The bulls are making a whopper of a claim: that stock prices are reasonable because rates will stay incredibly low forever. Discard that fantasy, and their rationale collapses. We’re in for a replay of what always happens when bond yields become super-slender. Rates will rise, and equities will languish so badly that on big caps, you’ll be lucky to pocket piddling returns in the years ahead.

Warren Buffett often talks about how interest rates exert a gravitational pull on stocks. What matters to investors isn’t the so much the total return they get from equities; it’s the margin on what, say, big-cap stocks offer over the relative safety of holding Treasuries. The less the 10-year is yielding, the higher the price folks and money managers are willing to pay for an S&P 500 index fund, or a blue-chip portfolio that more or less tracks the index.

The bulls who pushed the S&P 500 to a record close of 3945 on Feb. 12 swear that today’s incredibly low rates justify valuations that look overly stretched, if not hovering in bubble territory. Just discount back future earnings incorporating these super-low yields, they argue. You’ll find that the dollars you’re paying for each dollar in profits, though it looks high, makes sense, and even leaves air space for the S&P to push higher.

The optimists’ manifesto contains a faulty assumption: That by far the lowest rates in at least 60 years are here to stay. The idea is that stocks can stay this expensive because they’ll remain so much better than bonds. But that’s only true if you posit that “real” yields, adjusted for inflation, remain in today’s negative territory pretty much forever. It’s never happened before and won’t happen going forward. Put simply, earnings should be discounted back not depending on today’s rates, but projected rates going forward. Wall Street would rather believe that a 10-year Treasury paying less than inflation is the new normal. It’s not.

Right now, the P/E multiple for the S&P 500, based on the past four quarters of EPS through Q4 of 2020, stands at 40. That’s deceivingly high, reflecting the severe drop in mid-2020 caused by the COVID crisis. The Congressional Budget Office projects that U.S. GDP will regain its pre-pandemic reading in mid-2021. So let’s predict that when national income returns to that norm, so do profits, which stood at $139.47 per share at the close of 2019.

Using that “normalized” number, the S&P is selling at a P/E of 28.3 (3945 divided by $139.47). Hence, the S&P earnings yield, the dollars you can expect to pocket in dividends and price appreciation, is 3.53%. Add 2% inflation, and you can expect to garner 5.53% a year, so long as the P/E holds constant at around 28. That return comes in two buckets: the dividend yield of 1.58%, and earnings growth, courtesy of plowed-back profits, of nearly 4%.

That number doesn’t sound great. But it’s better than it looks, because the yield on the 10-year Treasury is just 1.09%. We’ll do the calculations adjusted for inflation, since it’s the “real” numbers that set P/E multiples. Inflation is now running at 1.65%, meaning the long bond yield is a negative 0.56%. No wonder bonds appear so unattractive. They don’t even keep you whole with prices at the mall and supermarket. But what makes bonds look bad makes stocks look better. Stocks are beating bonds, adjusted for inflation, by 4%. That is, the 3.53% real expected return exceeds the long bond yield that’s 0.56% underwater by just over 4%. Based on history, that 4% is a pretty good margin.

But you only keep logging those 3.53% (real) yearly returns if bonds stay incredibly cheap and the P/E remains at a lofty 28. It’s hard if not impossible to justify today’s valuations unless you assume that today’s negative real rates are permanent. If that’s the case, then the discount rate applied to future profits will remain extraordinarily low, and today’s stock prices look just fine.

But rates won’t remain at today’s bargain levels. That’s what both history and the Congressional Budget Office’s most recent forecast tell us. It won’t happen for a while. But the CBO projects a long bond yield of around 2.1% in 2026, rising to roughly 3.0% in 2028, and into the mid-3s by 2031.

Let’s assume investors will want to keep that 4% edge over bonds seven years from now, in early 2028. The CBO predicts inflation will be running at 2.1%. Hence, the “real” yield will jump from minus 0.56% to plus 0.9%. Add that number to the 4% margin, and you get 4.9%. That 4.9% is what stocks can be expected to pay you in the future, adjusted for inflation. Hence, the earnings yield would need to go from today’s 3.53% to 4.9%. And the real earnings yield can only be 4.9% if the P/E drops to 20.4. Then you’d be getting $4.90 (plus inflation of, say, $2.00) for every $100 you invest in the S&P 500.

We’ll assume you don’t sell any stock over those seven years. During that period, you’d get an annual average of 1.8% from dividends. But the almost eight-point drop in multiple from 28.3 to 20.4 would leave you with a minuscule return. The value of your portfolio would be almost exactly where it is today. Your total return would be that 1.8% from dividends. Betting on the S&P means losing to inflation. The best time to buy stocks is when rates are extremely high and P/Es are in the dumps.

Today’s scenario is the exact opposite. So, as they say in the COVID era, follow the science. And forget the fantasy.

More must-read finance coverage from Fortune:

• Why Biden’s climate policies could be good for the S&P 500
• With the Mustang Mach-E, Ford brings America’s consummate muscle car into the plug-in era—and sparks fly
• Tax season 2021: Deadlines, brackets, refunds, taxable income, and everything else you need to know about filing your 2020 taxes
• Reformed Bitcoin miner: Elon Musk’s $1.5 billion bet is “crazy”
• TikTok to the moon? The app is outgrowing its Gen Z stereotype