Every S&P 500 bottom since July 2016 — and what they had in common

Each signal, the regime it fired inside, and the S&P 500's forward closes from that date:

Returns are price-only from the signal close. Cells marked pending have not yet matured at the corresponding horizon.

Median forward returns across the signals that have matured:

The S&P 500 returns somewhere around +9% in an average year. On the matured set, the year after a signal came in at more than double that, since the signals only fire after real stress and not during calm. That does not make any single future signal predictive in isolation. What it does give is a useful base rate for reading the next alert when it shows up.

The best calls were the deepest ones: the COVID-crash low on 2020-03-26 (+51% over the next year, +74% over two), the December 2018 bottom (+30% / +50%), and the October 2022 bear-market low (+19% / +59%). Each one fired close to the actual trough.

The less flattering pattern shows up when the stress drags on. The first 2018 fire on 2018-10-31 was early — the market kept falling, and the December fire arrived closer to the real bottom. The same shape repeated in 2025: 2025-03-17 fired into the tariff selloff, the market continued lower, and 2025-04-09 fired again near the actual low.

This isn't a bug. The engine fires when the rule's three gates clear — oversold reversal, stressed regime, confirmation — and inside a deeper or longer drawdown the same shape can resolve more than once. Bear-regime fires have to clear a higher confirmation bar for exactly that reason: false starts are worst there.

Three things showed up over and over. First, a real drawdown — not a 2–3% wobble. The regime classifier requires it. Second, exhausted downside momentum on both Williams %R timeframes at once. Third, the first sign of a turn off the floor. The engine does not predict the bottom. It waits for those three things and flags the moment, so the dip turns into a dated event on your calendar instead of a feeling buried in the headlines.

For ETF investors holding VOO, SPY, or IVV, that distinction matters. A broad-index alert can support a contribution plan or staged cash deployment, but it should not be treated as an instruction to deploy all cash on a single date. The historical record is useful because it shows the standout cases and the early fires side by side.