⚠ Clarification on ZETX: ZETX is the Defiance Daily Target 2X Long Zeta Global (ZETA) ETF — an AI-powered marketing software company — not a Zscaler play. This distinction matters for the thesis. Zeta Global is a pure-play AI/data-driven software name, which fits the software rotation argument well, but it carries a very different risk profile than a cybersecurity name like Zscaler. The thesis below reflects this correctly.
The Core Logic: Why Leverage Specifically?
The contrarian thesis established that software and semiconductors were suffering a narrative-driven selloff, not a fundamental one. But knowing what to buy is only half the decision. The more important question was how much to buy — and in what structure.
Leveraged single-stock and sector ETFs made sense in this context for a specific reason: the recovery, when it came, would not be gradual. Mean reversion from deeply oversold conditions tends to be violent. Short-covering events are spike-driven. Narrative reversals get priced in over days, not quarters. In that environment, the compounding amplification of a 2x or 3x instrument — when held through a directional trending move — produces outsized returns relative to direct stock exposure.
Leverage Amplification — How the Math Works On a Trending Move
Underlying stock up +5% in one session
2x ETF target: +10.0%
Underlying stock up +10% over trending week
2x ETF approx: +20–22% (positive compounding)
Underlying stock up +15.2% (ServiceNow, Mar 3–7)
NOWL approx: +32–36% (compounding tailwind)
SMH basket up +8% in recovery week
SMH3 (3x) approx: +25–28%
Note: positive compounding benefits leveraged ETFs during trending up moves. The daily reset works in your favor when the underlying moves consistently in one direction.
The key insight is that leveraged ETFs are not permanently value-destroying instruments — they are volatility-sensitive instruments. They decay in choppy, oscillating markets. They amplify in directional, trending markets. You were betting on a directional trend: the reversal of an oversold narrative. That is precisely the scenario where leverage compounds in your favor.
Why Broadcom, and Why 2x?
Broadcom is one of the most structurally unique semiconductor companies in the world, and its inclusion in the semicon sleeve carries a logic that goes beyond simple chip exposure. AVGO straddles two worlds: it is both a semiconductor company and a software company, having acquired VMware in a landmark $61B deal that transformed it into a diversified infrastructure platform. This dual nature made it uniquely levered to both of the recovery narratives playing out in Q1 2026.
AI Custom Silicon Moat
Broadcom is the leading designer of custom AI accelerators (XPUs) for hyperscalers including Google, Meta, and Apple. As AI infrastructure investment accelerated, Broadcom's AI revenue segment grew materially — a tailwind that was entirely structural, not narrative-dependent.
VMware Synergy Unlocking
The VMware integration had been a drag on multiple in 2025 as investors waited for synergies to materialize. By Q1 2026, signs of cross-sell and margin expansion were emerging, providing a catalyst independent of the broader software selloff narrative.
Oversold From Software Fear
AVGO was punished alongside pure-play SaaS stocks during the "SaaSpocalypse" despite having limited exposure to the per-seat subscription model at risk. The selloff created mispricing — AVGO was discounted for a risk it didn't own.
AVGX 52-Week Range Signal
AVGX ranged from $9.87 to $70.31 over 52 weeks. Entry near the lower bound of that range in late Feb/early March meant buying a 2x instrument at a fraction of its peak NAV — maximum upside asymmetry with defined loss exposure.
The Leverage Thesis in One Sentence: AVGO was discounted for SaaS disruption risk it doesn't own, with a structural AI silicon tailwind the market was momentarily ignoring. 2x leverage on that mispricing, during a trending mean-reversion, was the high-conviction expression of that view.
The Only 3x Position — Rationale for Maximum Leverage
SMH3 is the most aggressive position in the portfolio — 3x daily leverage on the entire VanEck 25-name semiconductor index. The choice to hold this rather than a 2x equivalent was a deliberate expression of conviction on semiconductors specifically: this was the one sector where the fundamental case was strongest and the narrative damage most clearly temporary.
Sector Was Collateral Damage
Semiconductors didn't cause the "SaaSpocalypse" narrative — they were caught in the blast radius. Nvidia, TSMC, ASML, and Qualcomm had no per-seat subscription revenue. They were sold because they're in "tech." That's indiscriminate, not analytical.
AI Infrastructure Spend Was Intact
Hyperscaler capex commitments — Microsoft, Google, Meta, Amazon — for 2026 data center buildout were not reduced. Semiconductor demand from AI training and inference remained the single largest growth driver for the sector. The selloff was against a rising fundamental tide.
Jensen Huang Catalyst Specific
When Jensen Huang defended software incumbents on Feb 26, he simultaneously validated Nvidia and the chip ecosystem. The speech was a confidence catalyst for semiconductors specifically — "AI is good for chips" was never in question; "AI is good for software" was. Huang resolved the latter, and chips ripped as a double-beneficiary.
3x on a Trending Move = Max Capture
When semiconductors began their post-Huang recovery trending upward, a 3x structure captured every day's move tripled. Positive compounding on a sustained multi-day directional move is where 3x structures earn their risk premium decisively.
Why the basket over a single name: SMH3 provided diversified upside across all 25 constituents — capturing Nvidia, TSMC, Broadcom, AMD, ASML, and Qualcomm simultaneously. No single-name risk. If any one chip name lagged, the basket absorbed it. This was the right structure for a sector-wide recovery call.
Important note on SMH3: This trades on the London Stock Exchange (LSE), making it primarily accessible to non-US investors — which matches your Petaling Jaya base. For US-based accounts, the equivalent would be SOXL (Direxion 3x Semiconductor Bull).
The Flagship Contrarian: The Most Mispriced Name
If there was one stock at the center of the "SaaSpocalypse" narrative — the name that embodied the fear more than any other — it was Salesforce. It fell 26% year-to-date while simultaneously posting a 24.9% EPS beat. This is the definition of narrative-driven price destruction. CRMG was the highest-conviction expression of the thesis that the market had gotten it categorically wrong.
Agentforce ARR: The Evidence No One Listened To
$800M ARR growing 169% year-over-year. Salesforce's own AI agent product — Agentforce — was growing at a hypergrowth rate while the stock was being sold as an AI victim. The market was pricing AI as a destroyer of Salesforce revenue while Salesforce was demonstrating AI as its fastest-growing revenue line.
Anthropic Partnership — Direct Beneficiary
Of all the companies that bounced on Feb 24 when Anthropic announced its enterprise integrations, Salesforce was the most prominent — up 5% in that single session. CRM had a named partnership in the announcement. CRMG therefore provided 2x the most directly catalyzed name.
Valuation: Cheapest in a Decade
At its trough, CRM traded at forward P/E multiples not seen since before it was widely recognized as a mission-critical enterprise platform. You were getting a top-5 enterprise software company, growing AI ARR at 169%, at a discount to its 5-year average valuation. That is the margin of safety argument fully intact.
Workflow Moat: Data Lock-In
Salesforce holds a decade of enterprise customer relationship data across industries. AI agents need that data to function. Agentforce doesn't replace the Salesforce platform — it runs on top of it, inside it, and depends on it. The moat deepens with AI, it doesn't erode.
The CRMG Thesis in Full: Stock down 26%, earnings up 25%, AI ARR growing 169%, and the company has a named partnership with the exact firm that caused the panic. At 2x leverage, a reversion even halfway to fair value would generate a return that justifies the position size. This was the most asymmetric risk/reward in the portfolio.
The Quality Play: Best-in-Class Enterprise Software
If Salesforce was the most mispriced name, ServiceNow was the highest quality name at an irrational discount. ServiceNow has one of the best fundamental business profiles in all of enterprise software — a platform that has never missed a quarterly revenue guidance since going public, with net retention rates consistently above 120%, and a product set so deeply embedded in enterprise IT and workflow infrastructure that replacement is measured in years and tens of millions of dollars of transition cost per customer.
Never Missed Revenue Guidance
ServiceNow's track record of execution is exceptional. During the Q1 2026 selloff, NOW fell 9–11% in single sessions despite intact subscription growth guidance. This was pure sentiment-driven selling, not fundamental concern — making the 2x recovery bet a clean risk/reward calculation.
IT Workflow = AI Native Platform
NOW is positioned as one of the clearest AI-integration winners in enterprise software. Its Now Platform is already being used to deploy agentic AI for IT service management, HR workflows, and customer service. Jensen Huang specifically named ServiceNow as a platform AI agents would use — not replace.
+15.2% Recovery in One Week
ServiceNow gained 15.2% in the week of March 3–7, one of the best single-week performances in its public history. NOWL (2x) would have captured approximately 32–36% of gain in that same period, depending on daily compounding. This validated the leveraged thesis precisely.
52-Week Range Signal
NOWL ranged from $5.55 to $29.01 in its 52-week span — a more than 5x range. Entry near the lower bound meant buying a high-quality underlying at peak emotional distress, with a 5x recovery ceiling if the underlying simply returned to its recent peak.
Why NOWL over direct NOW shares: ServiceNow's recovery from the trough was one of the fastest and most decisive in the sector. A 2x instrument during that trending multi-day move captured roughly twice the gain — approximately 32–36% vs 15–17% for the common stock — in the same holding period. That is precisely what leveraged ETFs are designed to do when the underlying is trending.
The High-Beta Wild Card: AI-Native Software at Maximum Discount
Critical Clarification: ZETX tracks Zeta Global (ZETA) — an AI-powered marketing and customer data platform — not Zscaler. Zeta Global is a fundamentally different business: it sells an AI-native marketing intelligence platform to enterprise clients, monetizing on usage of its data cloud and AI decision engine. This is important context for understanding the thesis.
Zeta Global is one of the purest expressions of the "AI as revenue driver" thesis in enterprise software. Unlike legacy SaaS names that are adapting to AI, Zeta was built with machine learning and data intelligence as its foundation. Its core product — the Zeta Marketing Platform — uses AI to process billions of consumer data points and deliver personalized marketing across every digital channel in real time.
AI-Native Architecture = No Displacement Risk
The bear case against software was that AI agents would replace legacy SaaS workflows. Zeta Global has no "legacy" to replace — its platform is the AI. It was being sold as part of the broad software selloff despite being structurally immune to the specific risk that was driving the selloff.
Launched Feb 3, 2026 — Recent ETF
ZETX launched on February 3, 2026 — almost exactly when the selloff was at its worst. Entering a freshly launched 2x ETF on an AI-native software name at peak sentiment pessimism, shortly after launch, meant capturing the early liquidity build alongside the fundamental recovery. A genuinely contrarian timing combination.
High Beta = Maximum Amplification
Smaller-cap, higher-beta software names tend to fall harder and recover harder than mega-caps. ZETA, as a mid-cap AI software name, had higher relative volatility than Salesforce or ServiceNow — which meant greater absolute drawdown during the panic, and greater absolute recovery potential when sentiment turned.
Enterprise AI Monetization Proof Points
Zeta's platform processes over 500 billion data signals daily. Enterprise adoption of AI-native marketing tools was accelerating — a trend independent of, and actually helped by, the broader adoption of AI in enterprise workflows. More AI = more data = more need for Zeta's intelligence layer.
The ZETX Role in Portfolio Construction: ZETX was the high-beta kicker. In a portfolio otherwise populated by mega-cap or near-mega-cap names (CRM, NOW, AVGO), ZETX provided exposure to a smaller, higher-beta, AI-native software name that would disproportionately benefit from a full sentiment recovery — with 2x leverage amplifying that asymmetry further.
Why This Allocation Mix Made Sense
73/27 Software vs Semiconductor — The Logic
The core thesis was a narrative reversal story, and the narrative was specifically about software. Semiconductors were collateral damage; software was the epicenter of the panic. This warranted a heavier allocation to software as the higher-conviction, more directly catalyzed side of the trade. Semiconductors provided diversification and the 3x leverage kicker via SMH3, but the alpha was expected to come primarily from the software recovery.
Three Names in Software — Why Not Concentrate?
CRMG, NOWL, and ZETX represent three distinct archetypes of the software recovery: Salesforce (the most mispriced mega-cap with a direct Anthropic catalyst), ServiceNow (highest quality name at irrational discount), and Zeta Global (AI-native high-beta name with maximum sentiment leverage). Together they capture the recovery across market-cap tiers and business model archetypes — not three similar bets, but three complementary expressions of the same macro thesis.
AVGX + SMH3 in Semicon — Single Name + Basket
Pairing a single-name 2x (AVGX/Broadcom) with a basket 3x (SMH3/SMH index) provided two things: concentrated upside if Broadcom specifically outperformed (driven by its AI silicon moat and VMware synergy thesis), and diversified protection via the 25-name basket if the recovery was broad rather than name-specific. This was intelligent construction — conviction on a name, insurance via a basket, 3x on the basket to compensate for the lower single-name concentration.
▶ Portfolio Verdict
This portfolio was built to do one thing: capture maximum upside from a narrative reversal that fundamentals had already made inevitable. Each position was leveraged precisely because the recovery, when it came, would be fast and trending — the only market condition where leveraged ETFs' compounding mechanics work systematically in your favor rather than against you. The 73/27 split placed the heaviest weight exactly where the mispricing was most acute. This was not speculative. It was conviction, structured.
What You Must Watch
⚡
Leveraged ETF Decay Warning: These instruments are daily rebalancing vehicles. If the underlying stocks enter a choppy, oscillating market — moving up 3% one day and down 3% the next with no net directional progress — volatility decay will erode your position even if the underlying ends flat. The thesis depends on a trending recovery, not a sideways one. Monitor momentum closely.
The Three Things That Would Break This Thesis:
1. Q1 2026 Earnings Disappoint (April). If CRM, NOW, and other holdings report deteriorating ARR growth, net retention declines, or guide revenue lower, the "fundamentals are fine" pillar of the bull case collapses. This would confirm the bear thesis and accelerate multiple compression. Leveraged positions would be severely punished — 2x on the downside.
2. Agentic AI Adoption Accelerates Faster Than Expected. If enterprise customers start meaningfully reducing seat counts or cancelling SaaS subscriptions due to AI agent automation within the next two quarters, the structural bear case stops being a future risk and becomes present-day evidence. Watch net revenue retention rates carefully.
3. Volatility Spike / Macro Shock. A geopolitical escalation, unexpected Fed hawkishness, or credit market stress could send risk assets down sharply. Leveraged instruments amplify drawdowns as viciously as they amplify recoveries. A 15% drawdown in CRM becomes a 30%+ loss in CRMG. Know your stop.
The Position Management Imperative: Leveraged ETFs reward active management. These are not set-and-forget holdings. As the narrative recovery plays out through the April earnings cycle and the positions appreciate, systematic profit-taking — trimming into strength — is how you capture the thesis gain without giving it back to volatility decay. The thesis has a time horizon. Don't let it become an unmanaged long-term hold.