When you say 75%-80% is that after or before taxes?

After taxes.

@akratic – That’s the biggest problem with “being overly technical with a simplified model”. Here you assumed constant income and constant returns; something which is unlikely to show up in the real world. Changes can really throw the model off.

I’d really hate for anyone to make blanket assumptions about the percentages I mention. I remember reading in Numerical Recipes (book about computational math) in terms of whether to normalize your statistical set with N-2, N-1, or N. What they said, and I paraphrase (and fully agree) that if the choice makes a difference in the statistical conclusion you’re trying to make, you’re probably on thin ice.

This is also why specific number questions and answers (without considering the fuzzy factor) concern me. Anyone wishing to pursue this should be able to navigate the land for this (or hire a CFP to do it).

To me, there’s little difference between 75 and 80% (my savings rate has been anywhere between 50+ and 90%) If this difference matters, we’re splitting hairs. There’s also little difference between 5 years and 6 years. It’s a little bit like becoming a millionaire. I was briefly a millionaire in my native currency in 2007, not in 2008 and 2009, but then again in 2010. Expect to cross the mathematical point several times. What matters is the probability of staying at the right side of it.

I think the worst possible conclusion one can draw from my statements is that if I save exactly XX% then I can quit after exactly N years. Yes, this can be shown mathematically, but the assumptions never hold exactly and therefore the conclusions don’t either.

Makes sense?

Maybe I should be more careful in my statements whenever I use a number. To me the dangers of “math” is intuitive; probably that whole computational physics background.

Yeah, I don’t expect my projections to come exactly true, because you’re right, there’s a lot of unknowns and fuzziness.

Maybe the reason I react so strongly to fuzzy ERE math is that I’m currently fighting tooth and nail for each percentage point of savings. You mentioned saving anywhere from 50%-90% but here’s my last few months (81%, 81%, 80%, 84%, 77%, 74%, 70%). What I’m doing is fighting hard to drive my expenses down — for each percentage point — precisely because it brings ERE that much closer. For me, 75% vs 80% is not a wash: it took a fair bit of effort, and it should pay off with a whole year less 9-5ing, give or take quite a large fudge factor.

I don’t want you to have to qualify every number you use… but I also don’t think anyone is hitting FI after five years of 75% savings (unless they are very skilled at investing). Of course, your big picture ideas continue to stand regardless of my nitpicking, and it’s these big picture ideas that make your site so great!

PS: If you want to give Wall Street a shot, let me know. I like the way you think, and I could probably make it happen (if not today, in a few months or a year)

@akratic – How about 5.58 times 77.1%? 😎

Maybe if I had said slightly more than 75% or “three quarters”, or about 5 years. Frankly, when I say 75% it is not meant in the precise sense. It’s sufficiently close to a “round number” (3/4) that it implies a more limited precision.

My guess is that you’re fighting your percentages on the margin: you’re out of slack given your current living arrangements. My 50-90% represents 5 different living arrangements and 3 different salaries. Rent and income makes the biggest difference.

Now onto the meat of the matter.

From a theoretical/experimental perspective, there are two challenges. I’m assuming you understand the difference between precision and accuracy. If not, then that’s a very important point. Accuracy reflects how close the model comes to the real world. If our model is a one-foot ruler, and accurate ruler would be 1.0000000 ft long. An inaccurate ruler may be 1.4756344 ft long. Precision refers to how well the ruler measures. Precision can be found statistically. If the ruler has 1/64″ graduations, it is a precise ruler. If it has 1″ graduations, it’s not very precise. So,

A 1.05 ft ruler with 1/2″ grads is somewhat accurate but not precise relative to a 1.3 ft ruler with 1/32″ grads which is very inaccurate but also very precise.

Note how instruments are called precision instruments, not accuracy instruments 😉 … that’s a good way to remember the difference.

With statistics, you can achieve very high precisions (goes as 1/sqrt(N), where N is the number of measurements…also follows that going from 1 to 10 measurements gives you a lot more precision than going from 10 to 20). You can never achieve a very high accuracy through lots of sampling. This is strictly model/method dependent.

Now, there are two inputs to understanding. There’s a sampling of data points in the real world (experimental). But there’s also a sampling of models (theoretical). The combination of these two yields the understanding. In general, the simpler the model, the easier it is to understand and do math with. Complex models require simulations or brain power because they can’t be handled analytically.

There are four ways to deal with this.

Mathematically: (Few points, analytical/linear equations—what you’re doing)
Statistically: (Many points, analytical/linear equations—Monte Carlo)
Computationally: (Few points, nonlinear equations—simulations)
Not possible yet: (Many points, nonlinear equations—the real world, the stock market)

What I did in the book to answer your question was to use the mathematical approach with a parameter range of different SWRs and plot them. I bet some will complain that they can’t tell if the graph says 5 years or 6 years. That was intentional, however. I don’t want anyone to use this approach believing it’s in any way precise.
It is followed by a section expounding on that problem.

As you say, a brilliant… or lets just say lucky investor could have retired in 3 years, starting a savings account in 2006, learning all about investments and then dumping it all in an index fund in, oh say, March 2009. One could argue that’s cheating/unreliable. It certainly is, but so are the equation projections.

The best method in my opinion is to use cash flow. Invest as you would in FIRE. Do a dry run. This will make it much easier to gauge the impact. However, if we do cash flow, I’d argue that SWR is no longer completely relevant. Dividend investing has different return characteristics than total return. Heck, you could even be in real estate (I prefer REITs) that would be even stranger. It’s also country specific.

In the future, I will definitely say “more than three quarters of one’s income for around five years”. In this sense my numbers are accurate, but not precise.

—

“I probably still don’t get your point, because I’d rather be trying to quantify the model and sometimes wrong… than refuse to quantify.”

The risk is that quantification instills a false (inaccurate) sense of confidence. Example: http://en.wikipedia.org/wiki/Long-Term_Capital_Management … here they had models where the real world was inaccurately but very precisely approximated (to the tune of billions of dollars) by Gaussians. As it turned out (to the loss of billions) the Gaussian wasn’t accurate and since the precision had been leveraged into huge bets, the company had to be bailed out.

It’s like taking your 1.23 ft ruler and having your geniuses work on increasing precision from 1/32 to 1/64 and making bets that now you can really measure the real world because you just doubled the precision. Then it turns out you were 0.23 ft off.

I hope this wasn’t too trite.

@akratic – Okay, 75% is not precise. 5 years is somewhat precise. The model used is not accurate.

Also, an important point which you mentioned in the beginning. Even a small side income or work from time to time can significantly alter the picture, as can a market boom or a market crash, or getting hit by a bus, etc.

PS: This stuff is in my bones. I’m probably assuming that this is the case for everybody but for practical purposes accuracy, precision, uncertainty, etc. are probably not universally ingrained. I worked with this stuff for many years so it’s become second-nature. It’s always hard to know what others know/understand. Like, when I used to explain my work, my first question was always “Do you know what a proton is?” That’s why I was concerned about sounding trite.

PPS: Yeah, that’s because your “needs” are around 20% and your wants are 80% and you successfully cut your wants. The needs require macrochanges.

NB: I think this is all covered in Chapter 7 😉

@OurTakeOnFreedom and Jacob: I believe travel before retiring is a good thing, if only to find out whether you like it enough to budget for it and work an extra year or two. As long as you’re still saving, you might want to choose cheap destinations.

We traveled extensively before the kiddos.

Now, we’re too tired to travel.

We will be travelling when I decide to fold my local business. Right now, it’s too much fun. I’ll reevaluate in 5 years or so. I know geo arbitrage is in my future. I can almost taste it.

That’s one of my goals and why i keep working even though i’m fi now at 2.8 swr until 85. i dont just want to go tour all of germany in 2 weeks, i want to go for 6 months, maybe a year; as just like your native country; when you live there; you’ll find all these places to go that only the locals know of and i have wanted to go to novaya zemlya since i first learned russian and you’re not going there unless you plan on staying at least a week.

I love travel, but I can’t stand the costs. On the other hand, trying to skimp too much on holiday also annoys me. So I simply travel a lot less.

It’s interesting that for my generation (Y), travel has become practically the norm – people think you’re an oddball if you don’t do it at least yearly.

@Seattle Sam – See About Me. Chicago also happens to be a lot less expensive than the bay area.