Author Topic: Best rebalancing interval: day before US election?  (Read 809 times)

Radagast

  • Handlebar Stache
  • *****
  • Posts: 1236
  • Location: West of the Mountains, East of the Sea
Best rebalancing interval: day before US election?
« on: November 03, 2018, 12:53:25 AM »
The Monday (or Friday, but too late now) before US congressional elections seems like a strong candidate for the best rebalancing period. Once every year, perhaps your birthday, might be the most popular interval and I have seen posters mention quarterly. DIY investing guru Bill Berstein says that it should be in the range of 1-5 years; the reason is that rebalancing frequently goes against momentum, which markets are known to have. Frequent rebalancing causes you to react too soon and too often, throwing good money after bad. Imagine rebalancing into the Great Depression quarterly: throwing bonds and cash into stocks more than ten times! Painful and difficult. And just as bad, frequently selling too early on the way up. While I have not been able to test it, it seems like late every even year would have worked out at least as well as annually would have. November 2000, 2002, 2004, 2006, 2008, etc.

It is also a date that you will hear all over the news and will not forget. Further, the market is pricing in probabilities of various election outcomes. After the election each probability will go to 1 or 0, and the markets will then move to the new reality. While we can't predict what will happen, we know that something will. On the flip side, every four-year presidential election also seems attractive, but the intervals are so sparse it seems unreliable, plus you might forget what exactly you were doing four years ago.

All in all, at very worst this schedule should not give worse results than annual rebalancing, and it might even give better. When two or more options seem equally attractive, we should default to the more amusing one. Amirite?

Villanelle

  • Handlebar Stache
  • *****
  • Posts: 2249
Re: Best rebalancing interval: day before US election?
« Reply #1 on: November 03, 2018, 02:36:11 AM »
To me, this reads like you are attempting market timing without quite market timing.  If that's the day that works for you, cool, but I think it is no better than your birthday or the day you adopted your dog or Pie day or any other single day (or that day every even year). 

Your argument against quarterly rebalancing applies equally to any other day, and as for the election specific stuff, unless you have some special knowledge or prescience, you are just as like to rebalance in a way that loses you money as that gains.  Just like any other day.  As you said, you know something will happen.  But that could by an up market, a down market, or a more or less the same market (and for days, weeks, months, or years), or any particular segment going up or down or staying the same, too. 

Unless there is some historical data that says bonds almost always go up the day of/after an election, (or that they go down, or stocks go up, or down), knowing that some undetermined thing may happen isn't especially valuable.  And if you are rebalancing to a specific asset allocation, that's even more the case.  Because if you are low on bonds (for example) but you have reason to believe bonds will drop the day after the election, then you would either be buying against your own predictions, or not actually rebalancing on your designated day.  (And of course, waiting until after your designated day then absolutely becomes market timing.)

I'm a red panda

  • Walrus Stache
  • *******
  • Posts: 7117
  • Location: United States
Re: Best rebalancing interval: day before US election?
« Reply #2 on: November 03, 2018, 07:13:45 AM »
Does your Investment Plan say that's your rebalancing day? If so, go for it. If you're doing it because of the election it's market timing and just as likely to not work out for you as work in your favor.

cap396

  • 5 O'Clock Shadow
  • *
  • Posts: 38
  • Age: 46
  • Location: FIREd and Traveling the World
Re: Best rebalancing interval: day before US election?
« Reply #3 on: November 03, 2018, 09:09:54 AM »
You could go back 30-40 years and get the historical data, then calculate returns by rebalancing the day before the election vs rebalancing on any other arbitrary day.  My guess is that over the long run there will be very little difference, but who knows, maybe you're on to something here.  And whichever rebalancing method gave a higher return over the past might not give a higher return in the future.

Radagast

  • Handlebar Stache
  • *****
  • Posts: 1236
  • Location: West of the Mountains, East of the Sea
Re: Best rebalancing interval: day before US election?
« Reply #4 on: November 04, 2018, 10:57:48 PM »
To me, this reads like you are attempting market timing without quite market timing. 
Quote
(And of course, waiting until after your designated day then absolutely becomes market timing.)
Does your Investment Plan say that's your rebalancing day? If so, go for it. If you're doing it because of the election it's market timing and just as likely to not work out for you as work in your favor.
Right so lets get this out of the way. Rebalancing back to a target static allocation or an allocation that varies based upon age or other external life situation is, as far as I am concerned, never market timing. You can rebalance daily, based on whims, monthly, when tolerance bands are crossed, during the Olympics, or never rebalance and none of those would ever be market timing. Moving back to your desired risk profile is by definition not market timing, the only question is how much variance you are willing to accept, and as far as I am concerned any answer is acceptable. Deliberately changing your allocation away from your target based on past or expected market movements would be market timing, but not towards your target.

If that's the day that works for you, cool, but I think it is no better than your birthday or the day you adopted your dog or Pie day or any other single day (or that day every even year).

Your argument against quarterly rebalancing applies equally to any other day,
Nope, it applies very specifically to quarterly rebalancing and generally to periodic rebalancing based on intervals less than one year. Thus far, markets definitely have exhibited momentum and have not been random walks. Therefore, frequent rebalancing has caused worse results because you tend to sell assets that will continue to rise and buy assets that continue to fall. One reason to consider 2-year election day rebalancing is that all those you mentioned (and realistically most I can think of) are based on calendar years or shorter intervals. There are lots of amusing annual recurring dates. Quite a few four year ones as well. Not many at two years and I can't think of any at three years. There is no reason (that I realistically know of, though there could be) that election rebalancing would work better, but it is a two year recurring event, and while I have read Bernstein I have never had the chance to play around with longer rebalancing intervals. They seem intriguing, and thinking back it seems like they may have worked ok in practice too.

Quote
and as for the election specific stuff, unless you have some special knowledge or prescience, you are just as like to rebalance in a way that loses you money as that gains.  Just like any other day.  As you said, you know something will happen.  But that could by an up market, a down market, or a more or less the same market (and for days, weeks, months, or years), or any particular segment going up or down or staying the same, too. 

Unless there is some historical data that says bonds almost always go up the day of/after an election, (or that they go down, or stocks go up, or down), knowing that some undetermined thing may happen isn't especially valuable.  And if you are rebalancing to a specific asset allocation, that's even more the case.  Because if you are low on bonds (for example) but you have reason to believe bonds will drop the day after the election, then you would either be buying against your own predictions, or not actually rebalancing on your designated day.
You would like to think that election day is probably not special. In fact it probably is not. I probably have some recency bias because it was very obvious in 2016. Rebalancing is done to protect yourself against the unexpected, and while election day is one of those days where unexpected things are likely to happen, so are all the others and at best it is perhaps only fractionally more likely to be different.

In summary, this is a well known two-year recurring event and there is a paucity of backtesting intervals longer than annually. In theory longer rebalancing intervals are better, to avoid fighting against momentum (but probably not longer than four years). A secondary argument is that election day may be slightly more likely to result in a blip or change in direction than other days, even though nobody can predict which way it will go. And third, I personally would find it kind of amusing.

So far I have never carried through with my own backtesting, hopefully I can find time to test this out in a few months.

Radagast

  • Handlebar Stache
  • *****
  • Posts: 1236
  • Location: West of the Mountains, East of the Sea
Re: Best rebalancing interval: day before US election?
« Reply #5 on: November 04, 2018, 11:00:36 PM »
You could go back 30-40 years and get the historical data, then calculate returns by rebalancing the day before the election vs rebalancing on any other arbitrary day.  My guess is that over the long run there will be very little difference, but who knows, maybe you're on to something here.  And whichever rebalancing method gave a higher return over the past might not give a higher return in the future.
Yup, for sure. Using Portfolio Visualizer the difference between the different rebalancing methods is usually less than 0.5%. At best I think this would give the performance boost of 5/25 band rebalancing, but without the constant monitoring of bands.

MustacheAndaHalf

  • Handlebar Stache
  • *****
  • Posts: 1306
Re: Best rebalancing interval: day before US election?
« Reply #6 on: November 05, 2018, 07:29:54 AM »
... rebalancing frequently goes against momentum ... frequently selling too early on the way up ... should not give worse results than annual rebalancing
Note each of these comments focuses on the performance benefit, not reducing risk.  Since stocks tend to grow faster than bonds, over time stocks will tend to be more of your portfolio.  And if you allow a larger allocation to stocks, you will on average have a better return.  So while accurate, the emphasis on performance ignores the desired level of risk.

Radagast

  • Handlebar Stache
  • *****
  • Posts: 1236
  • Location: West of the Mountains, East of the Sea
Re: Best rebalancing interval: day before US election?
« Reply #7 on: November 05, 2018, 10:21:36 PM »
Right, so here is backtesting data from Portfolio Visualizer for a simple 80% US stock market, 20% US Intermediate Treasury fund since they were both available. There is nothing life changing here, but longer rebalancing intervals monotonically lowered risk by a tiny amount and increased return by a tiny amount, out to 12 months which is all we can test. I think longer intervals would also work out better for people who allocate to international, as it seems like international outperformance comes and goes on lower frequencies than the occasional sharp stock crash.

Rebalance
Interval
Months   CAGR   StDev   Max Loss
Never    8.93%    12.07%   42.96%
12         9.02%   11.22%   39.92%
6           8.90%   11.25%   40.57%
3           8.97%   11.33%   40.98%
1           8.88%   11.36%    41.35%

Edit to add never rebalancing. You can see that there was probably a sweet point in there somewhere between 1 year and never.

https://www.portfoliovisualizer.com/backtest-portfolio?s=y&timePeriod=2&startYear=1985&firstMonth=1&endYear=2018&lastMonth=12&endDate=11%2F05%2F2018&initialAmount=10000&annualOperation=0&annualAdjustment=0&inflationAdjusted=true&annualPercentage=0.0&frequency=4&rebalanceType=3&showYield=false&reinvestDividends=true&symbol1=VFITX&allocation1_1=20&symbol2=VTSMX&allocation2_1=80
« Last Edit: November 05, 2018, 10:47:40 PM by Radagast »

Villanelle

  • Handlebar Stache
  • *****
  • Posts: 2249
Re: Best rebalancing interval: day before US election?
« Reply #8 on: November 05, 2018, 11:57:15 PM »
To me, this reads like you are attempting market timing without quite market timing. 
Quote
(And of course, waiting until after your designated day then absolutely becomes market timing.)
Does your Investment Plan say that's your rebalancing day? If so, go for it. If you're doing it because of the election it's market timing and just as likely to not work out for you as work in your favor.
Right so lets get this out of the way. Rebalancing back to a target static allocation or an allocation that varies based upon age or other external life situation is, as far as I am concerned, never market timing. You can rebalance daily, based on whims, monthly, when tolerance bands are crossed, during the Olympics, or never rebalance and none of those would ever be market timing. Moving back to your desired risk profile is by definition not market timing, the only question is how much variance you are willing to accept, and as far as I am concerned any answer is acceptable. Deliberately changing your allocation away from your target based on past or expected market movements would be market timing, but not towards your target.

If that's the day that works for you, cool, but I think it is no better than your birthday or the day you adopted your dog or Pie day or any other single day (or that day every even year).

Your argument against quarterly rebalancing applies equally to any other day,
Nope, it applies very specifically to quarterly rebalancing and generally to periodic rebalancing based on intervals less than one year. Thus far, markets definitely have exhibited momentum and have not been random walks. Therefore, frequent rebalancing has caused worse results because you tend to sell assets that will continue to rise and buy assets that continue to fall. One reason to consider 2-year election day rebalancing is that all those you mentioned (and realistically most I can think of) are based on calendar years or shorter intervals. There are lots of amusing annual recurring dates. Quite a few four year ones as well. Not many at two years and I can't think of any at three years. There is no reason (that I realistically know of, though there could be) that election rebalancing would work better, but it is a two year recurring event, and while I have read Bernstein I have never had the chance to play around with longer rebalancing intervals. They seem intriguing, and thinking back it seems like they may have worked ok in practice too.

Quote
and as for the election specific stuff, unless you have some special knowledge or prescience, you are just as like to rebalance in a way that loses you money as that gains.  Just like any other day.  As you said, you know something will happen.  But that could by an up market, a down market, or a more or less the same market (and for days, weeks, months, or years), or any particular segment going up or down or staying the same, too. 

Unless there is some historical data that says bonds almost always go up the day of/after an election, (or that they go down, or stocks go up, or down), knowing that some undetermined thing may happen isn't especially valuable.  And if you are rebalancing to a specific asset allocation, that's even more the case.  Because if you are low on bonds (for example) but you have reason to believe bonds will drop the day after the election, then you would either be buying against your own predictions, or not actually rebalancing on your designated day.
You would like to think that election day is probably not special. In fact it probably is not. I probably have some recency bias because it was very obvious in 2016. Rebalancing is done to protect yourself against the unexpected, and while election day is one of those days where unexpected things are likely to happen, so are all the others and at best it is perhaps only fractionally more likely to be different.

In summary, this is a well known two-year recurring event and there is a paucity of backtesting intervals longer than annually. In theory longer rebalancing intervals are better, to avoid fighting against momentum (but probably not longer than four years). A secondary argument is that election day may be slightly more likely to result in a blip or change in direction than other days, even though nobody can predict which way it will go. And third, I personally would find it kind of amusing.

So far I have never carried through with my own backtesting, hopefully I can find time to test this out in a few months.

I'm not sure how any of what you said refutes any of what I said.  If you want to choose that to be your rebalancing day, go for it. 

But if you are choosing that day specifically to try to time the markets, then yes, that is market timing.  But if you stick with that day no matter what the markets or elections do or are projected to do, it won't matter because it's still an annual rebalancing. 

You seem to be conflating an argument for annual rebalancing with an argument for rebalancing based on election day.  Those are not the same thing.  Again, if you choose that day for your annual rebalancing, the results (when compared to choosing St/ Patrick's day or your god's birthday or the summer solstice or the date that coincides with your license plate numbers, or...) will be the same.  So certainly it it a perfectly appropriate strategy.  But as I said, choosing a day because you think it's beneficial market-wise (as opposed to just convenient or easy to remember) it market timing without the market timing.  The outcome will be the same as an annual rebalance not based around market timing.  But the rational for choosing that date specifically is still based on flawed thinking.  But since it is flawed thinking that won't hurt you as long as you fully stick with it no matter what elections and markets are doing on that day, it doesn't matter. 

So, rebalancing once a year is good an healthy, and picking a set date or event t trigger that is good and healthy.  Thinking that you can strategize that day in order to time the market is probably wishful thinking, but it's not going to hurt you.  Is it an attempt at market timing?  I suppose that's a matter of semantics. 

Radagast

  • Handlebar Stache
  • *****
  • Posts: 1236
  • Location: West of the Mountains, East of the Sea
Re: Best rebalancing interval: day before US election?
« Reply #9 on: November 07, 2018, 08:00:32 PM »
One key distinction is that I am proposing rebalancing only once every two years, not annually.

MustacheAndaHalf

  • Handlebar Stache
  • *****
  • Posts: 1306
Re: Best rebalancing interval: day before US election?
« Reply #10 on: November 08, 2018, 05:17:13 AM »
When you focus on max loss, did you realize you're only using 3 years of data?  The largest loss is the 2008 crisis, so all of your max loss numbers are from 2007-2009.

I'd encourage you to exclude those years in order to find other crashes which show different "max loss" results.  It might not change the decision of when to rebalance, but it will let you incorporate more data into your calculations of max loss.  Here's the first few I found using portfolio visualizer:

1972-1974
1987-1989
2000-2002
2007-2009