MRVL

Summary

**Rating: Overweight**

**Entry Strategy:** Maintain existing positions; initiate a 25% starter position at current levels (~$212) with a hard stop at $197 (below VWMA/Bollinger middle confluence). Add the remaining 75% on a confirmed close above the 50 SMA at $241 or on a pullback to the $197-206 support zone that holds.

**Position Sizing:** Full position = standard allocation. Deploy only 25% now given the binary CPI/earnings overhang. Scale in on confirmation, not ahead of it.

**Key Risk Levels:** Support at $197-206 (VWMA, Bollinger middle, 10 EMA). Resistance at $228.80 (Bollinger upper) and $241 (50 SMA). Hard stop at $197 — a close below invalidates the recovery thesis. July low at $163.40 is the line-in-the-sand for thesis break.

**Time Horizon:** 5-20 trading days. The CPI print today and upcoming earnings are binary catalysts that will resolve the current equilibrium. Position for the resolution, not against it.

The case

## Investment Thesis

Our analysis sharpens the MRVL picture around three key insights. First, the normalized P/E of 130-140x, the interest expense surge ($51M → $256M quarterly), and the crowded retail positioning (2.75:1 bull/bear, put/call 0.31) are real concerns — but they are valuation concerns, not business-direction concerns. Second, the stock sits in a genuine equilibrium between $197 support and $241 resistance with declining ATR, coiling for a directional move. Third, the Microsoft Maia 300 catalyst is being discounted now, not in 2027 — the supply chain for 300,000+ custom AI chips starts with qualification and design wins today, and the NVIDIA $2B alliance expansion is concrete validation, not narrative.

**What tips the scale:** The Maia 300 as a 2026 design win being priced now is the decisive insight. The market is a discounting mechanism, and the realistic downside to the first major support ($197-206) is only 3-7%, not the 35% to the 200 SMA that a worst-case scenario would imply. The risk/reward is asymmetric to the upside when measured against the nearest support, not the worst-case scenario.

### Why Not Hold?

The past-decision record is the most damning evidence against a Hold. Nine resolved calls on MRVL with a 0.25 hit rate and mean alpha of only +0.028. Five of nine were Holds, and 60% of those Holds resulted in opportunity loss — the stock moved directionally while we sat neutral. The most painful: [2026-06-01 | MRVL | Hold | +37.5% | +38.0% | 2d]. The recurring failure mode is clear: Hold or Underweight ratings on MRVL repeatedly miss large directional moves. This is systematic, not idiosyncratic.

Our own prior work acknowledges: "The bull is right about the direction of the business; the bear is right about the price being paid for it." But the past record shows that on MRVL, being right about the business direction matters more than being right about the price. The secular AI tailwind has repeatedly overwhelmed valuation concerns. Defaulting to Hold because "neither side has a knockout" is exactly the error that produced a 0.25 hit rate.

### Why Overweight, Not Buy?

I stop short of Buy because the binary event risk is real and concentrated. The July CPI report today and the upcoming earnings with 99.7% IV create genuine two-way risk. The stock is below its 50 SMA — the medium-term trend is unconfirmed. The crowded retail positioning is a contrarian yellow flag. An Overweight captures the asymmetric upside while respecting the near-term uncertainty. It says: "The edge is real but not overwhelming; size accordingly."

What would weaken it

• CPI/earnings binary event risk with 99.7% IV creates genuine two-way risk
• Stock below 50 SMA — medium-term trend unconfirmed
• Crowded retail positioning (2.75:1 bull/bear) is a contrarian yellow flag