Research cutoff: Thursday, August 27, 2026 at roughly 12:30 BST / 07:30 ET.
U.S. price cutoff: Wednesday, August 26 regular-session close.
Horizon: roughly the next 1–15 U.S. trading sessions.
Pre-open rule: extended-hours moves are context only. They do not count as confirmation for this screen.
Today's answer: UPS is holding, Broadcom finally has a real catalyst — but there is still no clean pre-open buy
There are two names worth watching today, but I would not call either one a completed catch-up signal before the U.S. cash session opens.
UPS is still the cleaner setup. Tuesday gave us the first real transition: UPS rose 2.36% while FedEx slipped 0.10%. Wednesday was less exciting, but it mattered because UPS did not immediately give the move back. The stock added another 0.49% to $105.65. Its 3- and 5-session relative performance versus FedEx remains slightly positive, while the 20-session gap is still meaningfully negative.
That is almost the shape this screen wants: the laggard itself has stopped falling, the short windows are improving, and a longer-window deficit remains. The missing piece is price confirmation. UPS is still sitting just under the $106–$108 repair area that has mattered for several sessions.
Broadcom (AVGO) is the new research watch. This one is more complicated. Broadcom has lagged Nvidia by about 8.1 percentage points over 10 sessions and 14.3 points over 20, yet its AI semiconductor business is still growing at a remarkable pace. Nvidia's Wednesday-night earnings and long-term AI demand outlook are a genuine read-through for Broadcom's custom accelerators and AI networking business.
But this is not a free mean-reversion trade. Broadcom's weakness has had company-specific ingredients too: the market has already shown that very strong AI numbers can be punished when expectations, customer concentration or forward commentary disappoint. Broadcom also reports its own fiscal Q3 results on September 2, so the next week contains a major company-specific event.
Nvidia is up sharply in Thursday premarket trading. Broadcom is up too, but much less. That is useful information, not confirmation.
So the answer is simple: UPS is the closest transition, AVGO is a catalyst-reset watch, and there is still no clean pre-open buy.
Candidate summary
| Rank | Candidate | Genuine relationship | What matters now | Status | What would change the view |
|---|---|---|---|---|---|
| 1 | UPS | Direct parcel/logistics peer FedEx (FDX) | UPS held Tuesday's turn on Wednesday; 3/5-session relative performance stays slightly positive while the 20-session gap remains negative | Closest transition watch | Regular-session close above roughly $106.20, preferably through $108, then another session holding the repair while relative strength versus FDX stays positive |
| 2 | Broadcom (AVGO) | AI infrastructure/custom silicon/networking exposure versus Nvidia (NVDA) and the broader AI-capex complex | Nvidia's blowout outlook is a genuine sector read-through, while AVGO remains a substantial 10/20-session laggard | Catalyst-reset watch | AVGO must participate in the catalyst during regular trading: reclaim roughly $360–$365, then ideally $370+ with a second session of confirmation |
| — | Riot Platforms (RIOT) | Bitcoin mining/treasury exposure versus IBIT / Bitcoin | Tuesday's 6.5% bounce failed the very next session | Rejected again | Must rebuild above ~$21 and then ~$22.20; one-day spikes no longer qualify |
| — | Cadence (CDNS) | Direct EDA peer Synopsys (SNPS) | CDNS now outperforms SNPS over 1/3/5/10 sessions; only the old 20-session lag remains | Graduated / stale | Wait for a genuinely fresh divergence after the Synopsys earnings reset |
These are observation conditions, not market-open instructions.
Relative performance through Wednesday's close
The figures below are candidate price return minus benchmark price return, in percentage points. Negative means the candidate lagged its benchmark.
| Candidate vs benchmark | 1 session | 3 sessions | 5 sessions | 10 sessions | 20 sessions |
|---|---|---|---|---|---|
| UPS vs FDX | -0.20 pp | +0.21 pp | +0.39 pp | -1.15 pp | -7.83 pp |
| AVGO vs NVDA | +1.27 pp | -1.13 pp | +1.73 pp | -8.09 pp | -14.32 pp |
| RIOT vs IBIT | -2.66 pp | +2.15 pp | -8.30 pp | -22.45 pp | -10.51 pp |
| CDNS vs SNPS | +0.54 pp | +1.86 pp | +4.07 pp | +4.14 pp | -9.14 pp |
A note on UPS: the company went ex-dividend during the longer comparison window. These are raw close-to-close price returns, so the 10/20-session gap modestly overstates UPS's total-return underperformance. The gap is still large enough to matter, but it should not be interpreted to the second decimal place.
Seven-session walk-forward: the rules still saved us from the obvious false starts
I re-ran the screen over the prior seven completed U.S. sessions — August 18, 19, 20, 21, 24, 25 and 26 — using only information that would have been available at each historical cutoff.
I am not publishing a hit-rate percentage. The more useful information is what the rules did when the setups looked tempting in real time.
- RIOT on August 20 produced the kind of move that can trick a mean-reversion screen: a sharp absolute gain and a positive day versus Bitcoin exposure. The next session it reversed hard while IBIT surged. The second-session rule kept a one-day bounce from becoming a completed signal.
- RIOT on August 25 did it again. The stock jumped 6.5% and beat IBIT by more than six percentage points. On August 26 RIOT fell 3.24% while IBIT fell only 0.58%. Two similar failures in one week are a strong argument against loosening confirmation.
- UPS has behaved better. It began improving around August 25 and then held its absolute price on August 26. But Wednesday did not add relative strength versus FedEx. That is why the screen keeps UPS in transition rather than promoting it just because it did not fail.
- CDNS shows the opposite risk: waiting for proof can mean missing part of the move. Cadence's first promising turn did not confirm immediately, then the stock jumped 5% on August 25. The stale-gap filter now keeps us from chasing it simply because the 20-session column still looks negative.
The framework does not need to be loosened today. The recent false starts are exactly what the confirmation rules are meant to catch.
I am adding one explicit refinement because Nvidia's earnings create a new kind of test for AVGO:
Catalyst-reset rule
When a genuine related leader or demand bellwether releases major new information after the close, the old divergence remains a setup, not a signal. The first regular session after the event must show that the laggard actually participates in the same catalyst.
Premarket does not satisfy the rule.
If the candidate badly underperforms the relevant leader or sector on the catalyst day, that is evidence the gap may be company-specific rather than a catch-up opportunity.
The active framework is now:
- Require a genuine economic, operating or underlying-asset relationship.
- Require multi-window divergence; one dramatic day is not enough.
- Reject company-specific bad news, guidance cuts, dilution stress, legal/product problems, balance-sheet trouble or structural competitive breaks that explain the lag.
- Require a credible catalyst or supportive estimate backdrop.
- Require a fresh 5- or 10-session gap; a large 20-session number alone can be stale.
- Require absolute stabilization or a reclaim by the candidate itself.
- Require a second regular session of confirmation after a reclaim.
- Keep failed setups in post-invalidation quarantine until broken support is rebuilt.
- Do not count relative strength created by a peer-specific collapse.
- If a direct peer has earnings within two sessions, require post-event confirmation or agreement from a second benchmark.
- Use an event-day freeze when a direct peer reports after the close.
- Catalyst-reset rule: a major sector read-through must be confirmed by regular-session participation from the laggard itself.
1. UPS: the transition is still alive, but the breakout is not here yet
Why UPS versus FedEx is a real relationship
This is one of the cleaner pairs in the screen.
UPS and FedEx compete across parcel delivery, business shipping, international air freight and time-sensitive logistics. Both are exposed to e-commerce, industrial activity, fuel, labor, trade volumes and network utilization. If one materially outruns the other without a clear company-specific explanation, the divergence is worth investigating.
The business is not obviously broken
UPS's second-quarter report was better than the stock's reaction suggested.
The company reported $22.83 billion of revenue and $1.76 of adjusted EPS, and raised its 2026 outlook to roughly $91.2 billion of revenue and $7.22 of adjusted EPS. Management is deliberately shrinking lower-margin Amazon volume and leaning harder into healthcare, B2B, international and complex supply-chain work.
Investors have not fully bought the story. Reuters noted after the Q2 report that the shares fell despite the higher outlook because investors questioned whether UPS could deliver the second-half targets amid inflation, tariffs, competition and the Amazon volume reset.
That skepticism is important. This is not a stock that "must" bounce because the peer went up.
On August 24, UPS added a fresh strategic catalyst: it detailed more than $2 billion of ongoing investment across International, Healthcare and Supply Chain Solutions through 2028, including network expansion in Asia-Pacific, Canada and Hong Kong.
That does not move next quarter's earnings by itself, but it supports the same higher-value-business thesis behind management's turnaround plan.
What Wednesday told us
UPS closed Wednesday at $105.65, up 0.49%. FedEx rose 0.68%.
So Wednesday was not another relative win. But it was also not the immediate reversal that killed previous setups such as RIOT.
Over three sessions UPS is still about 0.21 percentage points ahead of FDX, and over five it is about 0.39 points ahead, while the 20-session deficit remains roughly 7.8 points on raw price returns.
That leaves the setup alive, but not finished.
| Item | UPS plan |
|---|---|
| Observation area | Roughly $104–$106 |
| First useful trigger | Regular-session close above roughly $106.20 |
| Better confirmation | Push through roughly $108, then hold the repair into another regular session while 3/5-session relative performance versus FDX stays positive |
| Invalidation | A fresh close below roughly $101–$102 materially weakens the transition |
| Likely catalysts | Execution on healthcare/international investment, evidence the post-Amazon network reset is lifting volume quality and margins, supportive estimate revisions |
| Main risks | Weak parcel demand, restructuring costs, labor/fuel pressure, tariffs, competition from FedEx and Amazon |
Sell-side opinion remains constructive but divided enough to be believable rather than euphoric. Recent published targets range from strongly bullish views to much more cautious ones. Community discussion is also cautious, with recurring concerns around Amazon logistics, margins and whether the network reset will take longer than management expects.
Verdict: UPS remains the best catch-up transition on today's screen. I still want it to prove itself above $106–$108 rather than buying one inch below resistance because the relative table looks better.
2. Broadcom: Nvidia finally gives the laggard thesis a proper test
Why AVGO versus Nvidia is economically relevant — and why it is not a perfect pair
Broadcom and Nvidia are not interchangeable competitors.
Nvidia dominates merchant AI accelerators. Broadcom is a major supplier of custom AI accelerators and AI networking, with a large software business on top. The relationship is therefore not "NVDA went up, so AVGO must go up." The genuine common driver is hyperscaler and AI-infrastructure capital spending.
That driver is very much alive.
Nvidia reported second-quarter revenue of $96.2 billion, including $89 billion of data-center sales, and gave an outlook that reinforced a longer AI-spending runway. Reuters reported Thursday that Nvidia was up roughly 6%–7% premarket and that at least ten brokerages raised price targets after the results.
That is a meaningful demand read-through for Broadcom's AI networking and custom-silicon business.
Broadcom's own AI fundamentals remain strong
Broadcom's fiscal Q2 revenue reached $22.2 billion, up 48% year over year. Its AI semiconductor revenue was $10.8 billion, up 143%, and management guided fiscal Q3 AI semiconductor revenue to roughly $16 billion, more than 200% growth year over year.
Those numbers do not look like a broken AI business.
Analysts remain broadly positive. Recent published ratings include Buy/Outperform calls and targets well above the current share price.
Retail sentiment has also noticed the divergence. A fresh Reddit discussion on August 26 explicitly asked why Broadcom had lagged while other AI stocks rallied. The replies were mostly constructive, but not blindly bullish: several users argued it could still fall further or pointed to the ugly chart. That is actually a healthier setup than a unanimous "easy money" narrative.
Why this is still only a watch
The stock price has earned its skepticism.
AVGO closed Wednesday at $355.59. It is roughly 8.1 percentage points behind Nvidia over 10 sessions and 14.3 points behind over 20.
But part of Broadcom's weakness has been company-specific. After its June quarter, the market punished the stock despite huge AI growth because expectations were extraordinarily high and investors focused on forward AI assumptions, customer concentration and diversification risk.
That means today's divergence cannot be labeled "unexplained."
Broadcom also reports fiscal Q3 results on Wednesday, September 2 after the close, according to the company. That gives this 1–15 session setup a very obvious binary catalyst.
Thursday premarket is the first test. Nvidia is surging and Broadcom is also higher, but Broadcom is participating much less strongly. Under the new catalyst-reset rule, that extended-hours move is only context.
| Item | AVGO plan |
|---|---|
| Observation area | Roughly $350–$360 |
| First useful trigger | Regular-session reclaim of roughly $360–$365 while the stock participates in the Nvidia-led AI move |
| Better confirmation | Roughly $370+, then a second regular session holding the repair |
| Invalidation / rejection | Fresh close below roughly $350, or new company-specific evidence that weakens the custom-silicon/networking thesis |
| Likely catalysts | Nvidia's demand read-through, AI networking/custom accelerator demand, estimate revisions, Broadcom's September 2 earnings |
| Main risks | Customer concentration, hyperscaler supplier diversification, valuation/expectations, financing/debt sensitivity, earnings event risk |
The key question today is not whether Nvidia had good numbers. It did.
The question is whether Broadcom's stock is willing to respond to good industry information.
If AVGO cannot participate during the regular session when the clearest AI bellwether just reinforced the spending cycle, I would treat that as a warning that the gap is more company-specific than it appears.
Verdict: AVGO is the most interesting new catch-up research setup today, but it has to pass the catalyst test first. I would rather miss the first few dollars than call premarket sympathy a completed turn.
Why RIOT is rejected again
RIOT gave us exactly the failure the rules were built for.
Tuesday looked excellent: the stock jumped 6.50% while IBIT gained only 0.18%.
Wednesday RIOT fell 3.24% while IBIT fell just 0.58%. The one-session relative result flipped back to -2.66 percentage points.
The longer gap versus Bitcoin exposure remains huge, but that is no longer enough. RIOT has now produced multiple sharp one-day relative bounces that failed immediately.
Its growing data-center business also makes IBIT a less complete benchmark than it used to be.
Verdict: no active catch-up signal. RIOT stays out until it can rebuild above roughly $21 and then $22.20 with actual second-session confirmation.
Why Cadence is still not a fresh catch-up buy
CDNS is doing what we hoped a laggard would do — which is precisely why it no longer belongs in the early-laggard bucket.
Through Wednesday, Cadence is now ahead of Synopsys by roughly 0.54 / 1.86 / 4.07 / 4.14 percentage points over the 1 / 3 / 5 / 10-session windows. Only the 20-session gap remains negative.
Synopsys then reported Wednesday night and raised its annual revenue and profit forecasts, citing AI-driven chip-design demand.
That earnings event resets the direct-peer relationship anyway.
Verdict: CDNS has graduated. A stale 20-day deficit is not a reason to chase a stock after the catch-up has already started.
Other AI names I did not force into the list
Nvidia's report has lifted a broad group of AI-infrastructure stocks in premarket trading. Marvell, Vertiv, Arista, Micron and others are moving.
That does not automatically make them catch-up candidates.
Several of those names have already been leaders over the recent 10- and 20-session windows or have already made large moves. A strategy built around buying before relative convergence should not quietly become "buy whatever is green after Nvidia earnings."
Two real watches are enough today.
Macro backdrop: good AI news is fighting a less comfortable rates setup
Wednesday's U.S. session was subdued after July PCE inflation came in hotter than expected. Reuters reported the major indexes finished slightly lower as investors waited for Nvidia.
Thursday's premarket picture is more risk-on in technology: Nasdaq-100 futures were up roughly 1% after Nvidia's results, while the S&P 500 futures were also higher.
But the rates backdrop has not disappeared. Markets are waiting for Fed Chair Kevin Warsh's Jackson Hole remarks, and higher long-term yields have recently been one of the main pressures on growth-stock valuations.
That makes today's regular session particularly useful. If an AI laggard cannot respond when the sector has a strong fundamental catalyst and Nasdaq futures are higher, that failure deserves more weight than an ordinary down day.
Bottom line
The screen is finally giving us two different kinds of useful setup:
- UPS is a classic slow transition: short-window relative performance has improved, the longer gap remains, and the stock is sitting just below a clear repair zone.
- AVGO is a catalyst-reset test: the long-window divergence is large, fundamentals are still strong, and Nvidia has just delivered the most relevant industry demand signal we could ask for.
Neither is complete before the bell.
For UPS, I want $106–$108 reclaimed and held.
For Broadcom, I want to see the stock actually participate in regular trading, first through roughly $360–$365 and preferably $370+, rather than assuming Nvidia's success will automatically pull it higher.
RIOT failed its second-day test again. Cadence has already moved too far into its catch-up to qualify as early.
So today's conclusion remains deliberately boring: two good watches, zero clean pre-open buys.
Sources
Market and AI backdrop
- Reuters, Aug. 27 — Nvidia rises after signaling longer AI spending runway: https://www.reuters.com/business/nvidia-rises-after-signaling-longer-ai-spending-runway-2026-08-27/
- Reuters, Aug. 27 — Nasdaq futures lead after Nvidia forecast refuels AI trade: https://www.reuters.com/business/nasdaq-futures-take-lead-after-nvidia-forecast-refuels-ai-trade-2026-08-27/
- Reuters, Aug. 26 — Wall Street ends slightly lower after hot inflation data: https://www.reuters.com/business/us-stock-futures-subdued-run-up-nvidia-results-inflation-print-2026-08-26/
- Reuters, Aug. 26 — Investors await Warsh's Jackson Hole policy clues: https://www.reuters.com/business/anxious-investors-hope-clarity-warshs-fed-plan-jackson-hole-2026-08-26/
UPS
- Reuters, July 28 — UPS raises 2026 forecasts, shares fall on skepticism: https://www.reuters.com/business/ups-raises-full-year-revenue-forecast-2026-07-28/
- UPS, Aug. 24 — More than $2B investment across International, Healthcare and Supply Chain Solutions: https://about.ups.com/gb/en/newsroom/press-releases/customer-first/ups-invests-more-than--2-billion-to-give-customers-even-faster-s.html
- UPS price history: https://stockanalysis.com/stocks/ups/history/
- FedEx price history: https://stockanalysis.com/stocks/fdx/history/
Broadcom
- Broadcom, June 3 — fiscal Q2 2026 results: https://investors.broadcom.com/news-releases/news-release-details/broadcom-inc-announces-second-quarter-fiscal-year-2026-financial
- Broadcom, Aug. 3 — fiscal Q3 results scheduled for Sept. 2: https://investors.broadcom.com/news-releases/news-release-details/broadcom-inc-announce-third-quarter-fiscal-year-2026-financial
- Investing.com — current Broadcom analyst consensus and recent ratings: https://www.investing.com/equities/avago-technologies-consensus-estimates
- Reddit, Aug. 26 — retail discussion around AVGO's relative lag: https://www.reddit.com/r/Stocks_Picks/comments/1vyups7/avgo_broadcom/
- Broadcom price history: https://stockanalysis.com/stocks/avgo/history/
- Nvidia price history: https://stockanalysis.com/stocks/nvda/history/
Rejected / comparison names
- Reuters, Aug. 26 — Synopsys raises annual forecasts on AI-driven design demand: https://www.reuters.com/business/synopsys-raises-annual-forecasts-ai-driven-chip-design-software-demand-2026-08-26/
- RIOT price history: https://stockanalysis.com/stocks/riot/history/
- IBIT price history: https://stockanalysis.com/etf/ibit/history/
- Cadence price history: https://stockanalysis.com/stocks/cdns/history/
- Synopsys price history: https://stockanalysis.com/stocks/snps/history/
This is market research, not individualized financial advice. Catch-up setups can stay divergent or break further; predefined invalidation and position sizing matter more than the elegance of the relative-performance story.