It’s that time of year again to offer my readers the highest conviction idea for the coming year. Last year, my bullish call on gold worked out extremely well (see
2025 High Conviction Idea: Gold). Gold prices soared in all currencies and was one of the best-performing asset classes for the year.
This year, I am going to try something different. Instead of giving you a fish so you can eat for the day, I are going to teach you how to fish. I offer you my most reliable market timing models, each of which had shown success rates at, or nearly, 100% accuracy.
Long-Term Timing Model
My market timing models come in various forms, each with a different time horizon. My market timing models tend to work well as buy signals, but often don’t work as sell signals. That’s a feature, not a bug.
The degree of reliability of my signals is remarkable. While some analysts have called their buy signals “table pounding buys”, I would go further by calling them a “what’s the credit limit on my credit card” buy.
With that preface, let’s begin with my long-term market timing model. This model relies on trending positive price for its buy signal and non-confirmation of price momentum for its sell signal. Here are the trading rules:
- Buy when the monthly MACD (bottom panel) of the NYSE Composite turns positive.
- Sell when the 14-month RSI of the NYSE Composite flashes a negative divergence when the underlying index makes a new high but the 14-month RSI doesn’t.
This model possesses the rare quality of flashing reliable buy and sell signals. The buy signals are shown as dotted blue lines on the chart and the sell signals are shown as dotted red lines. While the buy signals tend to be slow and does not spot the exact bottom, the risk-reward of an entry on the long side has been excellent. By contrast, the sell signals tend to be more timely and allowed me to sidestep imminent downdrafts.
The Zweig Breadth Thrust
Marty Zweig wrote about the Zweig Breadth Thrust buy signal in his book,
Winning on Wall Street, in 1986. The ZBT buy signal is triggered when the market exhibits strong price momentum by surging from an oversold to an overbought condition within 10 trading days. This is a rare condition and there have been only eight buy signals since Zweig published his book. The S&P 500 has been higher every time six and 12 months later.
There were three instances when the ZBT buy signal fizzled in the short run when the market corrected significantly after the initial buy signal. My own research found that these cases occurred during periods when the Fed was raising rates, so some caution is warranted under those circumstances.
A ZBT buy signal is generated when the ZBT Indicator (bottom panel) goes from oversold to overbought within 10 trading days. Oversold conditions are marked by vertical pink lines, which are relatively frequent, and overbought buy conditions are marked by dotted blue lines, which are rare.
Incidentally, ZBT Indicator oversold readings can be useful short-term buy signals, but they are not reliable enough to be considered in the category of “what’s the credit limit on my credit card” buy signals.
As well, the ZBT buy signal is asymmetric. It tells you when to buy, but it’s silent on when to sell.
The NAAIM Exposure Index
Another asymmetric trimming model uses the NAAIM Exposure Index. The National Association of Active Investment Managers is an organization of RIAs who manage the funds of individual investors. NAAIM conducts a weekly survey of their members to gauge their degree of bullishness or bearishness on the U.S. equity market.
I found low-risk long entry points whenever the NAAIM Index falls below the lower band its 26-week Bollinger Band. A decline below the lower BB is an indication of panic among RIAs, which is contrarian bullish.
The accompanying chart shows the history of buy signals based on the full history of the NAAIM Exposure Index that began in 2006. Each buy signal has occurred at or near the exact market bottoms with low downside risk.
Unlike my previous market timing models, which have long time horizons, the NAAIM Exposure Index is a short-term trading model.
My S&P 500 Bottom Spotting Model
Another short-term trading model that devised is my “bottom spotting model”. My technical analysis research found that overbought and oversold indicators can be useful ways to measure risk and reward, but overbought markets can become more overbought and oversold markets can become more oversold. Similarly, sentiment indicators can measure the degrees of investor enthusiasm and panic, but they are inexact market timing indicators.
While each of those short-term indicators is useful, my experience as an equity quant taught me to combine uncorrelated models to build models with more reliable signals. The result is the “Bottom Spotting Model”. My combined five technical analysis market indicators, each designed to measure slightly different overbought and oversold conditions, and sentiment. Further tests found that traders should buy whenever two or more components of the model flash buy signals simultaneously or within two days of each other.
The components are:
- VIX Index rising above its upper Bollinger Band, which is an oversold condition for the market.
- An inversion in the VIX term structure. The futures curve of the VIX is normally upward sloping, and an inversion is an indication of panic in the option market.
- An oversold condition in the NYSE McClellan Oscillator (NYMO).
- TRIN, also known as the Arms Index, measures advancing stocks to declining stocks relative to their respective volumes. A TRIN reading above 2 is a signal of price insensitive selling that’s reflective of either margin liquidation or risk manager forced hedge fund derisking that usually occurs at market bottoms.
- Intermediate-term overbought oversold indicator, consisting of the ratio of stocks above their 50 dma divided by stocks above their 150 dma, is an indicator of price momentum.
The accompanying chart shows the history of my bottom spotting model. Most buy signals are exhibiting good long entry points with positive risk-reward, but there are some instances when oversold markets have become more oversold. This model can be early with its buy signals. Traders using this model should use some risk management to control downside risk.
Useful, But Less Reliable Models
Finally, I offer two timing models that are useful, with strong, but not nearly 100% reliability.
The first is insider buying. A buy signal is triggered whenever the level of insider buying (blue line) converges or exceeds the level of insider selling (red line) from this chart of insider activity from Open Insider.
Insider activity is reported with a slight delay and this model will not spot the exact bottom. The accompanying chart shows an evaporation of insider sales while insider buying remained steady in the wake of the “Liberation Day” panic. As a reminder, excessive insider selling is not a reliable sell signal.
While insider buying is generally a useful buy signal during plain vanilla drawdowns as a buy signal, the signal is less reliable during periods of major market dislocations. The accompanying chart shows the history of insider activity starting in 2007, which was the market top prior to the GFC. There were several clusters of insider buying, and the market didn’t bottom out until March 2009, which was a period marked by off-the-chart readings of insider buying.
Lastly, I offer the S&P 500 Intermediate Term Breadth Momentum Oscillator (ITBM) as a timing model. The ITBM model is remarkable inasmuch as it produces actionable buy and sell signals.
One drawback of my S&P 500 Bottom Spotting Model tends to be early in its buy signals. One remedy is to wait for the indicator to recycle from extreme to neutral before buying at the risk of missing some of the move. I found reliable, but not 100% reliable, buy signals whenever the 14-day RSI of ITBM recycles from oversold to neutral. The accompanying chart shows the history of the buy signals. They grey lines show successful buy signals and the pink lines show failed signals. The history of this model is useful, but traders using this model need to employ some risk management.
Here is the history of ITBM sell signals. The success rate is strong, but not perfect.
In conclusion, instead of giving you a fish for the day, I am offering to teach you how to fish. My high conviction idea for 2026 and beyond consists of four U.S. equity market timing models with near 100% reliability and differing time horizons. In addition, investors can consider two timing models with strong success rates, but will need some degree of risk management.
