Merrill Lynch just presented its 26th annual “Car Wars” study at the Detroit-based Automotive Press Association. Their analysis tries to predict which auto makers will fare best in coming years based on how much of its product line will be updated. Overall, the auto industry will be launching more new or fully redesigned models over the next four years than at any other time in the past two decades, with 2019 and 2020 extremely busy.

Over the past two decades product launches have averaged 38 per year. The coming model year and 2018 will be about average, with 44 and 49, respectively. But based on Merrill Lynch’s research auto makers will launch an incredible 70 new or redesigned models in 2019 and 68 in 2020.

The risk in this strategy: Merrill Lynch forecasts that U.S. auto sales will surge to a record 20 million units in 2018, then begin a cyclical decline. All of these new cars, crossovers, and trucks (including redesigns for all three major full-size pickups) could be launched into a declining market. Of course, if manufacturers realize soon enough that some programs will not pay off they could cancel them. Only the 2017s are virtually set in stone at this point.

Not all auto makers will be equally industrious. GM, Ford, and Honda will update an especially high percentage of their lines from 2017 to 2020. In contrast, the Koreans will be slowing down a little, and will be focusing far too heavily on cars (where GM, Ford, and FCA will be doing relatively little) rather than where the growth has been, in trucks and crossovers. Merrill Lynch sees this as a cause for concern. They also note that Nissan’s strategy seems “confused.”

Of perhaps greatest interest to anyone who’ll be buying a new car in the next few years: the report includes their findings of which car lines will be updated when. (Sorry about the image quality, I photographed a paper report.)

Any surprises? Anything you’re especially looking forward to? You can comment at the bottom.

But are these predictions accurate? Those made in the same report two years ago have turned out to be overly optimistic, with many launches actually happening a year or two later than predicted. For example, none of the predictions made two years ago for Toyota for the 2017 and 2018 model years remain in the new report. This probably isn’t entirely or even mostly the analysts’ fault. Programs get delayed and even canceled all the time.

My prediction based on the outcome of these past predictions: that huge surge for 2019 and 2020 won’t happen, at least not in 2019 and 2020. Many of the launches currently predicted for 2019 and 2020 will instead happen in 2021 and 2022, if ever.