When it comes to predicting stock returns, about the best you can do is present “a range of likely outcomes.”

After all, “the future” — and everything to do with the future, other than the fact that you don’t have one beyond your personal sell-by date — is “inherently uncertain.”

The bits in quotes are from a new note by Goldman’s David Kostin, who suggested investors prepare for lower annualized stock returns over the next decade. Kostin uses five variables for the model: The CAPE, the frequency of recessions, benchmark US yields, trailing ROE for the index and, notably, the market-cap ratio of the most valuable company to the 75th percentile.

Obviously, the CAPE’s high as hell right now, if readers will forgive the colloquial cadence. Specifically, it’s 38x.

As the figure shows, that’s basically — i.e., excluding the dot-com bubble — a record high.

Valuations are famously unreliable as a market-timing device, but generally speaking, buying at elevated starting valuations is a good way to “lock in” lower long-term returns. It’s not a coincidence that starting valuation is “the most important variable” for Goldman’s model forecast.

Hopefully, readers don’t need an explainer as to why the macro environment, ROE and 10-year US yields (three of the four other variables in Goldman’s model) matter for long run stock returns, but as Kostin wrote, the market concentration point deserves special attention currently. It’s “particularly important today because the US equity market is near its highest level of concentration in 100 years,” as he put it.

The figure gives you a sense of just how extreme current conditions really are on the market concentration front.

As Kostin explained in a well-put passage, “index performance in a high concentration environment will reflect a less diversified set of risks and will likely have greater realized volatility compared with returns in a less concentrated and more diversified market.”

That’s really all you need to know. That and the valuation point mentioned above. Kostin tied them together. “Although greater volatility does not necessarily imply an increased likelihood of downside relative to upside return, valuation should reflect a discount for higher volatility and less diversification during periods of extreme market concentration,” he wrote.

Bottom line: Goldman’s model suggests the S&P will produce an annualized nominal total return of just 3% during the next 10 years.

As the figure shows, that outcome, if realized, would be very poor by historical standards. By “very poor” I mean 7th percentile.

How much of a drag is extreme market concentration in the model? Kostin quantified that: Excluding the concentration variable, Goldman’s forecast would be four full percentage points higher, and instead of the -1% to 7% bands shown above, the range would be 3% to 11%.