“Essentially, all models are wrong, but some are useful.”

Those are the immortal words of statistician George E.P. Box, which underlie every bit of serious analysis that seeks to approximate real world processes and outcomes with math. Including our own.

What keeps models useful is to recognize when and how they’re wrong. For example, many years ago, we recognized that the projections of our dividend futures-based model of how stock prices work could be skewed off target because it incorporates historic stock prices as the base reference points from which it projects future stock prices. If those historic stock prices were to include a period of abnormally high volatility, the accuracy of the model’s projections would be negatively impacted – skewed off by the echo of that past volatility.

With past volatility events proving to be poor predictors of future stock prices, we experimented with different methods to compensate for the echo effect in our model, eventually settling on a very simple method of “connecting the dots” to bridge across periods where we recognized that the accuracy of our model’s forecasts would be affected by the echo of past volatility. We show those adjustments as a red-zone on the forecasting charts we produce.

In our latest redzone forecast, which covers the period from 7 November 2018 through 7 December 2018, we assumed that investors would remain focused on the future expectations associated with 2019-Q1. In the Thanksgiving holiday-shortened third full week of November 2018, something changed for the S&P 500 (Index: SPX) to break that assumption, and what had been our stellar track record for previous redzone forecasts….

Instead of falling within our redzone forecast, the trajectory of the S&P 500 has dropped below it, seeming to follow the unadjusted trajectory associated with investors focusing their forward-looking attention on 2019-Q1. That’s almost certainly a coincidence, which we’ll be able to confirm in the next several weeks.