How the models
actually work

Portfolio Navigator answers one question: is the broad market healthy enough to own? It doesn't forecast. It measures breadth — how many of an index's stocks are participating in the trend — and reacts to what it finds. Strong rallies are usually broad; major declines are usually preceded by participation narrowing underneath the index.

For each of five index ETFs — SPY, QQQ, DIA, MDY, and IWM — the model computes a proprietary breadth indicator from the stocks inside that index, using point-in-time membership so history isn't rewritten by today's survivors. Each index has its own independently optimized indicator. The portfolio holds the five in equal weight; any index whose signal turns bearish moves to safety (BIL or SGOV).

The model uses two thresholds, not one. The level that turns a signal on sits above the level that turns it off, which keeps the model from flipping on every wobble. The result: signal changes are infrequent by design — typically months to years apart — and reflect major shifts in market character, not short-term noise.

Momentum is one of the most extensively documented anomalies in financial markets: stocks that have performed well over the trailing several months tend to continue outperforming over the following months, while recent laggards tend to keep lagging. The effect was first formally documented by academics Narasimhan Jegadeesh and Sheridan Titman in 1993, and has since been replicated across nearly every major equity market and asset class studied.

The likely explanation is behavioral: investors are slow to fully react to new information, causing prices to drift in the direction of the news for longer than a purely efficient market would predict. Whatever the cause, the empirical pattern has persisted for decades.

Stock Navigator uses a momentum score as the primary ranking signal — it determines which stocks make the list and in what order. Fundamentals don't pick candidates; they grade the ones momentum has already surfaced.

Where the F-Score asks "is this company healthy," the M-Score asks a sharper question: "is this company's accounting hiding something." Developed by Messod Beneish in 1999, it's a forensic accounting model built from the financial statement patterns observed in companies that were later caught manipulating earnings.

The model combines eight financial ratios — covering receivables growth, gross margin trends, asset quality, revenue growth, depreciation rates, expense ratios, accruals, and leverage — into a single weighted score. Each ratio captures a different way that aggressive or fraudulent accounting tends to show up in the numbers before it's publicly known.

A resulting score above a calibrated threshold flags a company as statistically likely to be manipulating earnings; below it, the accounting looks clean by historical pattern-matching standards. In Stock Navigator's rankings, this acts purely as a risk screen — it doesn't rank stocks higher, it flags ones worth a second look regardless of how attractive the momentum or F-Score numbers are.