In four years—November 2020—the stock market is likely to be no higher than it is today.

No, this is not another forecast based on the presidential election-year cycle.

Instead, it comes from a market-timing model that is one of the select few of the many hundreds I have monitored over the past four decades that has actually beaten the stock market. It does this by increasing and decreasing market exposure at optimal times. Most market-timing models, by contrast, are no better than a coin flip.

This market-timing system is based on a single number that appears each week in the Value Line Investment Survey, a newsletter published by Value Line, an investment research firm based in New York. Its analysts closely monitor some 1,700 stocks, and the number—known as VLMAP, for Value Line’s Median Appreciation Potential—represents the median of the estimates made by those analysts of how much those stocks will gain over the next three to five years.

Market timers use the VLMAP to project where the stock market will be in four years, the midpoint of the analysts’ three-to-five-year horizon.

The VLMAP is currently at one of its lowest levels in years—as low, in fact, as it stood at the top of the bull market in October 2007, right before the worst bear market in the U.S. since the Great Depression.

Value Line itself does not recommend using the VLMAP as a market-timing tool, even though the firm is not against anyone using it or any of the other data it produces. As far as I can tell, the VLMAP-based market-timing model originates in work done in the 1970s and 1980s by Daniel Seiver, a member of the economics faculty at California Polytechnic State University and editor of an investment advisory service called the PAD System Report.

It’s worth noting that a casual reader of the Value Line Investment Survey wouldn’t immediately become alarmed upon viewing the latest VLMAP reading. It stands at 40%, which over four years is the equivalent to an annualized return of 8.8%.

The reason Seiver considers the current VLMAP reading to be worrisome is that, in the past, it has tended to be too optimistic. In fact, the VLMAP’s current level puts it in the 10% of lowest readings since the early 1960s. Its highest level came in late 1974 at 255, right as the 1972-74 bear market came to an end. It got as high as 185 in March 2009, as the 2007-09 bear market was ending.

In an academic study published in 2013 in the Journal of Wealth Management, Seiver and a colleague documented the market-beating potential of VLMAP. The particular market-timing system they proposed calls for investors to reduce their equity exposure whenever the VLMAP drops to within the lowest third of all historical readings. That would mean anything below 55%. Currently, Seiver is recommending that subscribers to his newsletter allocate at least 50% of their stock portfolios to cash.

To be sure, four years ago the VLMAP-based timing model was also cautious, when—like today—the VLMAP stood at 40. And, yet the stock market has performed very well since then, producing a 6.4% annualized total return, as judged by the dividend-adjusted Wilshire 5000 index.

It may be that we just haven’t given the model enough time to be proven right, however. In an interview, Seiver said it’s possible that the extraordinary monetary stimulus of the past few years has supported the stock market and thereby postponed—but not forever—the ultimate day of reckoning. In this regard, he noted that the VLMAP-based forecasting model actually has a better track record at the seven-year horizon than it does at four years.

Seiver therefore thinks it would be a mistake for the bulls to take too much comfort from the stock market doing so much better over the past four years than had been forecast by the VLMAP-based model. Seiver is betting that the VLMAP will have the last laugh.