Inside the MB Edge Model: 100% Invested or 100% T-Bills

One signal, two positions

The MB Edge model, launched for individual investors in December 2025, holds exactly one of two positions: fully invested in the S&P 500 (via broad, low-cost ETFs such as SPY or VOO) or fully in Treasury bills. A full round trip — out and back in — has historically occurred about once every 1.25 years.

In the hypothetical backtest from 1957 through early 2026, the model was invested in the S&P 500 about 80.9% of the time and in Treasuries about 19.1% of the time. There are no intraday alerts, no watchlists, and no discretionary overrides — by design. The service exists for investors with retirement-scale horizons who do not want to sift the daily flow of headlines, social-media posts, and economic data.

The 8% rule — an honest answer to a hard problem

Berg's research reaches an asymmetric conclusion. Market bottoms can be modeled with precision: his team has modeled every S&P 500 bottom following declines of at least 8% since 1957, building 2,049 distinct models from thousands of data points, and buy signals typically arrive within days of major lows. Market peaks are another matter. Most data-based sell signals, he says plainly, either “arrive too early in an ongoing bull market, too late after a substantial decline has already occurred, or simply misfire.”

So the model does not pretend to call tops. It stays 100% long until either a model-based sell signal fires or the S&P 500 falls approximately 8% from its latest closing high — whichever comes first. Then it stands aside in Treasury bills until a new buy signal appears. The 8% exit gives up the first slice of every major decline in exchange for sidestepping the rest; the buy-signal machinery handles the harder job of getting back in near the low.

How the discipline played out in past bear markets

All figures below are hypothetical and backtested — the model did not exist in these years, and no client traded it. With that stated plainly, the mechanics are easiest to see in the three worst environments since 1957:

• 1969–1970. After the market peaked on November 29, 1968, the model sold on February 24, 1969, down −9.60% from the high. It stayed in Treasury bills until May 22, 1970 — while the S&P 500 fell a further −29.28%.
• 1973–1974. The model moved to Treasury bills on March 21, 1973, after a decline of −8.11%, and remained there until October 4, 1974 — sidestepping a further −42.78% decline.
• 2000–2003. The model exited on April 14, 2000, down −11.19%, in a bear market that would eventually take the S&P 500 down −49.15% in total. Notably, it did not simply hibernate: four buy signals fired during that bear market (April 5, 2001; September 19, 2001; July 22, 2002; August 6, 2002), producing hypothetical gains of +3.74%, +6.83%, +1.87%, and +2.28% before the model's exit rule triggered again.

What the model is not

MB Edge is deliberately not a trading service. Berg's institutional research publishes more than three times per week, covers many asset classes, and includes long, short, and leveraged recommendations; its long-only equity model portfolio has produced hypothetical price-only returns of 538.06% since January 1, 2016, versus 228.59% for the S&P 500 over the same period. Berg has so far declined to build an intermediate product for self-directed traders, reasoning that he cannot yet differentiate one clearly enough in scope, depth, and frequency without cannibalizing the institutional service.

It is also not personalized investment advice. MB Edge publishes the state of a long term hypothetical model and the research behind it; what any individual should do depends on circumstances the model cannot know.

Who it's for

Berg's stated belief is that individual investors can outperform the market — and stay calm — by exiting equities early in major bear markets and re-entering near the subsequent lows. The point of the 8% rule is not to maximize every rally. It is to make the downside survivable enough, financially and emotionally, that the investor is still there — and still solvent — to act on the next buy signal.

The full signal history, methodology, and disclaimers are documented across this site; the sample reports are the best place to see the research voice firsthand.

Frequently asked questions