He obviously consider 12.5 to be a decent choice. Very well, suppose it is 1993 and you are at 50/50 stocks/bond and the Dow is at 2,500. The Dow is now at 16,000, which is over a 640% increase, so you would have decreased your stock allocation by 640% / 12.5 = 51.3%. That is, to zero. (Or would you be shorting them?) In other words, following this rule will inevitably (one hopes inevitably, because one hopes the stock market inevitable rises) reduce your stock allocation to zero, and a good deal faster age-related "glide slopes" would.If the market goes up 50 percent, maybe I want to reduce my stock allocation by 4 percent. So there’s a 12.5 ratio between those two numbers.... Pick your number, pick your ratio.
So, he hasn't really stated the rule operationally. There are some implicit assumptions about what it means for the stock market to "go up." Just going up over twenty years doesn't count.
Which isn't surprising because he doesn't give any backtested results to show how his valuation-adjusted allocation compares with staying the course at a high stock allocation, or staying the course at a low stock allocation. I think he's saying that following his method will give you about the same results but make you feel better; or give you about the same results but avoid exceeding your risk tolerance. Or, maybe even, effectively give you a higher risk tolerance by letting you have a higher stock allocation.
It is definitely a form of market timing, but I don't necessarily have a knee-jerk reaction to that. It is definitely not staying the course, but increasingly I am thinking that staying the course is somewhat misleading--nobody does it, nobody can do it, the people who advocate staying the course constantly change the course they advocate staying; staying the course means "make only slow gradual changes, and take 5 years to make them," even though it's justified by pointing to what-if-you'd-stayed-the-course-since-1926.