The AI infrastructure narrative has pushed mega-cap tech stocks to record valuations, with Alphabet, Microsoft, Amazon, and Oracle adding $1.9 trillion in market cap in just three days. But the foundation of that rally is cracking: the cost of debt for th

Verdict: avoid shorting mega-cap momentum until the technical reversal actually confirms

The bearish thesis has genuine teeth: Oracle's free cash flow swung from positive $11.8B in fiscal year 2024 to negative $23.7B, Amazon's Trainium 2 chips are reportedly being replaced after only 20 months, and per Bloomberg the era of cheap AI capital is ending as Treasury auction dynamics push yields higher. Yet the technical setup designed to time this reversal is painfully thin — the 24-month backtest on Amazon produced just 13 trades with a 38.5% win rate and a 3.1% net return against a 12.3% maximum drawdown, meaning the risk-reward was unfavorable even when the entry conditions did fire. More damning, the 60-month evaluation produced zero triggers across 1,788 bars, so the exact confluence of RSI above 65 with simultaneous MACD and EMA bearish crossovers has never historically co-occurred. The strongest counter-evidence is that Microsoft — the thesis's own reference point — posted a 46.8% operating margin (98.9th percentile) and $67.0B in free cash flow (99.3rd percentile), essentially self-funding its AI buildout with a negligible 0.07 debt-to-equity ratio. Oracle's 9-day EMA is only $4.03 above its 50-day EMA, making it the closest to triggering, but until the crossover confirms, you are shorting mega-cap momentum on conviction alone. No robust parameter setup was established, so this configuration stands as-is — wait for confirmation or stay away. **Conviction breakdown** - **Thesis support (70/100):** Oracle's negative $23.7B FCF and Amazon's $131.8B capex give the capital-destruction argument real weight, supported by Bloomberg's funding-cost analysis. - **Trade readiness (20/100):** Only Oracle's EMA crossover is near; all three tickers still lack the confirmed MACD and EMA bearish crossover required for entry. - **Risk quality (35/100):** The 2:1 reward-to-risk ratio (2.4% stop, 4.8% target) is sound on paper, but the 12.3% backtest drawdown against a 3.1% return and high kurtosis (AMZN 13.8, MSFT 16.3, ORCL 17.4) undermine execution confidence. - **Backtest evidence (20/100):** A 38.5% win rate over only 13 trades is statistically unconvincing, and zero triggers over 1,788 bars means the pattern has no historical precedent over a 5-year horizon. - **Fundamentals trend (65/100):** Oracle's FCF deterioration and Amazon's margin compression support the bear case, but Microsoft's 46.8% operating margin and Amazon's 50.3% gross margin show the largest names remain fundamentally robust.

Trade now

All three names — Amazon, Microsoft, and Oracle — currently sit in overbought territory with RSI (14) readings well above the 65 threshold the entry rules require. Amazon trades at $278.09 with an RSI of 69.9, Microsoft sits at $504.95 with an RSI of 80.0, and Oracle is at $151.04 with an RSI of 74.3. That condition is met across the board. What has not happened yet is the bearish crossover confirmation: the strategy demands a simultaneous MACD line cross below its signal line and a 9-day EMA cross below the 50-day EMA, alongside price tagging or exceeding the nearest resistance level. The nearest candidate to watch is Oracle, where the 9-day EMA ($146.26) sits only $4.03 above the 50-day EMA ($142.22) — the smallest gap among the three. By contrast, Microsoft's 9-day EMA is $60.37 above its 50-day EMA, and Amazon's is $18.96 above, so those two would require substantially larger momentum deterioration before the EMA crossover leg fires. On MACD, all three are still in positive-cross territory but are flagged as near a potential bearish crossover. In concrete terms, waiting means monitoring daily closes for the MACD line to roll below its signal line while the 9-day EMA continues converging on the 50-day EMA. Risk management is tightly defined. The strategy enforces a hard stop loss at 2.4% and a take profit at 4.8%, producing an effective reward-to-risk ratio of roughly 2:1. Additionally, if the 60-bar holding period is reached or a MACD bullish crossover occurs, the position exits on signal. The 24-month backtest on Amazon produced 13 trades with a 38.5% win rate, a 3.1% net return, and a 12.3% maximum drawdown — modest but positive results that reflect the strategy's restrictive entry posture. No robust parameter setup was established through sensitivity analysis, so the current configuration should be taken as-is without threshold adjustments. Because the entry rules require multiple bearish crossovers to confirm simultaneously while RSI remains elevated, the practical instruction today is to set conditional alerts on each ticker for MACD bearish crossovers and EMA…

Scores

• Conviction score breakdown: 42
• Thesis support: 70
• Trade readiness: 20
• Risk quality: 35
• Backtest evidence: 20
• Fundamentals trend: 65

Watch items

• ORCL — EMA (9) vs EMA (50) gap
• AMZN — RSI (14)
• MSFT — RSI (14)
• ORCL — MACD (12,26,9)
• AMZN — EMA (9) vs EMA (50) gap
• MSFT — EMA (9) vs EMA (50) gap
• AMZN — RSI (14) above 65
• AMZN — MACD (12,26,9) crossed below MACD (12,26,9)
• AMZN — EMA (9) crossed below EMA (50)