Opportunistic Rebalancing: A New Paradigm

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Topic Author
Barry Barnitz
Wiki Admin
Posts: 3216
Joined: Mon Feb 19, 2007 10:42 pm
Contact:

Opportunistic Rebalancing: A New Paradigm

Post by Barry Barnitz » Thu Jan 03, 2008 11:20 pm

Opportunistic Rebalancing: A New Paradigm for Wealth Managers by by Gobind Daryanani CFP®, Ph.D, FPA Journal (January, 2008)
Executive Summary

* Wealth managers traditionally rebalance portfolios quarterly or annually to control risk due to asset class drifts. This paper proposes a new paradigm for planners: rebalance less frequently, but look more frequently to find the best opportunities for rebalancing.
* The proposed approach, called opportunistic rebalancing, not only controls portfolio drift, but also provides significant return improvements by capturing buy-low/sell-high opportunities as asset classes sporadically drift relative to each other.
* The paper studies a wide range of market conditions to show that rebalancing return benefits can be more than doubled compared with the traditional annual rebalancing.
* These additional benefits, attributed to transient momentum and mean reversion effects, occur sporadically in time and can only be captured by monitoring portfolios frequently.
* The studies suggest these practical guidelines: (1) use wider rebalance bands, (2) evaluate client portfolios biweekly, (3) only rebalance asset classes that are out of balance—not classes that are in balance, and (4) increase the number of uncorrelated classes used in portfolios.
* The studies show that trading costs and tax deferral are small compared with rebalance benefits.
* Opportunistic rebalancing has already been adopted by a number of leading wealth management firms across the country.
regards,
Additional administrative tasks: Financial Page bogleheads.org. blog; finiki the Canadian wiki; The Bogle Center for Financial Literacy site; Wiki Bogleheads® España.

livesoft
Posts: 72087
Joined: Thu Mar 01, 2007 8:00 pm

Post by livesoft » Thu Jan 03, 2008 11:31 pm

Good paper. Good summaries of different rebalancing methods: calendar, range-bound and opportunistic.

It appears to justify buying on the dips ... if you call that opportunistic rebalancing.

InertiaMan
Posts: 264
Joined: Tue Oct 16, 2007 10:26 am

Post by InertiaMan » Fri Jan 04, 2008 9:16 am

That was a pretty fascinating read. Basically, you let the MARKET determine when to rebalance rather than the CALENDAR. The most frequently you should "look" at your portfolio (to see if the market wants you to rebalance) is once a week and the least frequently is two weeks (although even once per month is only negligbly less optimal).

If I were to summarize the optimal rebalancing strategy based on their conclusions, one should:

- use rebalance bands of 20%*
- use tolerance bands of 50%*
- check your portfolio every 5 to 10 trading days to determine if rebalancing is justified

* example: for an asset class targeted at 10% of portfolio, you rebalance when it is less than 8%, or greater than 12%, but you only rebalance halfway back to the target (ie, from 8 to 9, or from 12 to 11).

Pretty methodical, and surprising how rare it results in a trade: typically 3 trades per year for a 5 asset class portfolio. The paper claims the benefit is about 55bps (versus no rebalancing) and about 30bps (versus annual). They claim the tax and trading costs will suck away 15% of that benefit (8bps) AT MOST.

One final bonus of this: it gives buy-and-hold Bogleheads a justifiable reason to look closely at the status of their portfolio once a week! :-)

rwwoods
Posts: 1312
Joined: Thu Jun 07, 2007 7:54 pm
Location: The Villages, Florida

Post by rwwoods » Fri Jan 04, 2008 12:51 pm

InertiaMan said:
* example: for an asset class targeted at 10% of portfolio, you rebalance when it is less than 8%, or greater than 12%, but you only rebalance halfway back to the target (ie, from 8 to 9, or from 12 to 11).
Actually, the author said that is adequate to rebalance back to anywhere within the tolerance band. For your example, it would be adequate to rebalance anywhere within 9-11%.

InertiaMan
Posts: 264
Joined: Tue Oct 16, 2007 10:26 am

Post by InertiaMan » Tue Jan 08, 2008 10:22 am

rwwoods,

Good point. My example was incorrectly implying that you always rebalance only to the edge of the tolerance band, but that is probably only true for a subset of the rebalanced asset classes.

If one assumes that you seek to minimize the size of a rebalancing transaction (to minimize tax consequences, for example), then at least one of the asset classes would get rebalanced only to the edge of it's tolerance band, right? But one may have to take other asset class(es) deeper into their tolerance band(s), as necessary to bring the original class to the edge.

User avatar
tadamsmar
Posts: 9015
Joined: Mon May 07, 2007 12:33 pm

WLIB Test

Post by tadamsmar » Tue Jan 08, 2008 12:48 pm

This seems to pass the WLIB Test: If this idea is true, could you build the World's Largest Investment Bank using it?

Yep, if there is a pattern of reversal in the market that allows one to profit from opportunistic rebalancing, then apply leverage and you can build the WLIB.

Assuming the return is big enough to beat any costs and taxes. Rebalancing in a tax-deferred fund tends to be a no fee affair, so this can, in principle, fail the WLIB Test.

Tramper Al
Posts: 3629
Joined: Thu Oct 18, 2007 11:42 am

Post by Tramper Al » Tue Jan 08, 2008 1:23 pm

InertiaMan wrote: If I were to summarize the optimal rebalancing strategy based on their conclusions, one should:

- use rebalance bands of 20%*
- use tolerance bands of 50%*
- check your portfolio every 5 to 10 trading days to determine if rebalancing is justified
That's very interesting, in that it's very similar to the (presumed arbitrary) rebalance scheme that I have been using. I rebalance when a (major) asset class is down 20%, and I rebalance half way back to target.

It is just the one asset class that triggers the rebalance, and it goes half way back to its target. All others participate in proportion to however much each is off its own target, so they don't get overly near target either.

One thing I do a little differently, though, is I have made my re-balance limits symmetrical on the multiplicative (or log) scale. So my primary limits are 20% loss and 25% gain. For my sub-allocations, the limits are 33% loss, 50% gain.

I also will allow a re-balance to trigger only at end-of-quarter. However, I keep such close tabs I always know where to put any distributions or new money in the mean time. Sometimes that is enough to get back under the threshold.

I have to consider this very carefully, as the ratio of new money to existing balances is quite low. That is I have no hope of keeping AA in line by new contributions alone. A good year for equities (yes even like 2007) is likely to send the Emerging Markets class above its limit. Thus I try to practice AA risk management without excessive tax penalties.

User avatar
Diver
Posts: 107
Joined: Thu Nov 15, 2007 11:13 pm

Post by Diver » Tue Jan 08, 2008 10:02 pm

InertiaMan wrote: - use rebalance bands of 20%*
- use tolerance bands of 50%*
- check your portfolio every 5 to 10 trading days to determine if rebalancing is justified
it's really nice article and the research looks very sound. but i'm not quite sure what is the advantage of not rebalancing all the way to the target % numbers though? they said that 25-50-75% of tolerance bands don't make significant difference. since that number won't determine if you need to seel another asset to buy one that declined, it looks like it's up to the investor to decide if he/she want to rebalance all the way or just half way. the numbers they provide seem to be for 50% tolearance bands, so it looks like that's the number they recommend.

also articles referes to several figures, but there are only tables in there. am i missing something ?

and again - very interesting article, thanks !!

User avatar
czeckers
Posts: 1042
Joined: Thu May 17, 2007 3:49 pm
Location: Upstate NY

Post by czeckers » Tue Jan 08, 2008 10:20 pm

I definitely found the article to be very interesting. I've been wondering for a while why Swenson rebalances daily. I was making the assumption that it was just resetting everything to predetermined ratios daily, which did not make a lot of sense unless to take advantage of very choppy markets. I think this article has shed some light on the topic.

I too am a little confused about the tolerance bands, and why 0% is not recommended. My guess (mostly uneducated) is that they are relating this strategy to its efficiency in a taxable account with an eye toward minimizing trading costs. In that case the 0% tolerance band (rebalancing all holdings to target) would involve too many transactions to be cost effective in the long run. With a wider tolerance band, it may be possible to execute only 1 or 2 trades to reach targets, making for a more efficient strategy.

Other thoughts?

-K
The Espresso portfolio: | | 20% US Total Mkt, 20% Small Cap Value, 10% US REIT, 10% Developed Int'l, 10% EM, 30% Inter-term US Treas | | "A journey of a thousand miles begins with a single step."

User avatar
Diver
Posts: 107
Joined: Thu Nov 15, 2007 11:13 pm

Post by Diver » Tue Jan 08, 2008 10:26 pm

czeckers wrote:My guess (mostly uneducated) is that they are relating this strategy to its efficiency in a taxable account with an eye toward minimizing trading costs. In that case the 0% tolerance band (rebalancing all holdings to target) would involve too many transactions to be cost effective in the long run. With a wider tolerance band, it may be possible to execute only 1 or 2 trades to reach targets, making for a more efficient strategy.
as far as I understand the tolerance band will not drive if rebalance required or not - rebalance bands do that. and one the decision to bring one particular asset back is made, tolerance band will drive "how far back" the asset will be brought.

I think it makes sense to rebalance half way in taxable account to minimize tax implications, but then if it's in tax deferred, would the recommendation be to rebalance all the way ?

I think rebalancing half way has something to do with riding the momentum longer - if you don't go back all the way to your target allocation then you won't pour as much money in your asset that is down and there is some benefits to rip if that asset keeps going down so more. on the other hand it will not work if the asset reverts and starts coming back. so I'm still a bit confused about this particular part of the strategy.

User avatar
czeckers
Posts: 1042
Joined: Thu May 17, 2007 3:49 pm
Location: Upstate NY

Post by czeckers » Tue Jan 08, 2008 10:40 pm

Say a fund exceeds the rebalancing band and triggers a transaction to bring it back into line, it will necessarily require an exchange into or out of another fund to balance the trade.

Assuming two funds with equal weights within a portfolio: If Fund A exceeds the band by 20% and fund B drops below its band 20%, then one can sell Fund A and buy Fund B and get back exactly to target.

However, if Fund A exceeds its band by 20% triggering a transaction, but Fund B is only off by 5%, then one can rebalance back only part way so as not to bring Fund B beyond its band.
The Espresso portfolio: | | 20% US Total Mkt, 20% Small Cap Value, 10% US REIT, 10% Developed Int'l, 10% EM, 30% Inter-term US Treas | | "A journey of a thousand miles begins with a single step."

User avatar
Diver
Posts: 107
Joined: Thu Nov 15, 2007 11:13 pm

Post by Diver » Tue Jan 08, 2008 10:54 pm

i guess another case for tollerance bands would be the case when there is a need to sell 2 funds in order to bring another fund back in balance.the tolerance bands could justify only selling one fund and keeping the fund in question only half way back but within tollerance band range.

User avatar
czeckers
Posts: 1042
Joined: Thu May 17, 2007 3:49 pm
Location: Upstate NY

Post by czeckers » Tue Jan 08, 2008 11:22 pm

I agree. In several places in the article they refer to minimizing transaction costs. They also take tax implications into account, saying that by using the 20% band, that the extra 80bp or so they achieved via this method more than offset the costs. I'm sure the margin would be slimmer if additional transactions were required.

I do wish they had made mention about whether rebalancing exactly back to target is preferred in tax-exempt/non-profit accounts using mutual funds, and thus no taxes or overt transaction costs.
The Espresso portfolio: | | 20% US Total Mkt, 20% Small Cap Value, 10% US REIT, 10% Developed Int'l, 10% EM, 30% Inter-term US Treas | | "A journey of a thousand miles begins with a single step."

User avatar
Diver
Posts: 107
Joined: Thu Nov 15, 2007 11:13 pm

Post by Diver » Wed Jan 09, 2008 10:00 am

czeckers wrote:I do wish they had made mention about whether rebalancing exactly back to target is preferred in tax-exempt/non-profit accounts using mutual funds, and thus no taxes or overt transaction costs.
that's exactly what i thought ! i think it would be the preferred way and it other case, not balancing back all the way will be "tolerated" since it saves on taxes/transaction. hence the name

IlliniSigEp
Posts: 202
Joined: Thu Mar 01, 2007 8:54 pm
Location: Chicago

Mometum effect

Post by IlliniSigEp » Wed Jan 09, 2008 10:24 am

Hi Guys,
My take on not rebalancing all the way back to 0 is that this would not capitalize on the momentum effect. Momentum is a real force in the market and works in both directions. Fama and French even postulate that momentum might be 'the 4th factor' outside the FF 3-Factor model. If you have a winner that has exceeded your rebalancing band on the upside, it might have even more momentum left in it. Only moving back half way allows more money to ride on this upward momentum. On the downside, if you rebalance up only half way, you have exposed less capital to downside momentum than if you rebalanced all the way back to 0. Just my take!

-Dave

User avatar
Diver
Posts: 107
Joined: Thu Nov 15, 2007 11:13 pm

Post by Diver » Wed Jan 09, 2008 10:28 am

but wouldn't that be in effect letting your rebalance band strech further since you let it ride further ? in effect not rebalancing back to zero you allow an asset to go beyond 20% to 25-30% and that was proven to provide worse results. i understand that it's not 100% the same, but pretty close in my mind.

IlliniSigEp
Posts: 202
Joined: Thu Mar 01, 2007 8:54 pm
Location: Chicago

Post by IlliniSigEp » Wed Jan 09, 2008 1:23 pm

Diver wrote:but wouldn't that be in effect letting your rebalance band strech further since you let it ride further ? in effect not rebalancing back to zero you allow an asset to go beyond 20% to 25-30% and that was proven to provide worse results. i understand that it's not 100% the same, but pretty close in my mind.
You would not let your band expand past 20%. Lets say an asset gets to +30% relative to your target and your band is set at 20%. This asset is experiencing upward momentum for sure. You rebalance back to +10% of your target. You now have 10% more of the asset relative to your target, but you have less than the max 20%. If the asset keeps surging between now and the next time you check on your bands, you have slightly more allocated to it and capitalize on the momentum.

-Dave
Last edited by IlliniSigEp on Wed Jan 09, 2008 1:34 pm, edited 1 time in total.

User avatar
Diver
Posts: 107
Joined: Thu Nov 15, 2007 11:13 pm

Post by Diver » Wed Jan 09, 2008 1:26 pm

IlliniSigEp wrote:You don't let the asset 'ride further', you simply keep a little bit more in the asset than totally rebalancing back to 0.

-Dave
well, isn't it letting it ride further ? if you have positive momentum for that asset, you don't sell it all the way to its target allocation, but keep a little bit more in there, riding the moemntum with more money.

i wasn't implying expanding the rebalance band further, i was implying that not rebalancing it back to 0 let you ride momentum longer, the same thing as you had rebalancing band wider. sorry if i wasn't clear

IlliniSigEp
Posts: 202
Joined: Thu Mar 01, 2007 8:54 pm
Location: Chicago

Sort of agree

Post by IlliniSigEp » Wed Jan 09, 2008 1:37 pm

I sort of agree but if you were to expand your rebalancing band you would change your risk profile as well because the absolute amount of an asset could get really out of whack. By keeping the same bands but allowing a little more of an asset on a relative basis to ride, your absolute risk is limited. That is why in the paper they eliminate the 25% band because it caused the stock/bond ratio to vary widely (ie greater than an absolute 5%).

-Dave

User avatar
Diver
Posts: 107
Joined: Thu Nov 15, 2007 11:13 pm

Post by Diver » Wed Jan 09, 2008 1:44 pm

agreed

cherijoh
Posts: 6590
Joined: Tue Feb 20, 2007 4:49 pm
Location: Charlotte NC

Post by cherijoh » Wed Jan 09, 2008 2:27 pm

Diver wrote: it's really nice article and the research looks very sound. but i'm not quite sure what is the advantage of not rebalancing all the way to the target % numbers though? they said that 25-50-75% of tolerance bands don't make significant difference. since that number won't determine if you need to seel another asset to buy one that declined, it looks like it's up to the investor to decide if he/she want to rebalance all the way or just half way. the numbers they provide seem to be for 50% tolearance bands, so it looks like that's the number they recommend.
The reason you may not want to rebalance back to target is to ensure you don't overcorrect. (BTW this is standard process control theory, for all you non-engineers out there :-))

Say your EM funds are above the upper limit of 12% for a target of 10%. After you rebalance, the EM fund NAV can increase still further - in which case you may again have to sell some shares. On the other hand, the fund NAV could also fall back. This would bring you closer towards your target, if you had corrected only half way. But if you corrected back to target on your first sell, now you are below target. And of course the half-way correction results in having to pay less capital gains.

If you own an index fund without any other transaction costs, the partial correction would therefore be favored (esp. when you are checking frequently for the need to correct). If you own an ETF, you would need to factor in the costs of buying and selling before deciding how much to correct, since several smaller transactions will cost more than one big one.

C

User avatar
czeckers
Posts: 1042
Joined: Thu May 17, 2007 3:49 pm
Location: Upstate NY

Post by czeckers » Thu Jan 10, 2008 8:18 am

The tolerance band concept (or standard process control threory per cherijoh) in interesting... something I'm still trying to wrap my brain around. Does anyone know of any papers that address this specifically?

-K
The Espresso portfolio: | | 20% US Total Mkt, 20% Small Cap Value, 10% US REIT, 10% Developed Int'l, 10% EM, 30% Inter-term US Treas | | "A journey of a thousand miles begins with a single step."

rwwoods
Posts: 1312
Joined: Thu Jun 07, 2007 7:54 pm
Location: The Villages, Florida

Post by rwwoods » Thu Jan 10, 2008 10:09 am

I think you guys are overanalyzing the tolerance bands too much. The paper just indicated that it was sufficient to rebalance into the band. Rebalancing back to zero was not required to produce the reported rebalance bonus. Unfortunately, the paper did not report what the difference in the bonus was between the two methods. However, it was implied that the difference was mininal enough that it was inconsequential.

User avatar
czeckers
Posts: 1042
Joined: Thu May 17, 2007 3:49 pm
Location: Upstate NY

Post by czeckers » Thu Jan 10, 2008 12:44 pm

I would generally agree given the statement that tolerance bands of 25, 50, or 75 percent made no difference... but why no mention of 0%? Many investors are invested in tax-deffered accounts with mutual funds, there are no tax consequences and no overt trading costs removing the barriers to rebalancing back to exact AA.

-K
The Espresso portfolio: | | 20% US Total Mkt, 20% Small Cap Value, 10% US REIT, 10% Developed Int'l, 10% EM, 30% Inter-term US Treas | | "A journey of a thousand miles begins with a single step."

User avatar
RobertH
Posts: 219
Joined: Thu Mar 01, 2007 11:35 pm
Location: Bay Area

Post by RobertH » Thu Jan 10, 2008 2:57 pm

I liked this article very much. As I read it, the point of the tolerance bands is less about truing up the asset that triggers the rebalance and more about minimizing the number of compensating transactions that the first transaction triggers.

Wider tolerance bands enable one to rebalance out-of-balance asset classes with fewer compensating transactions in the rest of the portfolio.

This raises a question of methodology which is hard to resolve from the online version of the article (it references figures not present): does the rebalancing algorithm sell (or buy) the minimum amount needed to reach the tolerance band, or does the rebalancing algorithm try to rebalance to the target amount, accepting a lesser amount only when a single compensating transaction is not possible? I suspect the former.

Since the most volatile asset classes tend to have the smallest allocations, it is probably often possible to rebalance them all the way to the target (vs to the nearest tolerance band) with one compensating transaction. For my purposes, this is what I would be inclined to do, but it would be interesting to know what the authors actually examined.

Robert

User avatar
jeffyscott
Posts: 8866
Joined: Tue Feb 27, 2007 9:12 am
Location: Wisconsin

Post by jeffyscott » Fri Jan 11, 2008 9:58 am

This is basically what I have been doing for a while. I first started this a few years ago with selling whenever a fund got above its target by a certain amount. For a while this is all that I needed to do...sell, sell, sell.

I then got in trouble with T. Rowe's market timing police when bi-directional market volatility led to buying as well as selling. I had an emerging market round trip within 90 days, or something. (All of $1000 was involved, but they shut down on-line trading access until we called)

So I revised my policy, increasing the ranges and also only buying by increasing periodic purchase rate rather than lump sum. What I do is start buying when 15% below. I have been trying to decide what I should do when/if things fall further. I thought I'd like to buy a lump sum when something falls further to, say, 20% below...but I worried that I might get in trouble again, if I then needed to sell too soon.

This paper has given me some ideas, I may look at doing lump sum purchase, if anything falls to 20% or more below target, but only buy enough to get within 10% of target. This would hopefully leave enough room to avoid another arrest by market timing police.

I have semi-automatic, continuous (for ETFs) or daily (for funds) monitoring by using yahoo finance alerts. M* alerts could also be used.
Time is your friend; impulse is your enemy. - John C. Bogle

User avatar
Diver
Posts: 107
Joined: Thu Nov 15, 2007 11:13 pm

opportunistic rebalancing and dividends

Post by Diver » Wed Jan 30, 2008 3:47 pm

Studies like this suggest that rebalancing less frequently and with higher tolerance bands produce
better results since investor will take advantage of longer lasting trends. Or at least that what historical
data shows. But it seems logical and make sense. Which makes me wonder what is the best way to
handle distribution/dividends …

One option would be to invest the dividends into the asset that paid off the divided and keep the current
portfolio allocation and ride the momentum longer till the portfolio needs rebalancing. Another option
would be to use the dividends to buy more of the asset that went down, but this essentially will be
rebalancing "more often" or with smaller tolerance bands, which should be counter-productive according to
this study.

For example an investor has portfolio that consists of two asset classes - A and B. The desired asset
allocation is 60/40. The rebalancing bands - 20%, meaning rebalancing will occur when either asset
class drifts more then 20%.

Let's say that over time the allocation changes to 65/35, which is still within the boundaries. At this
point Asset A pays off dividends/distribution in amount of 5%. Now portfolio has the following:

Asset A - 61.75%
Asset B - 35%
Cash - 3.25%

Now my question is what to do with the cash?

The investor can buy more of the asset B which will bring the portfolio composition close to desired allocation.
Also he/she will buy more of the asset that went down in price, which make sense, since it's cheaper. But in
effect it will perform premature portfolio rebalancing that will not let the portfolio get to the desired rebalancing
bands and will not let to capture that little extra return that can be produced if the rebalancing happens at the
higher tolerance bands.

Or he/she can buy more of the asset A. That will match what the study suggests - let portfolio drift outside of
the tolerance bands before doing any rebalancing. But it also seem illogical in a sense that asset A is already
went up and is expensive and the investor will buy more of it.

The same would happen if investor has some extra money coming into a portfolio that is out of balance but still
within the tolerance bands. Should investor put money in according to the original allocation and not rebalance
till it gets to the tolerance bands, or should investor try to address the small imbalance with new money now ?

Sorry if my explanation is a little vague, I hope you can understand where I'm going with this ...

BPet
Posts: 58
Joined: Fri Feb 08, 2008 6:08 am

Tolerance Band

Post by BPet » Sun Feb 10, 2008 6:16 am

My take on the tolerace band as described in the article is as follows:

The goal is to 'attempt' to rebalance back to the target allocation. By doing this, you can lock in all of the 20% gain (assuming a 20% Rebalance Band), or on the downside, you can benefit by purchasing the maximum number of new shares at a 20% reduction in share price. The article seems to indicate that the additional capital gains would be offset by the rebalance benefits.

If it would require too many trades, such that the cost of each trade would have a material effect on your return, then it is "Tolerable" to rebalance to somewhere within the Tolerance Band. So, if your investments total $350K, then the costs of the additional trades are not worth the benefit gained by moving all the way back to the target allocation. If, however, your investments total is much larger, say $5M, then the costs of the trades (especially if its a flat $12-$20 fee), would be negligable, and so it would make more sense to re-balance all the way back to the target allocation and capture those full 20% gains or discounts.

Its my opinion that the Momentum gain is already achieved by using the 'wide' rebalance band of 20%, as opposed to using a narrow band of 5-10%. I don't think the tolerance band is for momentum purposes, the wide rebalance band achieves that.

p.s. Not sure if my logic is correct, but its what I got? , This is my first post :D

icehawk21
Posts: 3
Joined: Thu Dec 06, 2007 2:44 pm

Tolerance Bands

Post by icehawk21 » Thu Mar 20, 2008 1:13 pm

I was interested in the comments contained in this thread regarding the tolerance bands so I decided to email Gobind for the answer. To summarize, the tolerance bands are useful for minimizing transaction costs and capturing momentum effects. In fact, he suggests momentum plays a bigger role in the rebalancing premium he identified. Here is his reply to my email:
Please post for me
There are two reasons for tolerance bands
First is what you observed. Fewer trades. Even if all accounts in family were iras you still should use tolerace bands tro save trades
But the second reason is more important
Momentum
The classes within band are on a momentum swing . Very often, so you do not want to squash them to the target
So even if trading costs were zero use tolerance bands
Make the bands wide to capture sufficient momentum
Loolk frequently then you gget an average of 30 bps over annuall
Hope that helpas

Please correct spellling
I am sending from blackberry
Regards to all who are reading the article
I know you are a thinking bunch

----- Original Message -----
From: Bryan Ambrogiano <bambrogiano>
To: Daryanani, Gobind
Sent: Mon Mar 17 17:29:22 2008
Subject: Opportunistic Rebalancing question

Dear Gobind,

I have recently read (and re-read) your article on 'Opportunistic Rebalancing: A New Paradigm for Wealth Managers' and plan on incorporating your suggestions into my own rebalancing activities. However, I (and many others) have a question regarding the tolerance bands. In fact, you can review some of these thoughts/questions at the following link to the Bogleheads/Diehards forum:
LINK REMOVED
The issue surrounding the tolerance bands is whether or not one intends to rebalance back to the asset class' target allocation as opposed to rebalancing back to the tolerance band. My understanding is that the tolerance band exists to help alleviate additional transactions involving other asset classes to always rebalance back to the target allocation. Having the tolerance band in place mitigates a full rebalance to target and gives the investor some leeway to avoid excessive transactions/costs/taxes. Should the goal be to rebalance back to the target but settle somewhere within the tolerance range with the minimal number of transactions?

Thus, if one were rebalancing in a tax deferred/advantaged account, one would always rebalance back to target (presuming the small fixed transaction cost is relatively small compared to the size of the transactions)? While in a taxable account, it is desirable to simply rebalance to within the tolerance band with the fewest transactions?

Please read the forum link above and you'll note some of the questions that arose on the forum regarding your article. If you have the time and inclination, please respond to my email and I'll post the email thread on the forum.

Thanks in advance!

Bryan Ambrogiano
Hope this helps!

Bryan

P.S. I did not correct his spelling as he requested. He sent his reply on his Blackberry and anybody who uses one (incl. myself) knows the difficulty in spelling/grammar on a Blackberry :D

icehawk21
Posts: 3
Joined: Thu Dec 06, 2007 2:44 pm

Tolerance Bands

Post by icehawk21 » Thu Mar 20, 2008 1:22 pm

As a follow-up to my previous post here is Gobind's reply to another email I sent him.
No I'd stay with 50%. My experience with iRebal tells me that's a pretty good place.



--------------------------------------------------------------------------------
From: Bryan Ambrogiano [mailto:bambrogiano@shaw.ca]
Sent: Thursday, March 20, 2008 2:07 PM
To: Daryanani, Gobind
Subject: Re: Opportunistic Rebalancing question


Gobind,

Thanks you kindly for your reply. I will post the response to the forum today.

Another question...You mentioned that there didn't appear to be much difference with tolerance bands between 25% and 75%. As such, should one randomly choose a tolerance band (such as 50%) or is there some logic to the appropriate choice?

Thanks in advance!

Bryan Ambrogiano
Looks like 50% is probably the best choice for the tolerance bands.

Bryan

User avatar
sewall
Posts: 1341
Joined: Sat Mar 15, 2008 1:57 pm
Contact:

Post by sewall » Fri Mar 21, 2008 8:10 am

The tests in the article were done on a portfolio for which the manager has finer control over asset classes than is the case for me (and many investors). Most of my money is in index and balanced funds, which are rebalanced for me daily (despite the proven suboptimality!!!).

I do have some index funds I use to set stock/bond ratio, within stock I monitor domestic/international and within bond I monitor the %inflation protected. That's it. I'm not looking at capitalization or any other dimension because that is taken care of within the index and balanced funds.

So my question is, if the vast majority of my funds are being rebalanced daily and I am only looking at a much coarser level than was the hypothetical manager in the "Opportunistic Rebalancing" article, how much of the lessons of that article apply? Do they apply at all? Should I really let my domestic/international ride out to 20% of target? It seems that there is a chance I'd be extrapolating the article's results out of sample and that a different test would need to be done to see how best to rebalance under the cirucmstances I describe.

bawelch99
Posts: 5
Joined: Tue Mar 25, 2008 11:58 am

FPA Session this week - Opportunistic Rebalancing

Post by bawelch99 » Mon May 19, 2008 11:58 am

I read the discussion here regarding Gobind Daryanani's paper on opportunistic rebalancing, so I wanted to let you all know that Gobind will be giving an FPA-sponsored talk on this subject on 5/21 at noon EST. This forum will not allow me to post the link, but you can find the session on the FPA site (fpanet dot org) in the Virtual Learning Center under Live Virtual Seminars. Alternately, please feel free to email me directly, and I will send you the link - benjamin.welch (at) tdameritrade (dot) com.

Hope that some of you can attend.

blackball
Posts: 99
Joined: Thu May 01, 2008 9:36 pm

Post by blackball » Mon May 19, 2008 6:44 pm

From reading the article, it appears that the optimal 20% rebalance band is applied to all five asset classes.

Is this optimal?

I would think that each of the asset classes should have a different rebalance band to reflect their different volatilities and "momentum" swings. For example, don't commodities usually have a longer cycle compared to plain equities? This would seem to indicate that commodities should have a higher rebalance band than equities.

As well, each asset class has a different "expected return". Over the long run, you would expect to move funds from the higher performing asset class (equities) to the lower performing asset class (bonds) by rebalancing in general (using either calendar or opportunistic rebalancing). This reduces volatility but "dilutes" return.

I think that there must be some better way to incorporate both the expected return and volatility of individual asset classes to optimize the benefits of rebalancing. Maybe by pairing asset classes with similar expected long term returns (i.e. commodities and reits)?

bawelch99
Posts: 5
Joined: Tue Mar 25, 2008 11:58 am

Post by bawelch99 » Tue May 20, 2008 2:30 pm

Good questions which could make for interesting Q&A on tomorrow's call. I know Gobind would acknowledge that there's a lot of room for further research around tactical questions like these.

leary
Posts: 57
Joined: Thu Aug 16, 2007 6:26 pm

Oportunistic rebalancing

Post by leary » Fri May 23, 2008 12:56 pm

I have a question about the tolerance bands in the rebalanceing process. When an asset class or index fund gets 20% out of balance it is time to rebalance regardless of whether it has been 1 year or 1 month since the last rebalancing. That is clear. But instead of rebalancing back to the target level of the asset allocation the auther recomends using tolerance bands and only rebalancing back part way.

My question is why not rebalance back to the target allocation? It does keep trading costs down to use tolerance bands, but if the cost of the trades were not a factor would it be better to do a complete rebalance back to the target allocation? At any one time one never knows whether momentum will continue or not. Maybe now is the time it will revert to the mean. Not continue in its present direction. Who can kmow? Thanks.

Charles

Rodc
Posts: 13601
Joined: Tue Jun 26, 2007 9:46 am

Post by Rodc » Fri May 23, 2008 2:53 pm

That is a better article on rebalancing than most. The author looks at both return and risk for different strategies.

Since his strategy is basically what I use, of course I like it. :)

But in the end like all such studies I have read, it is worth noting the differences between variations of rebalancing are pretty small.

And the study is dominated by the HUGE tech bubble and its big POP!, and so who knows if the next 10 years, 20 years will look much like this time period.

So, my take away again is it does not matter much which of the many rational schemes you use (although I can't help but be pleased my method may be a teeny bit better than some others. :))
We live a world with knowledge of the future markets has less than one significant figure. And people will still and always demand answers to three significant digits.

bawelch99
Posts: 5
Joined: Tue Mar 25, 2008 11:58 am

Post by bawelch99 » Fri May 23, 2008 4:27 pm

Leary wrote: "I have a question about the tolerance bands..."

Trading costs are one aspect, as you mentioned. Momentum is also important and usually much more so. The argument isn't necessarily that it's preferable to have the class that you're correcting at a tolerance ("noise") band rather than at target. What's more important are the other trades that potentially result from pushing the rebalanced class all the way to target. In practice, if you push it all the way to target, you often wind up being forced to correct many, if not all, of the other (in band) classes in order to balance out the trades.

Although you can't necessarily predict when momentum swings begin and end, you know they exist. Without a tolerance band, trades in one out-of-band class more often force you to clip other classes that may be on momentum swings. Based on the research in the paper, this can cost you significant rebalancing benefits in the long run.

leary
Posts: 57
Joined: Thu Aug 16, 2007 6:26 pm

Post by leary » Sat May 24, 2008 11:24 am

If one had an asset class such as emerging markets set as 5% of a portfolio and wanted to set a 20% band to rebalance in would that mean that you would rebalance when that class rose to 6% or fell to 4% of the portfolio? Just want to see if I have this right. Thanks.

Charles

User avatar
tetractys
Posts: 4830
Joined: Sat Mar 17, 2007 3:30 pm
Location: Along the Salish Sea

Post by tetractys » Sat May 24, 2008 11:54 am

leary wrote:If one had an asset class such as emerging markets set as 5% of a portfolio and wanted to set a 20% band to rebalance in would that mean that you would rebalance when that class rose to 6% or fell to 4% of the portfolio?
Yes leary, because 1% is 20% of 5%. But you might want to use a larger band of say 25%, or even 30% since emerging markets are a volatile asset class. And then to follow the strategic or opportunistic rebalancing, and taking all practical matters into consideration, like taxes and costs, you would put that money into the asset class or classes that has the least correlation.

"New paradigm"???, give me a break! -- Tet
RESISTANCE IS FRUITFUL

Easy Rhino
Posts: 3267
Joined: Sun Aug 05, 2007 11:13 am
Location: San Diego

Post by Easy Rhino » Sat May 24, 2008 12:18 pm

You know, the one thing that kind of confuses me about the use of tolerance bands, is what to do when one asset leaves the tolerance band, but the rest are still within it?

For instance, if my portfolio was 60 stock, 30 bond, and 10 commodity. And commodies goes to over 12%, then I'd probably want to rebalance. However, the stock and bond portions would still be well within their own tolerance bands, so do I just rebalance to the one that is the least on target?

User avatar
tetractys
Posts: 4830
Joined: Sat Mar 17, 2007 3:30 pm
Location: Along the Salish Sea

Post by tetractys » Sat May 24, 2008 12:55 pm

Easy Rhino wrote:... if my portfolio was 60 stock, 30 bond, and 10 commodity. And commodies goes to over 12%, then I'd probably want to rebalance. However, the stock and bond portions would still be well within their own tolerance bands, so do I just rebalance to the one that is the least on target?
Yes if your band is %20; and look for the assets most out of correlation. -- Tet
RESISTANCE IS FRUITFUL

leary
Posts: 57
Joined: Thu Aug 16, 2007 6:26 pm

Post by leary » Sat May 24, 2008 2:47 pm

Thanks for the response. I thought that was the way to get 20% bands, but wanted to be sure. I have always rebalanced yearly, but this is interesting. One thing that made me pause is that the author is a managing director of TDAmeratrade. I had always thought of them as a trading oriented discount broker but the author seemed to be talking about the effiencent frontier and MPT with index funds. I went to the TDAmertrade web site and they seem to offer a folio which buys 15 ETF's within the folio for your portfolio. All index funds(mainly I-Shares and some Vanguard & a few Claymore) They say the funds are all selected to match MPT and the effiecent frontier, but if one does not like their funds you can pick your own. They buy the 15 funds and let you rebalance as often as you want (as often as every 5 days) They have a spread sheat showing how far each fund is out of balance which you can check each day. And as I said rebalance every 5 days if you want. There is no transaction cost for buying the ETF's to set up the account and no transaction cost to rebalance as often as you want. The only cost is .75 basis points PER YEAR of the net asset value of the account. Collected in 4 quarterly installments. It looked really interesting. They recommend it mainly for retirement accounts. (IRA's & rollover IRA's)

My wife & my retirement account is in Vanguard Index Funds now and has been for years. Would like to hear others opinion on this.

Charles

livesoft
Posts: 72087
Joined: Thu Mar 01, 2007 8:00 pm

Post by livesoft » Sat May 24, 2008 3:09 pm

leary wrote:....The only cost is .75 basis points PER YEAR of the net asset value of the account. Collected in 4 quarterly installments. It looked really interesting. They recommend it mainly for retirement accounts. (IRA's & rollover IRA's)

My wife & my retirement account is in Vanguard Index Funds now and has been for years. Would like to hear others opinion on this.

Charles
I see no reason to pay for something that might gain you 0.5% a year and it's even something that you can do yourself! The fees is a sliding scale (0.75% of first $100K or 2.95% or $150 whichever is less; 0.5% for next $400K, and 0.35% for additional assets over $500K).

Full disclosure: I have a bought and I'm holding account at TDAmeritrade. I do my rebalancing outside of this account, so I pay no commissions or fees to TDAmeritrade to have them hold my ETFs and stocks.

One could implement the same plan without any extra fees in a WellsFargo PMA relationship that gives you 100 free trades a year. That would save you $4500 on your first million. $4500 is lot to pay for an extra 30 minutes of work a year.

Buddtholomew
Posts: 999
Joined: Thu Mar 01, 2007 3:29 pm

Post by Buddtholomew » Sat May 24, 2008 6:53 pm

Can someone distinguish between opportunistic rebalancing to target allocation and market timing? The main distinction I see is that oportunistic rebalancing relies upon +/- % changes from target allocations and market timing uses +/- % changes in an asset class.

What would you recommend in situations when an asset class is up 20% and is still within your target allocation? It seems that we should take advantage of this scenario as asset classes revert towards the mean...

leary
Posts: 57
Joined: Thu Aug 16, 2007 6:26 pm

Post by leary » Sun May 25, 2008 11:15 am

If an asset class is up 20% and still within the band it means that the other classes must be up also. Otherwise it would not be within its band. But that should not be true in a well diversified portfolio. All the asset classes should not be moving together in the same direction.

Charles

Buddtholomew
Posts: 999
Joined: Thu Mar 01, 2007 3:29 pm

Post by Buddtholomew » Mon May 26, 2008 10:26 am

leary wrote:If an asset class is up 20% and still within the band it means that the other classes must be up also. Otherwise it would not be within its band. But that should not be true in a well diversified portfolio. All the asset classes should not be moving together in the same direction.

Charles
I rebalance to target allocations with new contributions. Also, my rebalancing bands are set at different percentages depending on volatility in that specific asset class.

diasurfer
Posts: 1839
Joined: Fri Jul 06, 2007 8:33 pm
Location: miami-dade

Post by diasurfer » Mon May 26, 2008 11:08 am

livesoft wrote:
leary wrote:....The only cost is .75 basis points PER YEAR of the net asset value of the account. Collected in 4 quarterly installments. It looked really interesting. They recommend it mainly for retirement accounts. (IRA's & rollover IRA's)

My wife & my retirement account is in Vanguard Index Funds now and has been for years. Would like to hear others opinion on this.

Charles
I see no reason to pay for something that might gain you 0.5% a year and it's even something that you can do yourself! The fees is a sliding scale (0.75% of first $100K or 2.95% or $150 whichever is less; 0.5% for next $400K, and 0.35% for additional assets over $500K).

Full disclosure: I have a bought and I'm holding account at TDAmeritrade. I do my rebalancing outside of this account, so I pay no commissions or fees to TDAmeritrade to have them hold my ETFs and stocks.

One could implement the same plan without any extra fees in a WellsFargo PMA relationship that gives you 100 free trades a year. That would save you $4500 on your first million. $4500 is lot to pay for an extra 30 minutes of work a year.
30 minutes a year? To rebalance a portfolio of 10-15 ETFs every 5 days? Livesoft, I've studied your posts on buying ETFs, as I will be buying ETFs (on dips!) with Wells Trade within a few days now. For me it is the cheapest way to additional asset classes outside of VG. But I certainly expect it will take more than 30 minutes a year even for my relatively simple portfolio.

livesoft
Posts: 72087
Joined: Thu Mar 01, 2007 8:00 pm

Post by livesoft » Mon May 26, 2008 11:14 am

diasurfer wrote:30 minutes a year? To rebalance a portfolio of 10-15 ETFs every 5 days? Livesoft, I've studied your posts on buying ETFs, as I will be buying ETFs (on dips!) with Wells Trade within a few days now. For me it is the cheapest way to additional asset classes outside of VG. But I certainly expect it will take more than 30 minutes a year even for my relatively simple portfolio.
You must've misread the opportunistic rebalancing paper. You LOOK every 5 days or every day or every hour :twisted: , but you only rebalance rarely when you are OUT OF RANGE. It takes only 5 seconds to look. It takes BIG market movements to go out of range. :)

ETA: Also note that I used the word "extra" in "... extra 30 minutes a year ...". Presumably whether you use the TDAmeritrade stuff or something else, you have to look before you click on the rebalance button, so looking doesn't count towards your extra 30 minutes a year.
Last edited by livesoft on Mon May 26, 2008 11:37 am, edited 3 times in total.

diasurfer
Posts: 1839
Joined: Fri Jul 06, 2007 8:33 pm
Location: miami-dade

Post by diasurfer » Mon May 26, 2008 11:24 am

livesoft wrote:It takes BIG market movements to go out of range. :)
Movements needed for 5/25 to trigger:

http://www.diehards.org/forum/viewtopic ... highlight=

User avatar
jeffyscott
Posts: 8866
Joined: Tue Feb 27, 2007 9:12 am
Location: Wisconsin

Post by jeffyscott » Mon May 26, 2008 5:12 pm

livesoft wrote:You LOOK every 5 days or every day or every hour :twisted: , but you only rebalance rarely when you are OUT OF RANGE. It takes only 5 seconds to look.
You can even automate the looking to a great extent. I have a bunch of yahoo alerts set to do my looking on a continuous basis.
Time is your friend; impulse is your enemy. - John C. Bogle

Post Reply