. The StockTwits read is 82% bullish among labeled posts, but that's only 11 messages. The tone is chasing: regret about missing 150, calls for 225 this week, and one poster asking whether we're celebrating AI again. Retail is leaning in on an unfinished bar, which is a good time to sell into them.

Finally, the plan doesn't trim into the 160-163 support cluster, where the weekly SuperTrend stop, the October 20 low, and the lower Bollinger band overlap. If a weekly close below 159.83 flips the primary trend, the remaining position goes. But if the price gets there and holds, that's where I'd want to press hardest, and I can only do that if I trimmed at 184.

So I'd trim into strength, keep a third, and keep the cash for the next move.

FINAL TRANSACTION PROPOSAL: SELL Aggressive Analyst: Let me start by giving back three things, because you were both right and my case is stronger without them.

First, the dry powder. I said I'd press hardest at 160. Neutral is right that this collides with the plan, which puts the stop at 159.83 and bars new buying until the filing clears. You can't hold a stop and a bid at the same price. But the fix makes the argument more aggressive. Use Neutral's own re-add rule: a clean filing, then a weekly close that holds support. Buy at 165 with the stop at 159.83 and you risk about 3% to reach the 197.6 middle band, roughly 20% higher, which is about six to one. Compare that with where we stand. At 184.31 the weekly stop is 13% below, the 197.6 middle band is 7% above, and the 50 SMA at 201.84 is 9.5% above. A full weight here risks 13 to make 7 to 10. That is the real reason to sell, not fear. The cash isn't for catching a knife. It's for the one setup on this chart with real asymmetry, a confirmed hold at support.

Second, I called 184 the upper part of the range, and it isn't. It's 31% of the way up the Bollinger band and about 60% of the way up the 160 to 200 chop the technical report calls the likely default. What I'm actually selling is a 7.5% one-day pop on a bar that's still printing. That's why Conservative's front-loading point lands. If the gap fills toward 173 to 177, the window is gone, and the plan as written could leave us holding well over a third with no rally coming. I'll take Neutral's version: most of the trim now, 195 to 202 preferred for the rest, and a deadline of the week before NVDA on 11/18. From a full weight, that last tranche is about two-ninths of the position. Even if it fills 7% below today, that costs roughly 0.06% of the portfolio, which is cheap for not depending on a rally that ADX at 11 doesn't promise.

Third, I leaned on eleven labeled StockTwits posts and on insider selling being programmatic. The first is too thin to carry weight. The second was the fundamentals report's impression, not a checked footnote. The data does add one thing, though. The CTO sold $6.09M on 10/6 and $6.07M on 10/13, with 27,500 shares one week and 31,290 the next. Same dollars, different share counts, which is what a dollar-denominated schedule looks like. It isn't proof, and the Form 4 footnote settles it in two minutes. But that $6.07M is also the single insider headline in the news feed. And no open-market buys, at a company that granted about 420,000 shares to four officers in May, is the absence of a signal, not a signal.

Now Conservative. Your one-sixth rests on the 50% drawdown, and Neutral caught that it wasn't a gap. Run the pieces. From 302 to 226 took ten weeks, about 25%. Then 27% came in two sessions. Then another 9% over the next couple of weeks. The one real gap in the record is the 27%, which is the number the plan sized for. You're right that a weekly-close stop doesn't protect against it, which is exactly why the plan says size does the work.

You call a third a position held on faith. The faith position is the full weight. A third with a gate, a halve-then-exit rule, and no adds through three catalysts is sized to survive being wrong. Cutting to zero before anyone has read the notes is faith in the other direction. I'll grant that the filing is today's homework, since the quarter ended August 1 and EDGAR has it. Make it step one. One possibility is worth holding in mind. A stock-funded acquisition would explain the opex step, the margin dip, and the $500 to $650M non-cash jump in assets and equity all at once, and at that size it's 1.4 to 1.9% of a $34.6B market cap. That's my inference and it could be wrong. Your bad version, bought revenue, fits just as well, which is why the notes decide it. So I'll take your rule: if the item is unexplained or the revenue was bought, the residual goes to a sixth or zero. But I won't fight over the default before the reading. The gap between a third and a sixth is about 0.6 points of weight, which on the outcomes we're debating is between a tenth and a sixth of a percent of the portfolio. That default only matters for the days until someone reads the filing.

On the daily-close exit, yesterday's 161.00 low looks scary until you notice it closed at 171.51, nearly 12 points above the stop. A close below 159.83 does break the bottom of the range, so I'll take Neutral's compromise: halve on a daily close, exit on a weekly close. I wouldn't go to zero on one daily close in a range-bound market, because you'd be selling the middle at 184 and the bottom at 159. On copper versus optics, you're right that no filing resolves it. That's gap risk, and sizing is the tool for gap risk.

Neutral, your best point is September 2. A 20% drop on 30 million shares, with Credo's usual early-September reporting, looks like a print. The news report only gave that timing from past patterns and flagged it as unconfirmed, so it's an inference, but a good one. If it's right, the market has had the margin and opex numbers for seven weeks and still bought 47% off the low. That's why I keep saying this is a sizing call, not a bearish thesis. It also shows what the third is for. A stock that rallies 47% in 16 sessions off a post-print low is one you want some of when the filing is clean and Q2 beats. And it puts the next gap on a calendar, early December, so we're sizing for an event rather than a ghost.

I'd push back on one thing. If the numbers are known and priced, the filing is a model-risk check, not an edge. So I'd keep the gate narrow: an unexplained item, or bought revenue. Whether the margin drop "looks structural" is a judgment the market made seven weeks ago, and our read of one 10-Q shouldn't override it by itself. Your +1 to +2% expected value is the cleanest line in this debate. On a stock that moves 8% a day, conviction can't set the size, the risk budget does. That is the aggressive position, because it means being decisive about the sale rather than tinkering.

So here's where I land. Sell most of the trim now into the gap, and the rest by the week before NVDA, preferring 195 to 202 but not waiting for it. Read the filing first. If it's clean and the revenue is organic, keep roughly a third through the print. If not, go to a sixth or zero. Halve on a daily close under 159.83 and exit on a weekly close. Re-add only after a clean filing and a weekly close that holds support, where the stop is 3% away and the first target is 20% away. Anyone already at or below a third should hold what they have. That's decisive now, small enough to survive being wrong, and loaded for the one setup on this chart that pays.

FINAL TRANSACTION PROPOSAL: SELL Aggressive Analyst: We're down to arguing over basis points, so let me name what's settled and then spend my time on the three things that aren't.

Everyone agrees to sell most of it, read the 10-Q and the Form 4 footnotes first, halve on a daily close under 159.83, exit on a weekly close, and size the leftover by a loss budget instead of fractions of a standard weight. I'm taking Neutral's version of that without a fight: about 1% of the portfolio, so a 30% gap costs about 0.3%. That sits between Conservative's quarter and the trader's third, and it's the right number. The 34% figure runs to the closing low, which is the worst possible exit, and Conservative's own daily-close halving rule exists to prevent that exit. And if the winner has run to double its standard weight, the instruction is still the absolute number, even if that means selling far more than two-thirds.

First open item is the last tranche. It's about two-ninths of a full position, roughly 0.8% of the portfolio. Sell it 7% lower than today and it costs about six basis points. Sell it into 197 to 202 and you gain about six. That's the entire gap between "all today" and "by October 30." But look at Neutral's trigger. A close under 180.69 is 2% below the print, about a quarter of one day's ATR, so on this stock it fires on noise, probably this week. Neutral's plan is Conservative's plan with a short fuse, and the 195 to 200 target is a lottery ticket, not something to wait for. I'm fine with that as long as we say so. Neutral also said every gap-prone date comes after the deadline, and that isn't quite right. The Meta print is this week, and the world affairs report flags its capex commentary as the read-through for AI hardware. It cuts both ways, and a 10% move on 0.8% is only eight basis points, so it doesn't change the answer. Conservative is also right that nothing Credo-specific drove today's gap, which is why the time-sensitive part should go into it. I'd sell two-thirds of the trim today, put the last two-ninths on a standing order at 196 or better through October 30, and keep the 180.69 close as the backstop. The cost of being wrong is six basis points, and the payoff is the only exposure we keep to the move that actually pays.

Second is dry powder. I'll stay conceded on the six-to-one, since Neutral's two or three to one is the realistic figure. It still beats risking 13% to make 7 to 10% at 184. But our own rules undercut the argument. We want a clean filing, a weekly close that holds support, and no adds into NVDA, the December print or the 12/9 FOMC. Stack those and the cash sits idle until after December 9 almost regardless of what the chart does. Neutral's own framework says the budget is the cap, and that budget already covers a catalyst gap. So re-adding back up to the cap can't raise our worst case above what we accepted, and the blackout is redundant. Either drop it and let the budget govern, or keep it and admit the cash is for December, not for 160. I'd drop it. The cash also has no obligation to return to CRDO.

Third is the gate. Conservative, on "unexplained" I'm with you: zero, not a sixth. A filed 10-Q has to say what the $500 to $650M is, and if the notes don't, that's a disclosure problem, not a modeling gap. On "bought revenue" I'd make the test measurable. Implied revenue went from about $437M to $479M, roughly $42M sequentially. If a deal supplied most of that, growth was bought and I'm at zero with you. If it supplied a few million, a stock-funded tuck-in isn't bought growth, and cutting to zero on it would be reacting to a footnote. You're also right that a clean filing earns no extra size. It removes a tail, and the print gets the same 1%.

On why not zero by default: you said a stock that goes 150 to 220 and back to 170 in five weeks hasn't priced anything stably. I agree, and that cuts against being flat as much as against being overweight. A name that rallies 47% in 16 sessions punishes the holder who has to call the re-entry on an 8% ATR, and that's what the 1% is for. It isn't a bet that the answers are good. At a 30% gap it costs about a tenth of a percent, and I think that's cheap for avoiding a re-entry decision made from zero.

On insiders, I'll let Neutral's point stand. Six million a week against more than a billion in daily dollar volume is noise, and no buying means we don't borrow conviction from insiders. That's the whole signal.

So the bold move here is deciding now. Get to about 1% of the portfolio, taking the first two-thirds of the trim today into the gap. Put the last slice on a 196-plus order through October 30, with a sale the next session on a close under 180.69. Read the filing this week, and go to zero if the item is unexplained or the deal bought most of the sequential growth. Halve on a daily close under 159.83 and exit on a weekly close. Rebuild only after a clean filing and a weekly close that holds support. If you're already at or under 1%, hold, and if you're flat, don't buy this gap.

FINAL TRANSACTION PROPOSAL: SELL Aggressive Analyst: We're now arguing about seven trading sessions and roughly half a point of portfolio weight, so I'll say where I move and where I hold.

Neutral caught a real flaw in my plan. I put a 196 order above and a 180.69 close below, then called the 180.69 trigger noise myself. Your random-walk number checks out. The stock is 3.6 points above the 200 SMA and 11.7 points below 196, so it hits 196 first about one time in four. The combined rule sells near today's price on average and adds a way to dither. I also called 196 a lottery ticket, and a one-in-four shot isn't one, so I'm retracting that. I'll take your version: five-sixths of the trim today, the last sixth by the close on October 30 on a pure date rule, and a 196 limit resting as a free option. That moves me toward Conservative. A date I can't talk myself out of beats a price trigger that fires on noise.

I won't go all the way to selling everything today. Conservative, your point is that holding the last tranche carries 1.8 to 2 percent against a budget built for 1 percent. As discipline, that's fair. But at Neutral's split we hold about 1.5 percent, so the gap between your plan and his is about 0.46 percent of the portfolio for seven sessions. On a 15 percent sector move that costs about seven basis points, and on a 30 percent undated gap about 14. You said a loss limit exists so we don't re-run the expected-value argument every time the stock gets lively. I agree, and that's the case for a date rule. But "sell it all while the window is open" is a timing claim, and you've told us all debate that we have no edge on timing. Your point that the realistic path for that tranche is a sale below today's price is also a selection effect. You only sell lower in the paths where the stock went lower, and you sell higher in the paths where it went up. Averaged, that's today's price, which is Neutral's point.

On dry powder, the two of you have worn my argument down to what it is. Conservative is right that if we hold the 1 percent we're at the cap, and that the budget covers one gap, not a sequence of stop, re-add, stop. I'll take Neutral's fix: rebuild at half size until the December print, and no new buying in the three sessions before NVDA, before the CRDO print, or before the FOMC. A blanket ban through December 9 idles cash for no reason, and the narrow rule does the job. Neutral, you're also right that what's left of my case is that the cash doesn't have to come back to CRDO. I'll own that. The point of selling is that it turns a 3.5 to 4 percent bet nobody priced into a 1 percent bet we did price. That's the aggressive part.

On the gate, you two are fighting over about $3.5M of revenue. A quarter of the $42M sequential increase is $10.5M and a third is $14M, which is under 1 percent of the quarter. I'd use Neutral's organic-growth test because it's measurable. I'd fold in Conservative's concern in a form we can check. If a deal does explain the margin and opex step, find out whether the acquired business carries a permanently lower margin. If it does, 64.5 percent is the new base, not a dip. That matters more than where the threshold sits. Also, the Zacks headline cites a $525M to $535M outlook. If that's the Q2 guide, it implies about 10 to 12 percent sequential growth from $479M, but only about 96 to 100 percent year over year, so the deceleration continues. The period isn't stated and it's unverified, but it tells us what the early-December print has to clear. The margin line in that guide matters more than the revenue line.

I accept Conservative's consequence for a missed deadline. The filing is an hour of reading and has been on EDGAR for weeks. So read it before the last sixth goes. If the notes say zero on the 23rd, the last sixth and the residual leave together, and the October 30 deadline never matters.

So here's where I land. Sell five-sixths of the excess into today's gap and the last sixth by the October 30 close, with a 196 limit resting. Read the 10-Q and the Form 4 footnotes first. Go to zero if the $500 to $650M item is unexplained or organic growth falls under about 6 percent. Halve if the margin dip is unexplained and still sliding, or if the notes are unread by the 30th. Halve on a daily close under 159.83 and exit on a weekly close. Rebuild only after a clean filing and a weekly close that holds support, at half size until December and not in the three sessions before each catalyst. If you're already at or under 1 percent, hold, and if you're flat, don't buy this gap.

The bold move isn't predicting where this stock goes. It's writing the rules down while the bar is still printing, so none of them depend on a rally ADX at 11 doesn't promise.

FINAL TRANSACTION PROPOSAL: SELL Aggressive Analyst: Neutral and Conservative have mostly argued me out of my position, so I'll say where I'm moving and where I'm holding.

First, the last sixth. Neutral is right that nothing in the filing read tells us to keep it. A clean filing earns no extra size, and a bad one cuts it. My own line that the sixth and the residual leave together if the notes say zero on the 23rd already made Friday the decision day. The October 30 date was me defending a price trigger I'd already called noise. So the last sixth goes at Friday's close, tied to the read. I'll take Neutral's other hit too. If the bold move is deciding now, the five-sixths goes into today's session with no price floor. A sale that depends on 184 holding is a hope, not a decision.

I'd keep the 196 limit resting through Friday, with the 180.69 trigger gone. By Conservative's own yardstick of 13 to 15% over seven sessions, the stock moves about 5% a day. A 6% pop to 196 inside two sessions is under one standard deviation, so on a rough read it's maybe a one-in-three shot. It's a fill price, not a reason for the date. The whole argument over the sixth is about 0.46% of the portfolio, so a one-sigma two-session move is worth about three basis points. I'm not spending credibility on that. I'll also drop the trader's rule against selling into 160-163. That carve-out made sense when the trim was discretionary. With a budget and a date it becomes a reason to hold exactly when the stock is weakest, and if price is in that cluster on Friday, the stops are about to rule anyway.

On the gate, I'm with Neutral. One revenue line in dollars: if a deal supplied more than roughly $14M of the $42M sequential increase, the residual goes to zero. Pull the real Q4 revenue from the May release first, because that $437M base is implied. Conservative's middle rung turns a $3.5M difference on a derived base into a rule, and the margin line already covers the bought-growth-at-worse-margin worry. One practical wrinkle: the guide isn't in the 10-Q. The Friday read is the business-combination note and equity statement, the release's gross margin outlook, and the Form 4 footnotes on the CTO's plan. That's a focused hour, not a full read, but it is three documents, so name them and do them.

On the rebuild, the remaining-budget-divided-by-30% formula settles half versus a third, and I'll take it. It also fixes the stop, re-add, stop sequence Conservative worried about, if it's applied as a cap on the total position and not an add-on. After a halve at 159.83, the other half stays until the weekly close decides it. Nothing gets bought back until the filing is clean and a weekly close holds support, and then only up to the formula's total.

Here is what I'm still fighting for, and it's why this isn't a timid plan. Conservative said three big swings in six weeks isn't a cycle, it's the regime, and I agree. That regime is why 3.75% was too big. It's also why a name that went 150 to 220 in 16 sessions is worth keeping a seat in and re-buying by rule. The plan is built to be wrong cheaply and right on a schedule. What would move me above the cap isn't the chart. It's the margin line in the guide, plus a clean note on the $500-650M item, plus a weekly close that holds support. If those line up, the rebuild gets the asymmetry the 184 entry never had. The Zacks headline cites a $525-535M outlook. If that's the Q2 guide, it implies about 10 to 12% sequential growth from $479M but only about 96 to 100% year over year. The period isn't confirmed, so check it, but it shows what the early-December print has to clear.

So here's the plan. Cut to about 1% of the portfolio, or your own loss budget divided by 30% if it differs. Sell five-sixths of the excess into today's session with no floor, and the last sixth at Friday's close, with a 196 limit resting until then. Read the three documents by Friday. Go to zero if the $500-650M item is unexplained or a deal supplied more than about $14M of the sequential dollars. Halve if the margin step is permanent with no recovery in the guide, or if the notes are unread. Halve on a daily close under 159.83 and exit on a weekly close. Rebuild only after a clean filing and a weekly close that holds support, sized by the remaining budget, and never in the three sessions before NVDA, the CRDO print or the FOMC. If you're already at or under the cap, hold. If you're flat, don't buy this gap.

FINAL TRANSACTION PROPOSAL: SELL