retired at 48:

You lastly stated: "With any luck (sarcasm alert), you will have raised your own risk to a 100% chance of running out of money before death."  If one stops "starting over" at the first down year, his chance of running out of money are the same as anyone retiring that year with same numbers, 5% risk.  It would never get to 100%.  Am I missing something?

My point (which you agreed with) was that the person who doesn't re-retire in the second year doesn't have as much risk as the person who starts in year two.  Thus, they can't both have a risk of 5%.  Here is an extreme example.

Number 1 retires at 64 with $900,000 and withdraws $36,000 at the beginning of the year to fund his expenses.  The market has a very good year, and at the end of it he has $10,000,000.  If he doesn't re-retire, he takes $37,080 for his second year withdrawal.  His buddy, who just retired with the same $10,000,000, takes out $400,000 at the beginning of the year. That year, the market drops by 95%.  Number 1 has (10,000,000 - 37080) * 0.05 = $498,146 left at the end of year two.  His buddy has (10,000,000 - 400,000) * 0.05 = $480,000 left.  On January 1, number 1 will be taking out $38,192.40 while his buddy will withdraw $412,000, leaving only $68,000 in his account.  Unless the market has another huge run up, the second person will have nothing left at the beginning of the next year when his withdrawal will be $424,360.   By retiring just before a 95% drop in the market, his buddy's risk of running out of money is 100%. 

The assumption of the 4% SWR is that you take the 4% (increased for inflation) and spend it.  If you are taking 4% and saving 2%, you aren't really using a 4% withdrawal rate. Number 2 has been able to take his family on several round the world cruises in the penthouse of the Queen Elizabeth II with his $812,000 in spending over the first two years of his retirement, but he will still be broke after the third year.

 To repeat, the 5% risk does not apply to each person separately.  The 5% risk says that if you take 20 different starting years, those who start in the worst year go broke and those who start in the other 19 years don't (only 1 out of 20 or 5% of the years lead to poverty; 95% of the starting years lead to comfort).  When you retire, you don't know if you are in the one bad year or one of the nineteen good years.  Accordingly, you have a 5% chance of starting in that one bad year.  If you keep re-retiring until you hit that one bad year, you will be part of the group that goes broke (guaranteed for everyone who has the bad luck to retire in the one bad year out of the twenty). Your risk is thus 100% of going broke since you kept re-retiring until you finally got to the one bad year out of twenty.

Al