Shifting to higher stock allocation during crashes

[jcw]

Hello,

Wondering if anyone has backtested a 3-fund portfolio in which you were to go 100% stock on any crash (> 20%). Seems in the last 15-20 years, the market has been more volatile than ever with two 40% drops in the last 12 years. Obviously, we would get some benefit in the standard buy/hold/rebalance but seems like we can capture additional value by taking advantage of the volatility. As the market came back up, you would slowly shift back to your normal stock-bond mix. Some might see 100% stock as too risky, but you could shift whatever percent you are comfortable with, beyond your normal rebalancing allocation.

Note that I did this in the last crash, I started shifting all bonds over when S&P was at 1200 and was 100% stocks at around S&P 950. Ran out of money when S&P was < 800 but would have continued buying down if I had the funds. I made out like a bandit obviously and I'm back at my 80/20 allocation now waiting for the next crash.

Thoughts?

[dickenjb]

I did this as well, going to 92% in stocks in Nov 2008. I would not do it again nor would I recommend it. There were some pretty gut wrenching moments when the market made new lows in March of 2009.

I guess if you are young it is less risky than when I did it two years from retirement. I figured I needed to double down and I would either end up retiring at 55 as planned or work until 60.

Since then I have fully adopted the BH way and am maintaining 60% equities at all times. Before I could always extend my working years, now I can not.

[gravlax]

Sounds similar to the "buy on the dips" strategy.

The only problem is no-one ever knows where the dips were (or ended), except for in hindsight.

What if the stock market drops 20%, you adjust your allocation to 100% equities, and then the market drops another 40%? Or what if the market does not recover for 5 years and bonds do well during that period?

[Call_Me_Op]

What if the next crash happens to be another depression, and stocks lose 90% of their value. Do you really want 100% allocation to stocks in that scenario?

[Stryker]

I find major down markets like 2008 - 2009 a great time to find out any minor or major weaknesses in the portfolios and shore them up. At the time, I was kept busy buying individual stocks until I ran out of cash.

[Jebediah]

Definitely, you should go 100% equities at the exact market bottom, then 100% bonds at the exact top.

[jcw]

Ok, there is obviously no way to call exact bottom. One strategy could be for every 20% total drop from the market tip, you adjust a total of 25% of your bond allocation into stock. So if you had a 20% bond allocation and the S&P high were 1500, for every 20% drop of 300 points, you would allocate an additional 5% of your total portfolio into stock. It would look like:

S&P 1200 : 85/15 [stock/bond]
S&P 900: 90/10
S&P 600: 95/5
S&P 300: 100/0


Thus, you can withstand an 80% total market drop from the top and still buy on the dip. Adjust the percentage however you wanting according to your risk appetite. The point is, any additional allocation (even 1% if you are ultra conservative) to stock during a 20% drop is, more than likely, going to give you a higher return especially if you have a 30-40 year time horizon like I do.

Obviously, don't do this is you are going to retire in the next 5-10 years or whatever.

[Rodc]

What if the next crash happens to be another depression, and stocks lose 90% of their value. Do you really want 100% allocation to stocks in that scenario?

Yeah, buying at a mere 20% down does not seem wise. That is a normal yawner of a dip, hardly worth noting. Even doing so in the last couple of crashes was way too early.

Frankly rebalancing during a -50% or more crash (done so twice) is enough excitement for me.

[freebeer]

Ok, there is obviously no way to call exact bottom. One strategy could be for every 20% total drop from the market tip, you adjust a total of 25% of your bond allocation into stock. So if you had a 20% bond allocation and the S&P high were 1500, for every 20% drop of 300 points, you would allocate an additional 5% of your total portfolio into stock. It would look like:

S&P 1200 : 85/15 [stock/bond]
S&P 900: 90/10
S&P 600: 95/5
S&P 300: 100/0


Thus, you can withstand an 80% total market drop from the top and still buy on the dip. Adjust the percentage however you wanting according to your risk appetite. The point is, any additional allocation (even 1% if you are ultra conservative) to stock during a 20% drop is, more than likely, going to give you a higher return especially if you have a 30-40 year time horizon like I do.

Obviously, don't do this is you are going to retire in the next 5-10 years or whatever.

OK so construct a model and see how following this strategy in the past would have compared to a normal BH strategy like 60/40 with annual rebalancing. Throwing out a strategy without any data showing how it would have worked isn't particularly helpful IMO.

[jcw]

Ok, there is obviously no way to call exact bottom. One strategy could be for every 20% total drop from the market tip, you adjust a total of 25% of your bond allocation into stock. So if you had a 20% bond allocation and the S&P high were 1500, for every 20% drop of 300 points, you would allocate an additional 5% of your total portfolio into stock. It would look like:

S&P 1200 : 85/15 [stock/bond]
S&P 900: 90/10
S&P 600: 95/5
S&P 300: 100/0


Thus, you can withstand an 80% total market drop from the top and still buy on the dip. Adjust the percentage however you wanting according to your risk appetite. The point is, any additional allocation (even 1% if you are ultra conservative) to stock during a 20% drop is, more than likely, going to give you a higher return especially if you have a 30-40 year time horizon like I do.

Obviously, don't do this is you are going to retire in the next 5-10 years or whatever.

OK so construct a model and see how following this strategy in the past would have compared to a normal BH strategy like 60/40 with annual rebalancing. Throwing out a strategy without any data showing how it would have worked isn't particularly helpful IMO.

Agree. That's why the very first sentence in the first post asks if any has backtested this. I could model it myself since I have some downtime, but why bother if the work is done here.

[rustymutt]

The flaw I see with this ideal is that the markets could move sideways for years. Look at Japan.

[jcw]

The flaw I see with this ideal is that the markets could move sideways for years. Look at Japan.

True but in the 3-fund portfolio, you have total international. So you have exposure to the world. Unless securities around the world stay flat, then it shouldn't be a problem. Since there is a positive correlation between domestic and international, a big drop in one usually effects the other.

[jcw]

Ok, I have backtested this strategy over the last 30 years from Jan 1983 to Jan 2013. I was able to show a 122% total portfolio improvement using this strategy.

Strategy: shift AA from bonds to stocks on down months, and then reset AA back to normal when the market reached new highs.
Initial Investment: $1 in Jan 1983
AA: 80% Bonds, 20% stock, rebalanced monthly
Final value in monthly rebalanced 80/20 portfolio: $9.40 [840% growth]
Final value in adjusted portfolio to buy on dips: $10.62 [962% growth]
Difference: 122%

Also had similar but less gain when testing on a daily basis and buying on daily drops of >8%. In the monthly case, I only readjusted when a total monthly drop of more than 10% occurred and I would shift 10% of my bonds to stock (ie: 90/10 instead of 80/20), then reset my AA when a new high was set. I assumed a constant fixed income bond return of 5%. Note that a 10% monthly drop only occurred about 6 times in the last 30 years so it's a fairly rare event. You can probably choose some better rules, like a total % drop from the market high instead of a monthly drop but this was easiest to model and shows a significant improvement for very little work.

I have a spreadsheet where are the parameters are tunable so you can play around with it yourself. How do I attach that?

[freebeer]

Ok, I have backtested this strategy over the last 30 years from Jan 1983 to Jan 2013. I was able to show a 122% total portfolio improvement using this strategy.

Strategy: shift AA from bonds to stocks on down months, and then reset AA back to normal when the market reached new highs.
Initial Investment: $1 in Jan 1983
AA: 80% Bonds, 20% stock, rebalanced monthly
Final value in monthly rebalanced 80/20 portfolio: $9.40 [840% growth]
Final value in adjusted portfolio to buy on dips: $10.62 [962% growth]
Difference: 122%

Also had similar but less gain when testing on a daily basis and buying on daily drops of >8%. In the monthly case, I only readjusted when a total monthly drop of more than 10% occurred and I would shift 10% of my bonds to stock (ie: 90/10 instead of 80/20), then reset my AA when a new high was set. I assumed a constant fixed income bond return of 5%. Note that a 10% monthly drop only occurred about 6 times in the last 30 years so it's a fairly rare event. You can probably choose some better rules, like a total % drop from the market high instead of a monthly drop but this was easiest to model and shows a significant improvement for very little work.

I have a spreadsheet where are the parameters are tunable so you can play around with it yourself. How do I attach that?

This is a 13% improvement in results not a 122% percent improvement.

[jcw]

oh yeh. 122% on your initial $1 but 13% on your total portfolio. Maybe not great, but perhaps not so bad considering it takes a few minutes of work over many years.

[nedsaid]

Easy to say, hard to do.

I "backtested" my emotions back to 2008-2009 and I was too scared to rebalance into stocks from bonds. But I didn't sell any stocks either. What I did do was put 100% of my new contributions into the stock market for about a year during the depths of the bear market. Then I went back to my normal plan of putting 60 percent of new contributions into stocks and 40 percent of new contributions into bonds.

You never really know what kind of investor you are until you have experienced a bear market. You never really know your risk tolerance either until you have felt the emotional pain of large market losses.

[555]

Backtesting can fool you.

[dkturner]

oh yeh. 122% on your initial $1 but 13% on your total portfolio. Maybe not great, but perhaps not so bad considering it takes a few minutes of work over many years.

41 basis points of annualized return seems trivial for the additional risks you outlined in the OP.

[EternalOptimist]

I'm confused, isn't this a form of 'timing' which I thought was frowned upon

[Dandy]

In general I would stick to your target allocation and re balance plan. In extreme times more like 2008 than just 20% I have no problem doing a bit more - lets call in over balancing or a tactical allocation say by another 5 or 10% but would not go all in.

It is not possible to call the bottom. So, going all in after a 50% drop in equities doesn't mean you have reached bottom. If you go all in you could lose more and have no fixed income assets to buy when the market is lower. And may have to sell "undervalued" equities to re balance or pay expenses.

To me it when people are reaching for that extra that they tend to underweight risks and their "greed" and overestimate their smarts. What looks like easy money with moderate risk often turns out to be far different.

[Rodc]

Ok, I have backtested this strategy over the last 30 years from Jan 1983 to Jan 2013. I was able to show a 122% total portfolio improvement using this strategy.

Strategy: shift AA from bonds to stocks on down months, and then reset AA back to normal when the market reached new highs.
Initial Investment: $1 in Jan 1983
AA: 80% Bonds, 20% stock, rebalanced monthly
Final value in monthly rebalanced 80/20 portfolio: $9.40 [840% growth]
Final value in adjusted portfolio to buy on dips: $10.62 [962% growth]
Difference: 122%

Also had similar but less gain when testing on a daily basis and buying on daily drops of >8%. In the monthly case, I only readjusted when a total monthly drop of more than 10% occurred and I would shift 10% of my bonds to stock (ie: 90/10 instead of 80/20), then reset my AA when a new high was set. I assumed a constant fixed income bond return of 5%. Note that a 10% monthly drop only occurred about 6 times in the last 30 years so it's a fairly rare event. You can probably choose some better rules, like a total % drop from the market high instead of a monthly drop but this was easiest to model and shows a significant improvement for very little work.

I have a spreadsheet where are the parameters are tunable so you can play around with it yourself. How do I attach that?

This is a 13% improvement in results not a 122% percent improvement.

Does one data point make a trend?

[Call_Me_Op]

The real problem is that in the midst of a major crash - since we do not know for sure where it will stop - a great many people simply cannot bring themselves to rebalance - and it's perfectly understandable. It's easy to look back and say you would have done it - but it's a different thing to be able to make a major stock purchase when they are going "to hell in a hand-basket."

[jeffyscott]

Easy to say, hard to do.

I "backtested" my emotions back to 2008-2009 and I was too scared to rebalance into stocks from bonds.

Similar here, but I don't believe it can just be put down to "emotions". The actual facts at the time were we did not know if the economy as we had known it was coming to a brutal, sudden end. I did not know if I would ever get the money that I was adding to stocks back.

I kept my 50% allocation to stocks, but my plan going in had been to keep my stock allocation in dollars the same in a decline. This plan did not really anticipate the level of decline that occurred. I abandoned that strategy in October 2008 and just tried to keep the allocation at 50%.

[midareff]

Had you followed this simple plan you would have rebalanced all the way down and back up again.

http://www.irebal.com/docs/Opportunisti ... ancing.pdf

[midareff]

The real problem is that in the midst of a major crash - since we do not know for sure where it will stop - a great many people simply cannot bring themselves to rebalance - and it's perfectly understandable. It's easy to look back and say you would have done it - but it's a different thing to be able to make a major stock purchase when they are going "to hell in a hand-basket."

It's like religion..... either you believe or you don't.

[jeffyscott]

Another factor to consider, for those who shifted to a higher stock allocation in the 2008-09 decline, is how many dollars were you actually risking. I think it is a far different thing to increase your stock percentage when you have a $100,000 portfolio then when you have $1,000,000.

Adding, say, $20,000 to stocks after a 50% or more decline that resulted in losing $40,000 or so is a far different thing from adding several hundred thousand dollars after having a loss of several hundred thousand dollars.

[tibbitts]

The real problem is that in the midst of a major crash - since we do not know for sure where it will stop - a great many people simply cannot bring themselves to rebalance - and it's perfectly understandable. It's easy to look back and say you would have done it - but it's a different thing to be able to make a major stock purchase when they are going "to hell in a hand-basket."

It's like religion..... either you believe or you don't.

As with religion, your belief can be shaken depending on whether current conditions are affecting you, and how much you believe in your own ability to recover. For example, if you felt your job (or a similar one) was secure, and you had many years of valuable human capital left, it's easy to say you "either believe or you don't." If you'd already lost your job due to the economic conditions, and didn't have any valuable portable skills or many years left in which to use them, then that's another situation. It's not always how many dollars you're risking, it's how many relative to what you believe is your ability to replace them even without a market recovery.

So you might get a false sense of your own ability to rebalance, or possibly over-rebalance as is being suggested here, based on how you dealt with the last crisis.

It's also true that we could easily face decades where, for example, stocks worldwide suffer repeated 19.99% declines, while you were waiting to implement a strategy based on a 20% decline. Then surely someone will advocate discarding total portfolios in favor of county-specific ETFs or something, because some countries would have suffered more than 20% declines, followed by robust recoveries. And then that too will fail to work out, because as someone will discover in that future environment, countries suffering greater than 20% declines will actually have horrible returns going forward. And then there will be another new obvious strategy proposed, and the process will repeat.

Paul

[555]

"Had you followed this simple plan you would have rebalanced all the way down and back up again.

http://www.irebal.com/docs/Opportunisti ... ancing.pdf"

Rebalancing all the way back up again just undoes rebalancing all the way down.

[Rodc]

"Had you followed this simple plan you would have rebalanced all the way down and back up again.

http://www.irebal.com/docs/Opportunisti ... ancing.pdf"

Rebalancing all the way back up again just undoes rebalancing all the way down.

On the way down you sell bonds to by cheap stocks.

When stocks rebound, the way up, you sell now appreciated stocks to buy bonds.

Buy low, sell high, as it were.

Rebalancing on the way down (one way) and on the way up (the other way) is the way you want it to work.

In no way does this "undo" anything.

[555]

"Had you followed this simple plan you would have rebalanced all the way down and back up again.

http://www.irebal.com/docs/Opportunisti ... ancing.pdf"

Rebalancing all the way back up again just undoes rebalancing all the way down.

" On the way down you sell bonds to by cheap stocks.
When stocks rebound, the way up, you sell now appreciated stocks to buy bonds.
Buy low, sell high, as it were.
Rebalancing on the way down (one way) and on the way up (the other way) is the way you want it to work.
In no way does this "undo" anything. "

If you rebalance n times on the way down and then rebalance n times on the way up, then the "rebalancing bonus" is roughly proportional to 1/n^2. Unless n is small you are basically buying at average prices on the way down, and selling at average prices on the way up. To first order there is no gain. The "rebalancing bonus" is a second order effect.

[Rodc]

"Had you followed this simple plan you would have rebalanced all the way down and back up again.

http://www.irebal.com/docs/Opportunisti ... ancing.pdf"

Rebalancing all the way back up again just undoes rebalancing all the way down.

" On the way down you sell bonds to by cheap stocks.
When stocks rebound, the way up, you sell now appreciated stocks to buy bonds.
Buy low, sell high, as it were.
Rebalancing on the way down (one way) and on the way up (the other way) is the way you want it to work.
In no way does this "undo" anything. "

If you rebalance n times on the way down and then rebalance n times on the way up, then the "rebalancing bonus" is roughly proportional to 1/n^2. Unless n is small you are basically buying at average prices on the way down, and selling at average prices on the way up. To first order there is no gain. The "rebalancing bonus" is a second order effect.

That seems to ignore the sequence of returns issue. For example this is why you can't get SWR from average returns.

At any rate I agree that you do not want to rebalance frequently if for no other reason than you want to remember falling knives and the momentum that is often involved.

And accept for the wildest rides, any rebalancing bonus is small and secondary to the point of holding risk in check.

[555]

"That seems to ignore the sequence of returns issue. For example this is why you can't get SWR from average returns."

Maybe you're confusing this thread with another thread.

[spanky123]

I made out like a bandit obviously and I'm back at my 80/20 allocation now waiting for the next crash.

When will be the next crash? What constitutes a crash?

[spanky123]

As with religion, your belief can be shaken depending on whether current conditions are affecting you, and how much you believe in your own ability to recover.

It depends on the degree to which you truly believe in the region. A true believer will never be forsaken.

[spanky123]

The only problem is no-one ever knows where the dips were (or ended), except for in hindsight.

What if the stock market drops 20%, you adjust your allocation to 100% equities, and then the market drops another 40%? Or what if the market does not recover for 5 years and bonds do well during that period?

Then the number of posts claiming timing success will significantly reduce.

[jcw]

I made out like a bandit obviously and I'm back at my 80/20 allocation now waiting for the next crash.

When will be the next crash? What constitutes a crash?

I think of it as a double digit drop over a period of time, usually 1 day, but could be over a month(s). No clue when the next crash will be. What i do know is that in every single crash in the history, except the current one, we have recovered to higher highs. We have currently not reached in new high in the market but it seems to be coming back. My money and your money (assuming you own stocks) are betting that history will repeat and we will see new S&P highs one day, even if it's in 10 years.

The point of this strategy is that these events only occur a few times in our investing lifetimes so we do not have to constantly try to time the market and can generally stick to the BH style of investing. It seems any additional allocation to stocks on big drops will show some improvement. Thus, even if you are super conservative and allocate an additional 1% to stock, you would come out ahead.

[Rodc]

"That seems to ignore the sequence of returns issue. For example this is why you can't get SWR from average returns."

Maybe you're confusing this thread with another thread.

No

The approximation that the average on the way down is the average on the way up so they should cancel ignores sequence of returns and the timing of cash flows between assets.

[Akiva]

"Had you followed this simple plan you would have rebalanced all the way down and back up again.

http://www.irebal.com/docs/Opportunisti ... ancing.pdf"

Rebalancing all the way back up again just undoes rebalancing all the way down.

" On the way down you sell bonds to by cheap stocks.
When stocks rebound, the way up, you sell now appreciated stocks to buy bonds.
Buy low, sell high, as it were.
Rebalancing on the way down (one way) and on the way up (the other way) is the way you want it to work.
In no way does this "undo" anything. "

If you rebalance n times on the way down and then rebalance n times on the way up, then the "rebalancing bonus" is roughly proportional to 1/n^2. Unless n is small you are basically buying at average prices on the way down, and selling at average prices on the way up. To first order there is no gain. The "rebalancing bonus" is a second order effect.

I think this ignores the fact that rebalancing will bring your CAGR closer to arithmetic returns.

[leonard]

On this forum, the answer to "Should I try this flavor of market timing?" is going to be "No". Did you expect otherwise?

[jcw]

On this forum, the answer to "Should I try this flavor of market timing?" is going to be "No". Did you expect otherwise?

I have been lurking here for about 10 years so I suppose I should not be surprised about that. Well maybe if I had been more strategic in the way I broached the topic, people would be more responsive. For example, apparently, 70% of bogleheads believe in strategic rebalancing on a 50% market drop (see thread: viewtopic.php?f=1&t=109401&newpost=1590262). Isn't this market timing? If you define market timing as trying to buy equities outside of your standard purchasing behavior to optimize returns, then yes, that is market timing.

What I am suggesting is not that far off from strategic rebalancing. I am saying to go one step further and allocate slightly more in equities. If you believe in strategic rebalancing, then you inherently believe that buying on a 50% dip will produce superior returns. If you believe this, why not put more in equities, if even a tiny amount?
Seems like even if I were to backtest this for every 30 year period over the last 100 years and show that it produced superior returns, I would still meet with much skepticism. This surprises me because the BH principles rely solely on data-driven analysis that shows that buy/hold passive investing has beaten active investing in the majority of cases. Even though that has been shown to be true through backtesting, no one here questions that even though we all know "past performance is no indicator of future success".

[Akiva]

On this forum, the answer to "Should I try this flavor of market timing?" is going to be "No". Did you expect otherwise?

I have been lurking here for about 10 years so I suppose I should not be surprised about that. Well maybe if I had been more strategic in the way I broached the topic, people would be more responsive. For example, apparently, 70% of bogleheads believe in strategic rebalancing on a 50% market drop (see thread: viewtopic.php?f=1&t=109401&newpost=1590262). Isn't this market timing? If you define market timing as trying to buy equities outside of your standard purchasing behavior to optimize returns, then yes, that is market timing.

No rebalancing isn't market timing, the reason this works has nothing to do with market timing and everything to do with making the portfolio's CAGR better approximate the arithmetic average returns. (Which is what rebalancing accomplishes; that's why it's the one free lunch in finance.)

What I am suggesting is not that far off from strategic rebalancing. I am saying to go one step further and allocate slightly more in equities. If you believe in strategic rebalancing, then you inherently believe that buying on a 50% dip will produce superior returns. If you believe this, why not put more in equities, if even a tiny amount?

Your logic doesn't work because you aren't rebalancing because you believe equities are temporarily undervalued, you are rebalancing to keep your geometric returns in line with arithmetic ones.

Seems like even if I were to backtest this for every 30 year period over the last 100 years and show that it produced superior returns, I would still meet with much skepticism. This surprises me because the BH principles rely solely on data-driven analysis that shows that buy/hold passive investing has beaten active investing in the majority of cases. Even though that has been shown to be true through backtesting, no one here questions that even though we all know "past performance is no indicator of future success".

Look, simple market-wide valuation measures like the median P/E of the index are passably good predictors of expected return. So hypothetically, you could increase your allocation as valuations fell and decrease it as they rose. But again this isn't market timing, it's having a dynamic allocation that accounts for changes in expected returns. (And even this isn't in the Boglehead spirit. Even if it works, most people probably won't do it right or would be psychologically uncomfortable with the behavior required.)

[Rodc]

Seems like even if I were to backtest this for every 30 year period over the last 100 years

Are you willing to use data from stock markets that are now gone because their underlying economies crashed and did not recover?

The challenge is if you use markets currently in existence you have at best three independent 30-year periods; all your data come from averaging (close to correct) those three data points. And those data points are biased high due to survivorship bias.

If markets snap back in some reasonable amount of time then rebalancing at about 50% down will look pretty good. You'll miss the opportunity if the market only drops 45%, and you will be in early if the market drops 90% (but will still do ok). Nowhere near enough data to make a convincing case.

What you miss is what happened to many in the 1930's: sell bonds to buy stocks cheap, then lose your job and have to sell your stocks cheaper. It works out fine on paper, but did not work so well in practice, in part because income is tied to the same economy that is crashing.

So, sure there are examples where this works, and I don't think it is crazy. But it is not the slam dunk a simple paper exercise might make it look. IMHO

[Akiva]

Seems like even if I were to backtest this for every 30 year period over the last 100 years

Are you willing to use data from stock markets that are now gone because their underlying economies crashed and did not recover?

The challenge is if you use markets currently in existence you have at best three independent 30-year periods; all your data come from averaging (close to correct) those three data points. And those data points are biased high due to survivorship bias.

If markets snap back in some reasonable amount of time then rebalancing at about 50% down will look pretty good. You'll miss the opportunity if the market only drops 45%, and you will be in early if the market drops 90% (but will still do ok). Nowhere near enough data to make a convincing case.

What you miss is what happened to many in the 1930's: sell bonds to buy stocks cheap, then lose your job and have to sell your stocks cheaper. It works out fine on paper, but did not work so well in practice, in part because income is tied to the same economy that is crashing.

So, sure there are examples where this works, and I don't think it is crazy. But it is not the slam dunk a simple paper exercise might make it look. IMHO

There's also the fact that even in the US market there are significant extended periods where equities did poorly. For the thirty-six years between 1896 and 1932, stocks went down on average. Similarly, you lost money if you held from 1962 to 1974. So it isn't like buy and hold stock returns are the sure thing over even reasonably long horizons...