Factoring in the Crash vs Performance?

[marko]

So, if I'm considering a stock or fund, I look at the 3, 5, 10 and 15 year performance via M*.

When I look at the 10 year performance, I find myself saying "...it did 6.5% over ten years, but that includes the crash, so maybe it really does (has potential for) something like 8% because the crash was a (hopefully) once in a lifetime event..."

Yes, I know..."past performance doesn't guarantee future returns etc". And I realize that the stated performance IS the performance for that period.

I'm trying to assess a 'normal' (expected) performance by factoring in a unique negative event.

But is this a fair thing to do? Or has the past 5 years sort of cancelled it out by being a unique positive event?

 

[imoldernu]

Not likely to get a definitive answer.
While many will disparage any negativity, clear thinkers will always look at both sides of a story. If the market always followed the hope of investors, there would be no bears. Here's one view:

The Stock Market's Shaky Foundation: Chris Martenson | Peak Prosperity

 

[pb4uski]

I think a 25% bump to the 10 year historical total return probably brings the 10 year return into line with 10 year performance absent this hopefully once in a lifetime event. The 25% is essentially the 10 year total return of Vanguard Total Stock Admiral of 7.87% in relation to the 1926-2013 total return of 10%.

While my best guess is that time will bear that out, nonetheless for my own personal planning I continue to haircut the 1926-2013 return by about 1/3 to be conservative and to reflect a potential "new normal" (if there is one). Better safe than eating catfood.

 

[LOL!]

I'm chuckling a little bit here because if one is using passively-managed low-expense-ratio index funds, then one doesn't even have to look or compare anything. One knows they will get the return of the asset class that the index fund follows through thick and thin. That in and of itself is more peace of mind than paying off one's mortgage.

Full disclosure: I paid off my mortgage a while ago.

 

[Ready]

I always like to bring up "Reversion to the Mean" when looking at past performance of mutual funds. I used to think I should invest in funds that have beat the index for the past decade and made my investment choices accordingly. Now I understand that a fund that beat the odds over the past decade is more likely to underperform in the future due to reversion to the mean.

As crazy as it sounds, you're probably better off picking a fund that has underperformed the index and hoping it catches up in the next decade. Or, you could just keep it simple and buy index funds, and stop trying to guess which actively managed funds might be the future winners. History tells us that less than 25% of those funds ever beat the index in the long run anyway. If you had a 25% chance of winning a bet in Vegas, would you want to bet your retirement money on that?

 

[samclem]

. . . if one is using passively-managed low-expense-ratio index funds, then one doesn't even have to look or compare anything. One knows they will get the return of the asset class that the index fund follows through thick and thin.

Yes, and there are good figures for returns for those asset classes going back many years, and during all types of market/economic gyrations. That makes it a lot easier to plan.

 

[Spanky]

"I'm trying to assess a 'normal' (expected) performance by factoring in a unique negative event."
How you might approach this? In other words, what's your method?

 

[photoguy]

When I look at the 10 year performance, I find myself saying "...it did 6.5% over ten years, but that includes the crash, so maybe it really does (has potential for) something like 8% because the crash was a (hopefully) once in a lifetime event..."

But is this a fair thing to do? Or has the past 5 years sort of cancelled it out by being a unique positive event?

I don't think this is a fair way of estimating future returns. First, crashes happen all the time. The most recent meltdown might have been larger than most (-55%) but this is only a little bigger than dot com crash (-47%) and 73/74 was (-45%).

Second, the most common way folks look at future potential returns is by looking at current stock valuations and not performance in the past X years. Look at measures like schiller PE (aka PE10 or CAPE) and forecast 10 year returns. These measures suggest that next ten years will be below average but these are still not very predictive (although they are the best we have).

 

[Animorph]

I'm not sure too many diversified funds did great during 2008-2009, but during 2000-2003 funds that were heavily into value stocks actually did pretty well. So even though you might be looking at a sort of long 10 year span, it could be just one key decision or characteristic that resulted in a good result. And it probably won't repeat next time. Certainly value stocks didn't save anyone in the last crash.

On the other hand, estimating future returns your way is probably no worse than any other method.

 

[marko]

"I'm trying to assess a 'normal' (expected) performance by factoring in a unique negative event."
How you might approach this? In other words, what's your method?

Nothing scientific; just goosing the number a little based on market watching for the past 40 years. 6.5% might be an "8%".

The crash took a lot of wind out of the sails of stocks so I just add a percent or two as a compensation as long as the fund/stock is otherwise relatively solid.

Again, not sure how good an approach this might be.

 

[ziggy29]

In many cases, the longer term performance results can be even less indicative of future performance if the fund underwent a style change or a change in fund managers during that time. Just food for thought.

 

[karluk]

When I look at the 10 year performance, I find myself saying "...it did 6.5% over ten years, but that includes the crash, so maybe it really does (has potential for) something like 8% because the crash was a (hopefully) once in a lifetime event..."

When assessing the probability of future profits, I would either not take this sort of past performance into account at all, or I would try to incorporate into my thinking some of the other predictors of stock market performance that have shown some reliability in the past. In particular, look at PE10, which has been a frequent topic on this forum. Looking back at the end of the 2000-2003 bear market, I see that PE10 bottomed out at around 22. That means it was predicting below average returns for the next decade, and indeed you say your stock (or fund) did 6.5% compared to a historical average of 8%.

But PE10 is higher now than it was ten years ago. So to the extent that PE10 has some validity in predicting long term stock returns, you should actually be expecting less than 6.5% over the next decade.

The underlying problem with looking only at the last decade or even 15 years is that valuations were sky high during the entire period, and topped out at their all time peak in 2000. Unless you expect a permanent state of high valuations going forward, the good but not great returns we've seen in the last decade are just one way for the market to revert to the mean. So below average future returns may possibly persist for some time.

I personally am a buy and hold rebalancer, so my philosophy is to disregard past performance and PE10 and just stick with my asset allocation. YMMV

 

[audreyh1]

So, if I'm considering a stock or fund, I look at the 3, 5, 10 and 15 year performance via M*.

When I look at the 10 year performance, I find myself saying "...it did 6.5% over ten years, but that includes the crash, so maybe it really does (has potential for) something like 8% because the crash was a (hopefully) once in a lifetime event..."

Yes, I know..."past performance doesn't guarantee future returns etc". And I realize that the stated performance IS the performance for that period.

I'm trying to assess a 'normal' (expected) performance by factoring in a unique negative event.

But is this a fair thing to do? Or has the past 5 years sort of cancelled it out by being a unique positive event?

No, I don't think you can factor 2008 out. It's probably reasonable to expect such an event every 20 years at least. There was 1999-2002 as well.

 

[SVHoper]

... The 25% is essentially the 10 year total return of Vanguard Total Stock Admiral of 7.87% in relation to the 1926-2013 total return of 10%.

Hi pb4uski,

I wondered, could you tell me how to find the 1926-2013, return. I was looking for Total Stock Market return and Small Cap or even S&P500 but I think Morningstar only went back to 1970s

Thanks!

To the op, IMHO I think past returns are driven mostly by the periodic table of asset allocations e.g. was the fund growth, emerging, small, large

Then when you compare a couple of funds that purport to be say large, you sometimes find one gets a bit more juice by some percentage of small or international.

 

[gerntz]

So, if I'm considering a stock or fund, I look at the 3, 5, 10 and 15 year performance via M*.

When I look at the 10 year performance, I find myself saying "...it did 6.5% over ten years, but that includes the crash, so maybe it really does (has potential for) something like 8% because the crash was a (hopefully) once in a lifetime event..."

Yes, I know..."past performance doesn't guarantee future returns etc". And I realize that the stated performance IS the performance for that period.

I'm trying to assess a 'normal' (expected) performance by factoring in a unique negative event.

But is this a fair thing to do? Or has the past 5 years sort of cancelled it out by being a unique positive event?

The issue is how big a loss are you prepared to deal with. 25%? 50%? Equities have lost up to 80% a few times over the last 100-150 years. Net, divide loss percentage by 80% get your equities percentage target.

 

[CaliforniaMan]

I always like to bring up "Reversion to the Mean" when looking at past performance of mutual funds. I used to think I should invest in funds that have beat the index for the past decade and made my investment choices accordingly. Now I understand that a fund that beat the odds over the past decade is more likely to underperform in the future due to reversion to the mean.

As crazy as it sounds, you're probably better off picking a fund that has underperformed the index and hoping it catches up in the next decade. Or, you could just keep it simple and buy index funds, and stop trying to guess which actively managed funds might be the future winners. History tells us that less than 25% of those funds ever beat the index in the long run anyway. If you had a 25% chance of winning a bet in Vegas, would you want to bet your retirement money on that?

Often people view reversion or regression to the mean as a "causal" phenomenon which at least over the short run it is not. It simply means that a recent outlier is likely not as good a measure of the actual mean as a longer average of past measurements. It is not causal, in that the low value does not cause the subsequent higher values, it is simply so far below the mean as to likely be an outlier.

So if the market or an individual stock has a bad year, that bad year does not cause subsequent years to be better, only that subsequent years are likely to be better. This is true for any selected time span.

For example, suppose a ball player usually hits a 250, then for some unexplained reasons starts batting 200. If there is no obvious reason, injury, etc, this is likely an outlier. Would we most likely expect that next season he would bat 300 to make up the lost time? Obviously that is not what we would expect, it would be much more likely that he will regress toward his mean of 250.

Similarly, we cannot expect a mutual fund to "make up for lost time" just because it had a low outlier, but rather just get back to the mean rate of increase.

Now of course a mutual fund manager might try all kinds of tricks to get ahead next time by increasing risk, but that is something else again. If it were as simple as choosing funds or stocks that had declined more than usual then heck, we would all be rich.

This of course is the definition for a stationary time series. If we look at whole economies and longer time periods, times of wartime destruction etc. we are getting into a different subject. But I don't think this is what most of us here are talking about.

Bottom line and I think historical testing has shown, you cannot predict future performance based on past. Sorry.

 

[marko]

For example, suppose a ball player usually hits a 250, then for some unexplained reasons starts batting 200. If there is no obvious reason, injury, etc, this is likely an outlier. Would we most likely expect that next season he would bat 300 to make up the lost time? Obviously that is not what we would expect, it would be much more likely that he will regress toward his mean of 250.

That was the basis of the question.

Do we view the stock as a "250" player with an outlier skewing the data or a true "200" player because that's what the raw numbers show?

Your example says that the player is really a "250" player (with, obviously, no guarantee of future performance). When comparing players it would be safe to say: "he just had one bad year but he's really a 250 player".