Research cutoff: Thursday, September 3, 2026, around 7:25 a.m. ET.
Price cutoff: Wednesday, September 2 regular-session close.
Horizon: roughly the next 1–15 U.S. trading sessions.
Premarket rule: premarket and after-hours moves are context only. They do not complete a catch-up signal.
Today’s answer: no clean catch-up entry
There are two interesting relationships on the screen this morning, but neither is ready.
UPS still has the old 20-session deficit versus FedEx, and its short-window relative numbers suddenly look better again. The problem is how they improved: FedEx fell 2.6% Wednesday while UPS was only slightly positive. That is not the same thing as UPS starting its own catch-up move.
Datadog is more interesting tactically. It has badly lagged the cloud-data/software basket over the last several sessions, and Snowflake’s strong quarter is a genuine demand read-through for AI-driven data infrastructure. DDOG is up roughly 5% premarket. But the catalyst arrived after Wednesday’s close, so by the rules of this screen the clock resets: I need a regular-session reclaim and follow-through before treating the sympathy move as evidence.
Marvell has a large 5- and 10-session gap, but its August 28 selloff was tied directly to disappointment over the timing of revenue from the Google custom-chip deal. That is company-specific information, not a clean peer-laggard anomaly.
KLA still trails SOXX, but Lam Research and Applied Materials are also lagging. That looks like a semiconductor-equipment group issue rather than one stock being left behind.
Bottom line: no clean U.S. catch-up buy before the September 3 open. UPS and DDOG are watches. MRVL and KLAC are rejects under the current rules.
Image source: Wikimedia Commons.
Candidate summary
| Name | Genuine relationship | What the screen sees | Status | What would change my mind |
|---|---|---|---|---|
| UPS | Direct parcel/logistics peer FedEx (FDX) | Large old deficit remains; 1/3/5/10-session relative numbers improved, but mostly because FDX sold off | Watch only | UPS itself needs to lead: regular close above ~$106.20, then follow-through with positive short-window relative strength |
| Datadog (DDOG) | WCLD cloud-software basket; Snowflake is a demand read-through for cloud data/AI workloads | Severe fresh lag, plus a credible new peer catalyst after Wednesday’s close | Catalyst-reset watch | Regular-session close back above ~$224, then a second session holding the repair rather than a one-day sympathy spike |
| Marvell (MRVL) | SMH / AVGO custom AI silicon and networking exposure | Large 5/10-session gap, but it was created by MRVL’s own earnings/deal-timing disappointment | Rejected | Needs a new base and stock-specific positive turn; the old gap cannot be treated as unexplained mispricing |
| KLA (KLAC) | Semiconductor process-control exposure versus SOXX, cross-checked with LRCX/AMAT | Negative relative performance across most windows, but equipment peers are weak too | Rejected for now | KLAC must outperform both SOXX and its close equipment peers for at least two regular sessions |
Relative performance through Wednesday’s close
The figures below are candidate price return minus comparator price return, in percentage points. Negative means the candidate lagged its comparator over that window.
| Candidate vs comparator | 1 session | 3 sessions | 5 sessions | 10 sessions | 20 sessions |
|---|---|---|---|---|---|
| UPS vs FDX | +2.91 pp | +2.13 pp | +3.29 pp | +3.75 pp | -6.23 pp |
| DDOG vs WCLD | -4.70 pp | -7.10 pp | -10.24 pp | -11.46 pp | -33.31 pp |
| MRVL vs SMH | -2.82 pp | -4.21 pp | -14.81 pp | -11.12 pp | +1.22 pp |
| KLAC vs SOXX | +0.58 pp | -0.46 pp | -3.59 pp | -4.51 pp | -5.14 pp |
The table is a good example of why the biggest gap is not automatically the best trade.
UPS now looks strong over four short windows, but the stock itself went from $104.23 on August 31 to $102.66 on September 1 and $103.01 on September 2. The apparent relative repair was manufactured largely by FedEx falling harder. DDOG has the cleanest fresh divergence, but Wednesday’s close came before the Snowflake catalyst. MRVL’s gap is explained by its own news. KLAC’s gap is shared by its subindustry.
Seven-session walk-forward: one more filter after replaying the recent signals
Before publishing this screen, I replayed the selection process at each of the prior seven completed U.S. decision cutoffs using only information that was available at that time.
I am not publishing raw hit rates. The useful output is the failure mode.
The replay again showed that the dangerous setups are not usually the obvious broken companies — those are easier to reject. The harder mistake is a false repair: the relative spread improves, but the candidate is not actually strengthening. A peer or benchmark is simply falling faster.
That matters directly for UPS today.
So I added a stricter confirmation rule and reran the seven cutoffs:
Candidate-led repair rule: relative improvement does not count as confirmation when it is produced mainly by weakness in the comparator. The candidate itself must show positive or clearly stabilizing regular-session price action over the confirmation window, in addition to improving relative performance.
This rule sits on top of the existing filters: genuine economic relationship, stale-gap rejection, absolute stabilization, second-session confirmation, post-invalidation quarantine, company-event freezes, peer-shock contamination, benchmark triangulation, and cross-sectional uniqueness.
The revised replay still keeps one-day turns and event gaps out, while preventing today’s UPS spread improvement from being promoted prematurely. The method is becoming a little less eager, which is exactly what I want from a 1–15 session catch-up screen.
1. UPS: the relative repair is real mathematically, but not yet real in the stock
Why the relationship matters
UPS and FedEx are unusually clean comparators. Both depend on parcel volumes, business shipping, international trade, air and ground network utilization, labor, fuel and pricing discipline. A persistent divergence can matter if one company has no obvious company-specific reason to remain behind.
UPS’s fundamentals are not obviously broken. In Q2 it reported $22.8 billion of revenue, $1.76 of adjusted EPS, and raised 2026 revenue, adjusted operating-profit and adjusted EPS guidance. Management also said the Amazon glide-down and related network reconfiguration were completed as designed.
More recently, UPS announced more than $2 billion of ongoing investment across international, healthcare and supply-chain operations, while its new global operating structure took effect September 1.
Likely catalyst
The near-term catalyst is less about one headline and more about whether investors start rewarding the post-Amazon network reset, higher-value healthcare/international mix and operating-model changes. That is a plausible catch-up story.
But Wednesday’s relative move was not confirmation. UPS rose just 0.34%, while FDX fell 2.57%.
Entry / observation zone
| Item | UPS plan |
|---|---|
| Observation zone | Roughly $102–$106.20 |
| First real confirmation | Regular-session close above ~$106.20 with UPS itself positive |
| Cleaner setup | Second session holding the reclaim while 3/5-session relative strength versus FDX remains positive |
| Invalidation | Break below roughly $101.80 without a quick regular-session reclaim |
Key risks
Parcel demand can soften quickly; oil and jet-fuel prices are elevated; restructuring can take longer than expected; and the competitive relationship with FedEx is not perfectly stationary. A FedEx-specific selloff can also distort the spread, which is exactly what happened Wednesday.
Analyst / news / community sentiment
The sell side remains mildly constructive rather than euphoric. StockAnalysis’ S&P Global aggregation shows 29 analysts, a consensus Buy, and an average target of $116.08. Recent calls are mixed: Wells Fargo and Bank of America maintained Hold ratings on September 1, while Citi reiterated Buy on August 27.
Retail/community discussion is much less useful. Recent UPS threads are split between the valuation/healthcare-logistics bull case and concerns about domestic volume and margins. I treat that as a sentiment temperature check, not evidence.
Verdict: watch only. The stock must lead its own repair.
2. Datadog: Snowflake gives the laggard a real catalyst, but Thursday has to confirm it
Why the relationship matters
Datadog and Snowflake are not interchangeable businesses, so I would not use a simple one-to-one pair trade. The economic connection is still meaningful: both benefit from enterprise cloud migration, rising data volumes and AI workloads, and both sit in the spending chain for companies building and operating data-intensive applications.
That is why I compare DDOG not only with Snowflake’s news but also with the WCLD cloud-software basket.
DDOG has materially lagged that basket over every measured window, including more than 10 percentage points over five sessions. Some of the 20-session gap is contaminated by Datadog’s own August 6 post-earnings de-rating, so I put more weight on the fresh 3/5/10-session weakness.
Fundamentally, the company is still growing quickly. Datadog reported Q2 revenue of $1.12 billion, up 36% year over year, with $279 million of free cash flow and roughly 4,720 customers above $100,000 of ARR, up 23% from a year earlier.
Likely catalyst
Snowflake’s Wednesday report is the cleanest new catalyst on today’s screen. Snowflake raised its fiscal-2027 product-revenue forecast to $6.07 billion, reported 37% product-revenue growth, and said AI products accounted for roughly half of the acceleration in growth. At least 22 brokerages raised Snowflake price targets after the report.
Datadog was up about 5% premarket Thursday in sympathy.
That is exactly where the event-reset rule matters: the market has not yet had a regular session to decide whether the read-through is durable.
Entry / observation zone
| Item | DDOG plan |
|---|---|
| Observation zone | Roughly $208–$224 |
| First real confirmation | Regular-session close above ~$224 |
| Cleaner setup | A second session holding above the repair area with DDOG outperforming WCLD |
| Invalidation | Break below Wednesday’s $208.22 low without a quick reclaim |
Key risks
Valuation remains sensitive, the 20-session gap includes the company’s own August earnings reset, and Snowflake’s consumption/data-cloud model is not identical to Datadog’s observability/security economics. A one-day software sympathy rally can fade fast.
Analyst / news / community sentiment
Sell-side sentiment is strong. StockAnalysis’ S&P Global aggregation shows 46 analysts, a Strong Buy consensus and an average target near $285. Recent August targets from Mizuho, Baird and JPMorgan were around $300 or above, although Bernstein remained more cautious.
Community chatter is broadly positive on AI-observability demand but also repeatedly flags valuation risk after earnings. The Reddit samples are low-volume and anecdotal, so I do not use them as a signal.
Verdict: the most interesting watch today, but still a watch. Let the regular session prove that Snowflake changed the tape.
3. Marvell: a big gap with the wrong explanation
Why the relationship matters
Marvell, Broadcom and the SMH semiconductor basket share genuine exposure to custom AI accelerators, networking silicon and hyperscaler capital spending. Broadcom’s latest results reinforce the underlying demand story: Q3 AI-chip revenue reached $16.7 billion, and Broadcom raised its fiscal-2027 AI-chip outlook to about $115 billion.
That would normally make a lagging AI-silicon peer interesting.
Why I reject it
MRVL’s five-session deficit versus SMH is almost 15 percentage points, but the gap was created by Marvell’s own August 28 repricing. Investors were disappointed by the timing of meaningful revenue contribution from the Google custom-chip deal even though management raised longer-term revenue forecasts.
Reuters reported that the Google deal could generate up to roughly $120 billion through fiscal 2033, but management indicated the contribution becomes more meaningful in fiscal 2029. That timing reset is exactly the kind of company-specific explanation this strategy is supposed to avoid.
The 20-session comparison reinforces the point: MRVL is actually ahead of SMH over that longer window.
Observation / invalidation
I would only revisit the name after it builds a fresh base and starts outperforming both SMH and close AI-silicon peers on its own. The recent $200–$218 area is a base-watching zone, not a catch-up entry.
Verdict: reject. Broadcom’s strong demand read-through does not erase Marvell’s own expectation reset.
4. KLA: this is still an equipment-group lag, not a single-stock anomaly
Why the relationship matters
KLA is directly tied to semiconductor manufacturing intensity through process-control and inspection tools. The AI infrastructure cycle, advanced packaging, leading-edge logic and memory complexity all increase the need for this equipment.
KLA’s own business is healthy. Fiscal Q4 revenue was $3.66 billion, and management said the trends supporting growth are strengthening into the second half of calendar 2026 and 2027.
Why I reject it today
KLAC trails SOXX over 3/5/10/20 sessions, but its closest equipment peers do too.
For context, Lam Research trails SOXX by roughly 5.1 points over five sessions, while Applied Materials trails by roughly 5.9 points over the same window. When several tight peers lag together, the simplest explanation is a subindustry rotation or multiple compression, not a stock-specific mispricing.
That fails the cross-sectional uniqueness filter.
Observation / invalidation
The catch-up thesis becomes more interesting only if KLAC starts outperforming both the broad semiconductor basket and its equipment peers for two regular sessions. Until then, the recent $170–$176 area is only a base to watch.
The setup is invalidated as a single-name catch-up idea if the semiconductor-equipment basket keeps making new relative lows versus SOXX or if the policy/tariff backdrop worsens.
Verdict: reject for now. Strong fundamentals are not enough when the whole peer group is lagging.
Other relationships screened and rejected quickly
A few names looked tempting but failed before reaching the main table. Microsoft had fresh Azure disclosure but is not a genuine laggard: its relative performance versus QQQ is positive over 5, 10 and 20 sessions. Occidental has almost no meaningful relative gap versus XLE. Oilfield services have already moved with crude, so OIH is not a laggard to chase. HPE also fails the clean-read-through test because its own earnings and supply-constraint commentary now dominate the stock.
That is the point of the process: a quiet screen is better than manufacturing a trade.
Market context for Thursday
U.S. futures were close to flat early Thursday as investors balanced strong corporate earnings against elevated oil prices and Middle East risk. Broadcom was down about 2.3% premarket after its total Q4 revenue forecast came in slightly below consensus despite a stronger long-term AI-chip outlook, while Snowflake was up roughly 24% after its results.
Treasury yields eased from recent highs, but Friday’s U.S. payrolls report is the next major macro event. That raises the bar for taking a marginal catch-up setup today: I want the stock-specific turn to be visible before adding macro-event risk.
What would change the call today
UPS: close above ~$106.20 with UPS itself positive, not merely FedEx weaker; then hold the reclaim for a second session.
DDOG: regular close above ~$224 after the Snowflake read-through, then hold or extend while outperforming WCLD.
MRVL: no catch-up call until a new base forms and the stock begins to outperform after its own expectation reset.
KLAC: needs a stock-specific turn versus both SOXX and close equipment peers, not just a bounce in the whole group.
Until one of those happens, there is no clean 1–15 session relative-laggard setup I would promote today.
Sources
Price and relative-performance data
- UPS price history — StockAnalysis
- FDX price history — StockAnalysis
- DDOG price history — StockAnalysis
- WCLD price history — StockAnalysis
- MRVL price history — StockAnalysis
- SMH price history — StockAnalysis
- KLAC price history — StockAnalysis
- SOXX price history — StockAnalysis
Company, analyst and market sources
- UPS Q2 2026 earnings — UPS Investor Relations
- UPS $2B international, healthcare and supply-chain investment — UPS
- UPS analyst consensus — StockAnalysis
- Datadog Q2 2026 results — Datadog Investor Relations
- Datadog analyst consensus — StockAnalysis
- Snowflake results and AI-demand read-through — Reuters
- Datadog premarket sympathy move after Snowflake — Investing.com
- Broadcom AI-chip outlook — Reuters
- Marvell selloff and Google-deal timing — Reuters
- KLA fiscal Q4 2026 results — KLA Investor Relations
- Thursday U.S. futures / oil / earnings context — Reuters
Community sentiment — anecdotal only
- UPS volume discussion — Reddit
- UPS post-earnings discussion — Reddit
- Datadog post-earnings valuation discussion — Reddit
This is market research, not personalized investment advice. Catch-up trades can fail because the relationship changes, new company information emerges, or macro conditions overwhelm the historical relationship.