History Says the Nasdaq-100 Does This in September. AI Investors Should Take Note.

The article highlights seasonality risk: from 2017-2025 the Nasdaq-100 fell an average of 2.1% in September and was down 67% of the time (e.g., -10.6% in Sept 2022). It argues against trying to time exits by showing that, over 1996-2025, 48% of the S&P 500’s best days occurred during bear markets and that missing the 10/20/30 best days would have cut ending value by ~56%/74%/84%. For AI-heavy investors, it notes Nvidia’s Septembers have tended to be weak (e.g., -19.5% in Sept 2022) but historically rebounds in Oct-Nov (average +5%+), implying potential opportunistic adding on pullbacks rather than selling into September.

Analysis

This is less a fundamental warning than a positioning and liquidity setup. In a tape where AI is the dominant index driver, the vulnerability is multiple compression, not earnings deterioration: crowded owners will trim beta first, and that mechanical de-risking usually transmits from NVDA into the rest of the semiconductor complex, including AMD, ASML, and MU. The first loser is often the highest-duration factor exposure, so even if order books stay intact, the market can still punish the group on higher vol and lower risk appetite.

The key horizon is 2-4 weeks: a weak September can be tradable, but only if it comes with risk-off flows rather than a true revisions break. What would reverse it is evidence that hyperscaler capex checks remain firm, rates fall, or NVDA holds up through the first calendar week without losing relative strength versus QQQ; that would tell you seasonality is being overpowered by fundamentals and positioning. If instead semis underperform while breadth narrows, expect a rotation into defensives and a sharper unwind in the most crowded AI names.

Contrarian take: the market may already know the seasonality, so the trade is not to short everything indiscriminately but to buy protection selectively when complacency returns. The better expression is relative-value hedging, because a benign macro print or strong earnings guidance could quickly end the seasonal scare and leave outright shorts exposed. The signal is probably overdone as a market call, but still useful as a timing tool for reducing gross and buying hedges on strength.

Key Decisions for Investors

• Short SMH or QQQ via 2-4 week put spreads into early September; use a 3-5% downside target and cut if QQQ regains its 20-day moving average with improving breadth.
• Trim overweight NVDA exposure on any 2-3% rally; this is the highest-beta expression of the seasonal trade and the most likely source of forced de-risking if vol rises.
• Pair trade: long XLP or XLU vs short a semiconductor basket (NVDA/AMD/MU) for the next 3-6 weeks; thesis is multiple compression in crowded growth names while defensives absorb flows.
• If you need AI exposure, rotate from NVDA into ASML on weakness only after seeing stabilization in semis; ASML is less momentum-sensitive and gives cleaner exposure if the selloff is only seasonal.
• Set an alert on NVDA relative strength vs QQQ: if it underperforms for 5 trading sessions while rates are stable, treat that as confirmation to increase hedges; if it outperforms, seasonality is likely failing.