Research cutoff: Sunday, August 23, 2026 at roughly 12:30 BST / 07:30 ET.
U.S. price cutoff: Friday, August 21 regular-session close.
Horizon: roughly the next 1–15 U.S. trading sessions.
Weekend note: the U.S. cash market is closed today. There is no Sunday price bar to manufacture, so every relative-performance number below is still anchored to Friday's completed session.

Today's answer: Cadence is still the best transition, but I would not front-run Wednesday

Friday finally gave this screen one setup that behaved like a real catch-up candidate rather than a stock that merely looked cheap versus a collapsing peer.

Cadence Design Systems (CDNS) rose 1.73% to $319.02 while its closest listed EDA peer, Synopsys (SNPS), was essentially flat. More importantly, Cadence's 1-, 3- and 5-session relative performance is now positive even though it still trails Synopsys over 10 and 20 sessions.

That is the shape I want: the old gap is still there, but the newest price action has actually turned.

The problem is the calendar. Synopsys reports fiscal Q3 results after the close on Wednesday, August 26. Nvidia reports the same day, and Cadence itself is presenting at the Deutsche Bank Technology Conference that afternoon. A direct-peer earnings shock and a major AI-spending readout can rewrite the relationship in a few hours.

UPS remains the second-best research watch. It still carries a very large 20-session deficit versus FedEx and has modestly improved over the 3- and 5-session windows, but Friday slipped relative again and UPS itself has not reclaimed the upper-$100s.

Fortinet and F5 are still quarantined. Both bounced Friday, but both remain below previously defined repair levels, and their attractive-looking short-window relative numbers are still contaminated by a week in which Palo Alto and CrowdStrike were hit much harder.

So the Sunday hierarchy is simple:

  • CDNS: best transition watch; the only name showing convincing absolute + relative repair.
  • UPS: secondary watch; the long gap is real, but the stock has not confirmed it.
  • FTNT: rejected / quarantine until broken support is rebuilt.
  • FFIV: rejected / quarantine until $400 is rebuilt and held.

Bottom line: there is no clean U.S. catch-up buy today. Cadence is the closest candidate for Monday, but I want another regular-session confirmation before the Wednesday event cluster.

NYSE trading floor

Candidate summary

NameGenuine relationshipWhat the screen seesStatusWhat I need next
Cadence (CDNS)Direct EDA peer Synopsys (SNPS)Positive 1/3/5-session turn while a meaningful 20-session deficit remainsBest transition watchAnother regular session holding roughly $316–$320; better confirmation through $325–$328 while CDNS itself remains positive
UPSDirect parcel/logistics peer FedEx (FDX)Small 3/5-session repair, but large 10/20-session deficit remainsSecondary watchReclaim roughly $106–$108, then hold into a second regular session while maintaining positive relative performance
Fortinet (FTNT)Cybersecurity peers PANW + CRWDAttractive short-window relative numbers, but FTNT remains below its old repair level and peer weakness distorted the comparisonRejected / quarantineReclaim $154–$155, then hold for two regular sessions; stronger repair above $158–$160
F5 (FFIV)App/API security versus PANW + CRWDFriday ended a six-session losing streak, but the absolute structure is still brokenRejected / quarantineReclaim $400, hold it for two sessions, then reassess the old low-$420s trigger

These are observation conditions, not opening-bell instructions.

Relative performance through Friday's close

For Cadence I use Synopsys, its cleanest direct public EDA peer. For UPS I use FedEx. For FTNT and FFIV I use the equal-weight return of Palo Alto Networks and CrowdStrike.

The figures are candidate return minus benchmark return, in percentage points. Negative means the candidate lagged.

Candidate vs benchmark1 session3 sessions5 sessions10 sessions20 sessions
CDNS vs SNPS+1.74 pp+2.41 pp+3.82 pp-1.60 pp-8.75 pp
UPS vs FDX-0.26 pp+1.08 pp+0.47 pp-4.43 pp-14.35 pp
FTNT vs PANW + CRWD+0.19 pp+4.25 pp+5.13 pp+2.22 pp-6.88 pp
FFIV vs PANW + CRWD+0.20 pp+4.81 pp+4.59 pp+2.17 pp-9.56 pp

The cyber rows are the trap in today's table.

If I ranked only by relative percentages, FTNT and FFIV would look healthier than Cadence. But Fortinet only just bounced back to $153.51 Friday and still finished below the old $154–$155 repair zone. F5 closed at $384.63, still well below the $400 level that previously invalidated its setup.

At the same time, Palo Alto and CrowdStrike suffered sharp weakness during the week. A stock does not become a healthy catch-up trade merely because its comparison basket falls faster.

Cadence is different. CDNS itself rose Friday, ended a four-day losing streak, and improved against Synopsys at the same time. That is why its smaller-looking relative numbers are more useful than the larger cyber numbers.

Seven-session walk-forward: no Sunday data means no reason to invent a new rule

I reconstructed the screen over the prior seven completed U.S. sessions — August 13, 14, 17, 18, 19, 20 and 21 — using only information that would have been available at each historical cutoff.

There is no new Sunday market outcome, so I am not pretending the sample changed overnight. The useful question is whether the current filters would have avoided the false starts that actually occurred.

They did.

Cadence shows why the absolute-price gate matters

CDNS began improving relative to Synopsys before Friday, but that was not enough:

  • August 17: CDNS beat SNPS on a relative basis, but Cadence itself fell. That failed the absolute-stabilization rule.
  • August 19: the 1-session relative gap was positive, but CDNS still closed lower in absolute terms.
  • August 20: the relative relationship improved again while Cadence itself fell again.
  • August 21: for the first time in the sequence, Cadence rose in absolute terms and the 1-, 3- and 5-session relative windows were positive while the 20-session deficit remained large.

That is why Friday is a genuine transition rather than simply another day where Synopsys happened to be weaker.

The other filters also prevented obvious false positives

  • FTNT's earlier one-day relative bounce failed immediately afterward. The second-session rule prevented it from becoming a completed signal.
  • FTNT then broke the predefined $154–$155 invalidation while PANW/CRWD were also collapsing. The quarantine and peer-shock rules correctly kept it out.
  • FFIV fell for six straight sessions before Friday's bounce. A relative-only model would have repeatedly promoted it because peers sometimes fell even faster.
  • UPS's short-window relationship improved, but the stock never reclaimed the roughly $106–$108 repair zone. Friday then slipped relative to FedEx again.

So I am not loosening or rewriting the core method today merely to make a Sunday article look new.

The current rules remain:

  1. Real economic or operating relationship.
  2. Multi-window divergence, not one dramatic day.
  3. Intact fundamentals — no guidance cut, dilution, balance-sheet stress, major legal/product problem or structural competitive break explaining the lag.
  4. A credible catalyst or supportive estimate backdrop.
  5. A fresh 5- or 10-session gap; a huge 20-session number alone can be stale.
  6. Absolute stabilization or a reclaim by the candidate itself.
  7. A second regular session of confirmation after the reclaim.
  8. Post-invalidation quarantine until broken support is rebuilt and held.
  9. Do not count relative strength produced by a peer-specific collapse.
  10. When a direct comparison peer has earnings within two sessions, require either post-event confirmation or agreement from a secondary benchmark / the candidate's own absolute strength.

That last rule matters more than any new tweak this weekend because Synopsys reports Wednesday.

I am still not publishing a raw hit rate from seven sessions. The audit is useful for identifying failure modes; a tiny sample does not deserve a probability badge.

1. Cadence: this is finally the right type of turn

Why the relationship is genuine

Cadence and Synopsys are the two dominant U.S.-listed electronic-design-automation companies. Their customers overlap across semiconductor designers, foundries and complex systems companies. Both benefit when chip designs become harder, advanced nodes proliferate, multi-die architectures expand and AI infrastructure pushes customers toward more design software and IP.

That makes CDNS/SNPS a much stronger pair than two stocks that merely share a broad technology ETF.

Why the lag does not look like a broken-business story

Cadence's Q2 results were strong. Revenue rose to $1.584 billion from $1.275 billion a year earlier. Non-GAAP EPS increased to $2.11 from $1.65. Backlog reached a record $8.1 billion. Management raised its 2026 revenue outlook to roughly 19% growth, lifted non-GAAP EPS guidance to $8.10, and raised its operating-cash-flow target to $2 billion.

The company explicitly tied the quarter to broad-based demand for AI-driven design tools across both Design for AI and AI for Design.

The sell side remains constructive as well. Recent analyst tracking still shows a strongly bullish consensus with many targets around or above $400, though that should not be confused with a short-term price guarantee. Community reaction to the quarter has also been broadly bullish on backlog and AI-driven demand, but I treat that only as sentiment context.

Friday finally changed the tape

CDNS closed $319.02, up 1.73%, while SNPS closed roughly flat near $397.87.

That leaves Cadence:

  • ahead over 1, 3 and 5 sessions;
  • only modestly behind over 10 sessions;
  • still materially behind over 20 sessions.

That is a useful transition shape because there is still potential mean-reversion room without asking the stock to catch a falling knife.

ItemCDNS plan
Observation areaRoughly $316–$320
First useful confirmationAnother regular-session hold in/above that area while CDNS remains positive in absolute terms and continues to beat SNPS
Better triggerClose through roughly $325–$328 with healthy technology/EDA breadth
InvalidationA fresh close below roughly $311–$312 would make Friday look more like another failed bounce
Likely catalystAI/EDA demand, analyst estimate support, Wednesday's Cadence conference appearance, Nvidia validating AI-capex demand
Main risksRich valuation, broad growth-stock de-rating, rising long yields, Synopsys earnings shock, AI-capex slowdown

Why Wednesday stops me from calling it a clean Monday buy

Synopsys reports Q3 after Wednesday's close. Cadence also has an investor presentation Wednesday afternoon, and Nvidia reports that evening.

That creates three overlapping signals:

  • SNPS: direct-peer earnings can mechanically alter the CDNS/SNPS gap.
  • NVDA: can move the entire AI/semiconductor design-spending narrative.
  • CDNS presentation: can give investors fresh company-specific commentary.

If Synopsys drops 10% on a company-specific issue, CDNS can suddenly look fantastic on relative performance without moving at all. Under the peer-event rule, that does not count as clean catch-up confirmation.

Verdict: CDNS is the best transition on the screen, but I want Monday to prove Friday was not a one-day bounce. I would much rather see $325–$328 earned than buy simply because the old 20-session gap is large.

2. UPS: the long gap survives, but Friday was not confirmation

Why the relationship is genuine

UPS and FedEx share the same parcel and logistics cycle: U.S. business shipping, e-commerce, international air freight, fuel, labor, network utilization and global trade.

This remains one of the cleanest operational peer comparisons in the U.S. market.

UPS's latest quarter also improved enough that the 20-session lag deserves investigation. The company reported roughly $22.8 billion of Q2 revenue, adjusted EPS of $1.76, and raised full-year expectations to approximately $91.2 billion of revenue and $7.22 of adjusted EPS. Management said the planned Amazon volume glide-down and related network reconfiguration had been completed.

But the discount has real reasons behind it. Investors still want proof that the post-Amazon network reset and cost program can produce durable domestic margin improvement. The sell side reflects the same disagreement: recent calls range from Buy ratings around the low-$130s to Hold ratings around $110.

Community discussion is also mixed. Some UPS employee threads talk about possible Amazon peak-season volume returning, while others remain skeptical about network volumes and restructuring. That is anecdotal operating color, not evidence that the stock should rerate.

Friday weakened the short-term case slightly

UPS closed around $102.01, down roughly 0.6%, while FedEx fell only about 0.3%.

The three- and five-session relative windows are still modestly positive, but the one-session reading turned negative again. The large 10/20-session deficit therefore remains a research lead, not a confirmed turn.

ItemUPS plan
Observation areaRoughly $101–$103
First useful repairReclaim roughly $106–$108
ConfirmationHold the repair into a second regular session while maintaining positive relative performance versus FDX
InvalidationA decisive close below roughly $101 would weaken the setup further
Likely catalystEvidence the network reset is improving volume quality and domestic margin; upward estimate revisions
Main risksWeak parcel volumes, restructuring charges, competition from FedEx/Amazon, fuel and labor costs, second-half execution

Verdict: still worth watching. The 20-session gap is large, but I am not buying the old gap while Friday's fresh relationship is negative.

3. Fortinet: Friday repaired one day, not the failed setup

FTNT rose 1.80% Friday to $153.51, but that close is still below the old $154–$155 repair/invalidation zone.

Its underlying quarter remains excellent — revenue rose 26% to $2.05 billion, earnings grew 41%, billings grew 33%, and Fortinet has continued adding AI-security capabilities.

So this is not a fundamental rejection. It is a price-structure rejection.

The tempting relative table is also contaminated. PANW and CRWD were hit hard during the week, so FTNT can look relatively stronger even without a healthy recovery of its own.

Verdict: keep the quarantine. Reclaim $154–$155, hold for two regular sessions, then reassess. Stronger repair starts around $158–$160.

4. F5: one positive Friday does not erase six losing sessions

FFIV rose 1.81% Friday to $384.63, ending a six-session losing streak.

Its fundamentals are still respectable: the latest quarter delivered $865 million of revenue, up 11%, product revenue grew 19%, and management raised fiscal-2026 revenue-growth guidance to roughly 9%–10%.

But the stock remains well below the old $400 invalidation, and the recent relative numbers are distorted by the same cybersecurity peer weakness that affects FTNT.

Verdict: no active setup. Reclaim $400, hold it for two sessions, and only then does the previous low-$420s catch-up zone become relevant again.

This week's event cluster makes patience more valuable

Friday repaired part of a difficult week, but the macro backdrop remains hostile to lazy mean-reversion trades.

Reuters noted that long-term global yields surged last week and the 30-year U.S. Treasury yield reached its highest level since 2007, pressuring long-duration growth stocks and semiconductor shares. The Philadelphia Semiconductor Index fell roughly 5% for the week.

Now several high-impact events land in the same few days:

  • Wednesday, August 26: Nvidia reports Q2 results after the close — a major test of the AI-infrastructure spending story.
  • Wednesday, August 26: Synopsys reports fiscal Q3 after the close — a direct event for the CDNS/SNPS relationship.
  • Wednesday, August 26: Cadence presents at the Deutsche Bank Technology Conference at 12:25 p.m. PDT.
  • August 27–29: the Federal Reserve's Jackson Hole symposium, with markets looking for clues on policy while long-term yields are already elevated.

This is exactly the kind of week when a statistically attractive relative gap can be overwritten by new information.

Bottom line

The screen is finally distinguishing real repair from relative arithmetic.

Cadence is the only candidate today that has both:

  • a real direct-peer relationship;
  • intact and improving fundamentals;
  • a still-unresolved longer-window gap;
  • positive 1/3/5-session relative performance;
  • and a positive absolute Friday close.

That is why CDNS is first.

But first does not mean clean. Synopsys and Nvidia both report Wednesday, Cadence has its own conference appearance the same day, and long-duration technology remains sensitive to elevated bond yields.

UPS is the secondary watch, but it still needs $106–$108 back. FTNT and FFIV remain quarantined because one bounce does not repair broken support or eliminate peer-shock contamination.

So Sunday's conclusion remains: no clean U.S. catch-up buy yet.

If Monday gives Cadence another positive absolute session, keeps it ahead of Synopsys, and pushes through roughly $325–$328, that would be the cleanest confirmation this screen has seen in more than a week. If it cannot, I am happy to wait for Wednesday's event cluster to clear.

Educational disclaimer: This report is for informational and educational purposes only. It is not personalized financial advice, investment research, or a recommendation to buy or sell any security. Relative relationships can break permanently rather than mean-revert.

Sources

Market / upcoming catalysts

Cadence / Synopsys

UPS / FedEx

Fortinet / F5 / cybersecurity

Community links are anecdotal and included only for sentiment / operating color, not as validation of a trade.

us stocksstock marketinvestingrelative strengthcatch up watch
Written and reviewed by /lico

Just writing down my thoughts, interests, and the things I learn along the way.