Called Before the Turn: Our Model Flagged the Correction Before the S&P 500 Fell 10%

Our Model’s Correction signal fired in November. On November 20, the S&P 500 crossed into correction. Here is what happened, what our model saw ahead of time, and what it is saying now.

By Kip Lytel, CFA · Founder and Managing Wealth Advisor, Montecito Capital Management

The week in brief

Stocks had their roughest week in months. On Friday, November 20, the S&P 500 fell 3.7% to close at 6,910.79, 11.6% below its October 6 record close of 7,818.93, which puts it in a correction (a decline of 10% or more from a record). The Dow Jones Industrial Average fell 1.9% to 48,147.42, and the Nasdaq Composite fell 5.3% to 22,617.73, erasing its gain for the year. Even after the drop, the S&P 500 is still up about 1.0% for 2026 and the Dow slightly positive, while the Nasdaq is down about 2.6% for the year.

IndexClose, Nov 20FridayWeek
Nasdaq Composite22,617.73−5.3%−10.0%
S&P 5006,910.79−3.7%−8.6%
Dow Jones Industrial Average48,147.42−1.9%−8.0%

Weekly change measured from the November 13 close. Sorted by size of the weekly decline, largest first. Index levels are price only.

Three forces drove the selling:

  • The AI trade unwound. Chip stocks led the decline. Anthropic, which listed this week, gave back its opening gains and traded near its IPO price. Reports that the SEC is examining potential accounting issues at Nvidia, a week after its CEO stepped down, hit the largest stock in the index. Overnight, Taiwan, Japan and South Korea each fell more than 5%.
  • Policy risk. President Trump threatened a heavy tax on U.S. companies that build data centers abroad, adding a new risk to the industry that had carried the market.
  • A Federal Reserve leaning toward higher rates. Minutes released November 18 showed most officials expect further rate increases, and Governor Lisa Cook warned that waiting risks falling behind. Market fear rose sharply: the Cboe Volatility Index climbed to its highest level since March. Investors moved into Treasuries, and the 10-year yield eased back below 5%.

What our model saw ahead of time

RegimeSignal™, the quantitative market model we use as one input in our investment process, runs four separate warning signals, each built for one kind of market move: Pullback (about 5%), Correction (about 10%), Bear (20% or more) and Recovery. In November, both the Pullback and Correction signals fired, days before the S&P 500 crossed the 10% line on November 20. The signal came first; the correction followed.

The model was also seeing damage under the surface well before the headline index showed it. On the November 19 close, 180 of the 497 S&P 500 stocks it tracks were already 20% or more below their highs, in bear markets of their own, including 74% of technology stocks. Its Composite Market Score, a single 0 to 100 reading of the whole market, has fallen to 46.6 as of the November 20 close, down from 48.3 a day earlier, in its Caution band (40 to 60).

On the historical walk-forward evaluation, where each prediction was scored only with data available at the time and the results were reviewed independently, the four signals were right 84% of the time on average when they fired: Correction 84%, Pullback 83%, Bear 86%. These are historical model statistics, not a guarantee, and not the results of any client account.

What the model says now

The model has not identified a bear market. The Bear signal, the one built for declines of 20% or more, has not fired. That is the question that matters now: does this decline stop as a correction, or become a bear market? History offers a clear guide to what usually decides it.

The silver linings

Earnings remain strong. The blended S&P 500 earnings growth estimate for the current quarter is 31.9%, against a five-year average of 13.3%, with revenue growth of 14.4%. As the history below shows, bear markets usually arrive with a recession, and often with falling earnings. Today’s selloff is driven by valuations, the AI trade and interest rates, while profits are still growing. That is the strongest reason this decline may stay a correction.

If that changes, we expect to see it early. The Bear signal has historically called bear markets ahead of time. If it fires, we would look to reallocate sleeves of client portfolios toward safer havens, such as short-term Treasuries and our alternatives sleeve, in line with each client’s objectives and risk tolerance. Until then, we are holding our positioning and watching the Bear signal closely.

When corrections stay corrections

We studied every time the S&P 500 closed 10% or more below its record since 1950: 24 episodes. The single best dividing line was the economy. We used two gauges available at the time: whether the decline ran into a recession, and how fast S&P 500 earnings were growing when the index first crossed the 10% line.

Backdrop at the 10% lineEpisodesStopped short of a bear marketHigher 12 months later
No recession, earnings growing over 15%65 of 6 (83%)5 of 6
No recession1410 of 14 (71%)12 of 14
Recession103 of 10 (30%)4 of 10

S&P 500 daily closes, 1950 to November 13, 2026. Recessions as dated by the National Bureau of Economic Research, counted when one overlapped the decline or began within six months of its low. Earnings growth is trailing 12-month S&P 500 earnings versus a year earlier (Shiller data). The 2025 episode is counted as no recession; earnings data for it were not in the series, so it is left out of the earnings row.

Today’s backdrop sits in the strongest row: no recession, and expected earnings growth of about 32%, more than double the 15% threshold. The one exception in that row was 2022, when the Federal Reserve was raising rates at the fastest pace in four decades. With the Fed leaning toward more hikes now, that is the risk we are watching most closely.

How long do they last? In the 10 corrections since 1950 that came without a recession and stopped short of a bear market, the S&P 500 fell 14% on average from its record. From the day it first closed 10% down, it typically fell only about 4% more, usually found its low within weeks (median about two weeks), and regained its record in a median of about 3½ months. It was higher 12 months later in all 10 cases, by an average of about 17%.

Fall corrections have happened before. Three times since 1950, a correction without a recession crossed the 10% line in October or November: 1955, 1997 and 2018. In 1955 the S&P 500 was back at a record by mid-November; in 1997 it was back by early December. 2018 was harder: the decline deepened through December, close to the 20% line (19.8%), before a low on December 24, and the index regained its record by late April 2019. In all three, the bull market resumed. 2018 is a fair reminder that a recovery is not guaranteed to arrive on the calendar’s schedule, especially with a Fed raising rates, as it was then.

The week ahead

A holiday-shortened week with key data packed into three days:

  • Monday, November 23: S&P Global flash manufacturing PMI for November.
  • Tuesday, November 24: Conference Board consumer confidence.
  • Wednesday, November 25: October PCE inflation, the Fed’s preferred gauge, and personal spending, plus the second estimate of third-quarter GDP. A hot PCE reading would raise the odds of a December rate hike.
  • Thursday, November 26: Markets closed for Thanksgiving.
  • Friday, November 27: Typically a shortened session, with the stock market closing early.

Beyond the data, we are watching AI and chip stocks for signs of stabilization, follow-through on the Nvidia reports, and Federal Reserve commentary ahead of its December meeting.

What December and January have looked like historically

Seasonality is not a forecast, but the calendar has historically been on investors’ side at this time of year. From 1950 to 2025, on S&P 500 price changes:

  • December rose in 55 of 76 years (72%), with an average gain of 1.4%.
  • The “Santa Claus rally” window, the last five trading days of December and the first two of January, rose in 58 of 75 years (77%), averaging 1.3%.
  • January rose in 45 of 75 years (60%), averaging 1.1%.
  • December and January together rose in 49 of 75 years (65%), averaging 2.5%.

Still up for the year: The S&P 500 remains, barely, in positive territory for 2026. Since 1950 there have been 11 years when the index was still up for the year on November 20 despite sitting 5% or more below its high for the year. In all 11, it finished the year above its November 20 level, by an average of 3.6%. The closest match is 1978: up just 0.2% for the year and 11% below its high on November 20, much like today. It added about 1% by year-end and was about 9% higher a year later. The pattern held through the end of the year; 12 months out the results were mixed, higher in 6 of 10.

One more silver lining: in all 13 corrections since 1950 that stopped short of a bear market, the S&P 500 was higher 12 months after first closing 10% down, by about 16% on average. The difference between a correction and a bear market is exactly what our Bear signal is built to tell us.

Our bottom line

Our model warned of this correction before it arrived. It is not warning of a bear market today, and earnings growth remains strong. We are staying disciplined: holding positioning, keeping our alternatives and short-term Treasury sleeves ready, and watching the Bear signal. If you have questions about your portfolio, please call us at 805-965-7955 or reply to this email. We are glad to talk it through.

Important disclosures. This commentary is provided by Montecito Capital Management (“MCM”), a registered investment adviser, for informational and educational purposes only. It is not an offer, solicitation or recommendation to buy, sell or hold any security, and it does not consider any individual’s objectives, financial situation or needs. Registration as an investment adviser does not imply a certain level of skill or training. Please consult us before making any investment decision.

Affiliation and conflict of interest. RegimeSignal™ is developed and published by Cronus Market Intelligence, LLC, a company owned by Kip Lytel, CFA, who is also the founder and Managing Wealth Advisor of MCM. Mr. Lytel therefore has a financial interest in RegimeSignal. RegimeSignal is one of several inputs MCM considers; investment decisions are not made solely on its signals, and MCM clients are not required to use or subscribe to it.

Model statistics. Signal accuracy figures (including the 84% average and the individual signal rates) describe the model’s performance on a historical walk-forward evaluation and in independent reviews. They are hypothetical, prepared with the benefit of hindsight in their design, and do not represent actual trading, actual client accounts or the results any investor achieved. Hypothetical results have inherent limitations. A signal can fail to fire before a decline or fire when no decline follows, and past model accuracy does not guarantee future accuracy. The Composite Market Score is a proprietary summary measure and is not a prediction of returns.

Market data and history. Index levels and changes are price only, exclude dividends, and are from third-party sources believed reliable but not guaranteed; closing values are as reported shortly after the close on November 20, 2026. Seasonal and year-to-date statistics are based on S&P 500 daily closes from 1950 through 2025; the correction study uses S&P 500 daily closes from 1950 through November 13, 2026, NBER recession dates and Shiller earnings data. These studies are MCM’s own analysis, cover a small number of episodes, and are not indicative of future results; different definitions or periods would produce different figures. Indexes are unmanaged, and an investor cannot invest directly in an index. News items are summarized from published reports and may change.

Forward-looking statements. Statements about the outlook, possible portfolio adjustments and future market conditions are opinions as of the date written, may change without notice and may not come to pass. Any reallocation described is general and would be made only as appropriate for each client’s objectives and risk tolerance. Investing involves risk, including the possible loss of principal. Past performance does not guarantee future results. MCM’s Form ADV Part 2A is available on request.

Montecito Capital Management · 225 East Carrillo Street, Suite 203, Santa Barbara, CA 93101 · 805-965-7955 · mcapitalmgt.com

https://twitter.com/MontecitoCapMgt

Post

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top