Regressing to the mean
In a comment you say,
if the market crashes, doesn't "regress to the mean" mean that I should still expect 7% over the long run? That being the case, wouldn't I benefit from intentionally unbalancing my portfolio and going all in on equities? I can can still rebalance using new savings.
No. Regress to the mean just tells you that the future rate is likely to average 7%. The past rate and the future rate are entirely unconnected. Consider a series:
7%, -21%, 7%, 7%, 7%, 7%, 7%, 7%, 7%, 7%
The running average is
7%, -7%, -2.33%, 0%, 1.4%, 2.33%, 3%, 3.5%, 3.89%, 4.2%
That running average is (slowly) regressing to the long term mean without ever a member of the series being above 7%.
Real vs. market value
Real markets actually go farther than this though. Real value may be increasing by 7% per year, but prices may move differently. Then market prices may revert to the real value. This happened to the S&P 500 in 2000-2002. Then the market started climbing again in 2003. In your system, you would have bought into the falling markets of 2001 and 2002. And you would have missed the positive bond returns in those years. That's about a -25% annual shift in returns on that portion of your portfolio. Since that's a third of your portfolio, you'd have lost 8% more than with the balanced strategy each of those two years.
Note that in that case, the market was in an over-valued bubble. The bubble spent three years popping and overshot the actual value. So 2003 was a good year for stocks. But the three year return was still -11%. In retrospect, investors should have gone all in on bonds before 2000 and switched back to stocks for 2003. But no one knew that in 2000. People in the know actually started backing off in 1998 rather than 2000 and missed out on the tail end of the bubble.
The rebalancing strategy automatically helps with your regression to the mean. It sells expensive bonds and buys cheaper stocks on average. Occasionally it sells modest priced bonds and buys over-priced stocks. But rarely enough that it is a better strategy overall.
Incidentally, I would consider a 33% share high for bonds. 30% is better. And that shouldn't increase as you age (less than 30% bonds may be practical when you are young enough). Once you get close to retirement (five to ten years), start converting some of your savings to cash equivalents. The cash equivalents are guaranteed not to lose value (but might not gain much). This gives you predictable returns for your immediate expenses. Once retired, try to keep about five years of expenses in cash equivalents. Then you don't have to worry about short term market fluctuations. Spend down your buffer until the market catches back up.
It's true that bonds are less volatile than stocks, but they can still have bad years. A 70%/30% mix of stocks/bonds is safer than either alone and gives almost as good of a return as stocks alone. Adding more bonds actually increases your risk unless you carefully balance them with the right stocks. And if you're doing that, you don't need simplistic rules like a 70%/30% balance.