Verdict: REMOVE

Benchmark weight 0.58%. The regression is unambiguous: CRWD is a 1.8x leveraged cybersecurity ETF with zero orthogonal alpha. Keeping it adds amplified sector concentration without compensation. The fundamentals are excellent — that's consensus, not edge.

The Setup

CrowdStrike just printed a blockbuster Q4 FY26 (reported March 4). Record net new ARR of $331M (+47% YoY), ending ARR $5.25B, non-GAAP operating income $326M (25% margin), free cash flow $376M (29% margin). Gross retention 97%, net retention 115%. FY27 guide: $6.47-6.52B ARR (+23-24%), $5.87-5.93B revenue, FCF margin 30%+. Q1 pipeline grew 49% YoY. Charlotte AI usage up 6x, AIDR product up 5x QoQ.

The stock is at $392, down 31% from its $567 52-week high. RSI 33. Below both the 50-day ($426) and 200-day ($468) moving averages. Forward P/E 63.6x. Next earnings June 9 — 11 weeks away.

A company reporting record everything, trading at oversold levels, in a sector with genuine secular tailwinds. The temptation is to call it a buy. The math says otherwise.

The Factor Regression Kills It

90-day regression (daily returns, 2025-11-14 to 2026-03-26):

r_CRWD = 0.06% + (-0.31)×r_QQQ + 1.78×r_CIBR + ε

Alpha (annualized):   +0.06%   (t=0.00, p=0.999 — statistically zero)
Beta to QQQ:          -0.31    (t=-1.60, p=0.113 — not significant)
Beta to CIBR:         +1.78    (t=12.72, p=0.000 — highly significant)
R-squared:             71.4%
Idio variance:         28.6%   (target: 75%+)
Orthogonal Sharpe:     0.00

CRWD's returns are almost entirely explained by the cybersecurity sector. The 1.78 beta to CIBR means CRWD moves 1.8x the cyber ETF — it's a leveraged sector bet, not a stock pick. The negative QQQ beta is not significant; once you control for cybersecurity, CRWD has no independent relationship with broad tech.

The univariate regression against QQQ alone is misleading — it shows 81.8% idiosyncratic variance, which looks great. But that "idiosyncratic" component is really CIBR sector exposure that the one-factor model can't capture. This is exactly the mistake the framework warns about: apparent alpha that's really unmodeled factor exposure.

At 28.6% idiosyncratic variance, CRWD is 71% factor-driven. In a portfolio targeting 75%+ idio variance, this stock dilutes the signal. Every dollar allocated to CRWD is a dollar of cybersecurity beta you can get from CIBR at 20 bps, not a dollar of stock-specific alpha.

Why Zero Alpha Despite Great Fundamentals

$100B market cap. 30+ sell-side analysts. 3.1% short interest (light). Forward P/E of 63.6x reflects the growth trajectory precisely.

The AI-tailwind thesis — AI expands the attack surface, agentic workflows create new endpoint types, AI-powered attacks demand AI-powered defense — is correct. It's also consensus. Every sell-side shop runs the same thesis: Wedbush ("AI inflection"), TD Cowen (Buy, "platform consolidation"), Stifel (Buy), Oppenheimer (Buy). When we searched 3,962 earnings transcripts, zero companies reported reduced cybersecurity spending from AI substitution. The category error of the February selloff (Anthropic's Claude Code Security is SAST, not XDR) has been thoroughly debunked.

All true. All priced. Edge = P_you - P_market = 0.

The counterparty on a CRWD long is every informed participant: Goldman, Citadel, Two Sigma, 30 sell-side desks, the entire institutional cybersecurity complex. Nobody is uninformed here. Nobody is forced. There is no ignorant counterparty to harvest.

What the Non-GAAP Numbers Hide

Three accounting details worth examining:

GAAP reality. Non-GAAP operating income was $1.05B for FY26. GAAP net income was -$161M. The gap: $1.097B in stock-based compensation — 23% of revenue, accelerating ($649M in FY24, $861M in FY25, $1,097M in FY26). CRWD is profitable on paper and loss-making in GAAP. At some point, the $1.1B annual equity dilution is a real cost to shareholders.

Commission amortization change. In February 2026, CRWD extended the amortization period for deferred sales commissions from 4 years to 5 years. This adds $85-95M to FY27 non-GAAP operating income — roughly 6% of the $1.42-1.46B guided number. It's an accounting change, not an operational improvement. The cash went out the door in prior periods; they're just spreading the P&L recognition over a longer window. Meanwhile, capitalized commissions sit at $1.103B on the balance sheet ($447M current, $656M noncurrent), with $704M capitalized against only $449M amortized in FY26.

July 2024 litigation. The July 19 Incident triggered multiple legal proceedings: securities class actions on behalf of stockholders, derivative suits against officers and directors, consumer class actions, and government inquiries. The 10-K states insurance "will not cover all costs, claims and liabilities." No liability has been quantified. CCP (Customer Commitment Package) accounts have higher retention than company average and expanded 2x+ the $80M ARR provided — the operational damage is lapped. The legal tail is not.

None of these individually are thesis-breaking. Together they illustrate why "record profitability" deserves scrutiny. The FY27 operating income guide bakes in a $85-95M tailwind from the amortization change, partially offset by $74-80M of acquisition opex from four recent deals (SGNL, Seraphic, Onum, Pangea). Net: the operational beat is smaller than the headline suggests.

Last Week's -8.3% Decline, Decomposed

March 24 was the worst day: CRWD -4.92% while CIBR -2.95% and QQQ -0.68%. That's roughly 2% of idiosyncratic damage in a single session — C-suite Form 4 filings (Kurtz, Podbere, Sentonas, Saha selling around March 23-26), though prior analysis confirms all recent CEO/CFO sales are sell-to-cover for RSU tax withholding, not discretionary. The headlines read worse than the signal.

Over 90 days: CRWD -25.9% vs CIBR -15.3% vs QQQ -5.6%. The regression predicts this nearly exactly: CIBR -15.3% × 1.78 beta = -27.2%, close to actual -25.9%. The model fits. There is no hidden alpha.

The Moat Is Real — And Irrelevant for Filtration

To be clear about what's working:

• Sensor network: 1T+ events/day across 176 countries, 15PB structured security data. No LLM replicates runtime endpoint telemetry.
• Expert-labeled training data: MDR analysts, threat hunters, incident responders produce labeled data as a byproduct of operations. Cyber RLHF at scale.
• Retention: 97% gross, 115% net — best in class and stable across 5+ years of "MSFT will kill them."
• Microsoft coexistence: Defender share gains come from legacy AV (McAfee/Symantec), not CRWD displacement. Gartner MQ: CRWD #1 both axes, 3rd straight year. The "MSFT bundling" bear case has been the #1 short thesis for half a decade without materializing.
• Platform expansion: Cloud $800M+ ARR (+35%), Identity $520M+ (+34%), Next-Gen SIEM $585M+ (+75%). Flex subscription model ($1.69B ARR, +120% YoY) unlocking unprecedented cross-sell.

All real. All reflected in the $100B market cap. A strong moat in a well-covered mega-cap is a reason to own it in a diversified portfolio — it is not alpha for a filtration strategy.

Kill Conditions (What Would Change This to KEEP)

• Idiosyncratic variance climbs above 50% — a company-specific catalyst detaches CRWD from the cybersecurity sector
• CRWD-specific litigation resolution creates a step-function repricing event
• Regression alpha turns positive and significant (new product cycle, acquisition synergies, etc.)
• Cybersecurity sector establishes persistent outperformance vs QQQ, making the 1.8x sector beta an asset rather than uncompensated risk

None of these are visible on the 15-week horizon.

15-Week Expected Path

Weeks 1-4: Technical digestion continues. RSI 33 may trigger a bounce, but with no catalyst, it's noise. Cybersecurity sector direction drives 70%+ of CRWD's variance — if CIBR rallies, CRWD rallies 1.8x. If CIBR fades, CRWD amplifies the loss. You're betting on the sector, not the stock.

Weeks 5-10: Dead zone. No catalysts, no earnings. Momentum negative. The $426 50-DMA acts as resistance. Management is on the conference circuit but no new data points until June.

Weeks 11-15: June 9 earnings (Q1 FY27). Guide is $249-251M net new ARR (+29-30% YoY), $1.36-1.36B revenue (+23-24%). Deceleration from Q4's 47% net new ARR growth already baked in. At 63.6x forward P/E, the bar is elevated — a "meet" won't re-rate the stock.

P(CRWD outperforms average QQQ constituent over 15 weeks): ≈40%. P(roughly matches): ≈25%. P(underperforms): ≈35%. Expected excess return: -2% to -3%, driven by amplified sector beta in a softening tech environment plus no idiosyncratic catalyst.

Portfolio Impact

CRWD at 0.58% weight. Removing it redistributes ≈1.5bp to each of ≈40 survivors. De minimis individual impact, but the principle matters: every name in the basket should earn its weight through idiosyncratic content, not factor beta. CRWD fails that test.