Well, I was close to 100% equities until about 1 year before retiring (very young), when I started to transition, so I never worried about what was the right AA during accumulation.
In terms of withdrawals - my straight % remaining portfolio withdrawal method results in me taking out large $$ amounts when the portfolio is high which appeals to me intuitively, since the higher portfolio is more likely to see a correction in the near future after several years of a runup.
By dropping the withdrawal percent using a valuation based VPW, you are actually taking less $$ out when the portfolio is overvalued, leaving more in the portfolio to suffer a potential haircut. That runs against my philosophy so far.
Looking at income changes during bad sequences, the real income from traditional VPW seems to drop almost as much as the models I have run using my %remaining portfolio method.
The main advantage I see of VPW is leaving less money at the end.
I understand where you're coming from. It's also exactly why I won't use classic VPW and instead use one that is based on valuations. Either version of VPW leaves nothing in the end, that's true. But classic VPW tends to have withdrawals that too often increase way too rapidly towards the end of retirement withdrawing money that I would have had sooner, rather than later. And a valuation based method attached to PMT/VPW helps solve that.
The problem with comparing a fixed % withdrawal to something like VPW or a valuations based version of VPW is deciding what % to use for the fixed withdrawal since you can choose anything you want. And if you can choose anything you want, it starts to become an apples-to-oranges comparison as I can always find a % that compares favorably one way or another with a different withdrawal method, when testing past results. I know because I spent a long time playing around with that method a couple of years ago and I found that I could backtest based on my portfolio and choose my own criteria for success. For example, I could choose a very low % that was much lower than the average growth of my portfolio and I would see the withdrawals increasing over time. Or I could choose something too high and the withdrawals would steadily decrease over time as my withdrawals overran my returns, eventually dropping below what I could survive on. Or I could choose something that optimized for the worse start year in history for my portfolio (somewhere between 1966 and 1969) where it would start off high, then drop for a number of years, then recover. You could even try and look at quasi-apples-to-apples and choose the same starting percentage for the first year that is used for classic VPW for the first year. In that case, VPW will clearly win because the percentage withdrawn increases each year from year 1. So whether the real income drops as much between classic VPW and a fixed percentage withdrawal completely depends on the % actually chosen. And as always, I don't have a good way to know if the percentage I choose on day one will overrun or underrun my returns in the future and, unfortunately, I don't have so big a buffer that I could safely choose a super small percentage to all but guarantee success..
Everybody's needs and thought processes are different but I eventually walked away from that method, not because it left money on the table at the end, but rather in doing so, I had less to spend throughout my retirement. I learned about VPW and how it was basically a reinvention of something called "the actuarial method", often used by financial planners. But I really didn't like the withdrawal trajectories I was seeing, but about the same time several people over on BH started to bring up smoothing methods, especially valuation based methods. To me, it solved my major concerns: It didn't sacrifice early withdrawals for later ones and it regulated the year-on-year withdrawals, made sense intuitively, and had an absolute withdrawal that was based on a % of the portfolio which also means if my portfolio does well, so do my withdrawals.
I think of it this way: If I have $1M in my portfolio and I would normally take out 4% with VPW using long term returns, that would be a withdrawal of $40K, leaving $960K in the portfolio. If now there is a 20% drop in the market before my next withdrawal, I now have $768K before my next withdrawal.
If instead I'm using valuations which tell me that my withdrawal should be 3%, then I'm now withdrawing $30K, leaving $970K in the portfolio. If now there is a 20% drop in the market before my next withdrawal, I now have $776K before my next withdrawal.
In this simple example, I am going into my next withdrawal with more money in the portfolio by using valuations that adjust the withdrawal downward in anticipation of lower future returns, thus preserving the portfolio and possibly allowing for a higher withdrawal next time if the market recovers.
Now this is a simple example, but using valuations does the something similar. If valuations are low, you have no idea what sort of trajectory year on year your returns will take, but you do have an idea whether over, say, the next 5-10 years or so they might be low-ish, middle-ish, or high-ish and the withdrawal percentage is adjusted accordingly.
As an electrical engineer, I find using a valuation based version of VPW is very much like a feedback loop and why it tends to keep the withdrawals from running away extremely high or extremely low. And the thing I intuitively like about it is that extreme accuracy isn't required for a valuation in order to smooth out the withdrawals reasonably - at least historically and of course the future might not be like that, blah blah blah.