Page 1 of 2

Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 1:36 am
by metta2006
I'm planning to keep adding money to pp. Will I not have to sell outperforming assets if I can rebalance by buying the worst performing asset? Would that decrease return as I never sell. I'm tempted to sell some stocks to capture gain. Thanks!

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 1:39 am
by craigr
I would buy the lowest performing asset and not sell the winners myself. For one, if these are taxable funds you will pay extra in taxes you can avoid. Secondly, there could be transaction fees when you move in and out of assets (commissions, bid/ask spreads, etc.). Finally, there could be some momentum in the assets you are sacrificing if you are constantly whittling them down instead of leaving them be until they hit an upper band.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 1:53 am
by Pointedstick
Seems like you're asking if you can avoid rebalance bands by always buying the lagging asset. I believe Sophie did an extensive backtest one time and determined that buying the lagging asset produced slightly lower returns compared to buying all four equally and letting them hit their rebalance bands, but better returns than putting new contributions in cash and waiting to hit rebalance bands.

I say do it if it makes you comfortable, especially in a taxable account where Craig is absolutely right that increased selling can generate more taxes, which could wipe out the advantage compared to buying the lagging asset.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 12:26 pm
by dragoncar
Yes historically buying the lagging asset underperformed other strategies.  I suspect this is because (investment mumbo jumbo) the volatility timeframe was long / momentum effects were strong.  The question is: will this continue in the future?  If going forward there is a lot of short term volatility but, for example, little drift and long term fluctuation, then buying the lagging asset (which is conceptually similar to more frequent rebalancing) could be a winner.  Since I can't predict the future like this, I just do what is easiest and
Buy the lagging asset.  I have never hit a rebalance band (in about two years).

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 1:08 pm
by Xan
I still prefer the truly neutral approach of buying the assets in their current proportions.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 1:38 pm
by dragoncar
Xan wrote: I still prefer the truly neutral approach of buying the assets in their current proportions.
How is that neutral?  It predicts that momentum will continue.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 1:51 pm
by Xan
I disagree; doing it this way isn't a prediction of anything.  With any other scheme, the amount and frequency of your contributions affects the performance of your portfolio.  This way it doesn't.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 3:15 pm
by dragoncar
Xan wrote: I disagree; doing it this way isn't a prediction of anything.  With any other scheme, the amount and frequency of your contributions affects the performance of your portfolio.  This way it doesn't.
Wut?  The amount and frequency of contributions affects performance either way.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 3:45 pm
by Xan
...how?  If, when I buy, I buy in proportion to what's already there, then the performance isn't affected at all.  Neither is the timing of any rebalancing.

I believe that my contributions should be disconnected completely from the rebalancing timing and from the performance (or else I'm engaging in market timing).  That's why I do it this way.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 4:02 pm
by Tyler
No need to over think it.

I personally simply buy the lagging asset(s) for a few reasons - both practical and psychological.

Practical : it's the easiest way to avoid rebalancing events (important in my taxable account)
Psychological : always buying low is good mental exercise for me, and keeping the assets at 25% weightings reduces volatility

Perhaps I could make slightly more money with a different scheme, but at some point you have to stop splitting hairs and move on to bigger problems.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 5:08 pm
by BearBones
Sounds like most people ignore Sophie's data. As I recall the difference was not trivial.

Want to create a poll? It may be like my gold insurance poll. Everyone suggests buying insurance, but very few actually do it.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 6:51 pm
by dragoncar
BearBones wrote: Sounds like most people ignore Sophie's data. As I recall the difference was not trivial.

Want to create a poll? It may be like my gold insurance poll. Everyone suggests buying insurance, but very few actually do it.
It's an interesting area for exploration but as I recall Sophie provided results rather than data.  In others words it was not clear to me if outperformance was due to picking a particular entry date (were multiple entry dates used?).  How many rebalances were there and what would the tax implications be?  Did it affect volatility?

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 7:19 pm
by Pat
For additional money, I remember Harry Browne's rule is: contribute to cash portion and rebalance into other asset classes when the cash portion reach 35% of the entire portfolio.

For me, this rule is simple to follow and I don't need to time the market (according to my experience, my prediction on the market always goes wrong - expensive asset can become more expensive and underperformed asset can continue to go bad!).

Any comments on my interpretation of HB's rules? Thanks.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Fri Apr 12, 2013 10:18 pm
by sophie
dragoncar wrote:
BearBones wrote: Sounds like most people ignore Sophie's data. As I recall the difference was not trivial.

Want to create a poll? It may be like my gold insurance poll. Everyone suggests buying insurance, but very few actually do it.
It's an interesting area for exploration but as I recall Sophie provided results rather than data.  In others words it was not clear to me if outperformance was due to picking a particular entry date (were multiple entry dates used?).  How many rebalances were there and what would the tax implications be?  Did it affect volatility?
It was overall (CAGR) returns for the time period of the test, and yes the differences were significant.  If I have time in the AM (not certain of that) will try to find the actual plots.

I rather like buying assets in equal proportions, as it is the most agnostic method in addition to best performing, but it's probably most realistic to mix different strategies.  Absolutely it makes sense to avoid rebalancing a taxable account by buying lagging assets.  Since physical gold is taxable by definition and it's kind of a pain to buy, I've been settling into a lagging asset strategy for that.  And it's not unreasonable to let extra savings accumulate in cash, and leave it there until rebalance time.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Sat Apr 13, 2013 6:48 am
by BearBones
sophie wrote: I rather like buying assets in equal proportions, as it is the most agnostic method
Less agnostic than Xan's method of buying in current proportions. Agree? As I recall, you did not look at Xan's method.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Sat Apr 13, 2013 9:17 am
by sophie
BearBones wrote:
sophie wrote: I rather like buying assets in equal proportions, as it is the most agnostic method
Less agnostic than Xan's method of buying in current proportions. Agree? As I recall, you did not look at Xan's method.
True on both counts.  Will add to todo list.

I was thinking a bit more about the tax issue.  The difference between buying the lagging asset & buying assets in equal proportions (which I called DCA) was $1829, or 2.7% in my test example.  Let's say you end up rebalancing at the end of the test period (my original post quoted below), selling 1/3 of your stocks, and that your capital gains tax rate is 25% including state.  That means you'd sell $5234.7 of your stocks.  If half of that is gains (a wild overestimate!), then your tax would be $654.34.  In the example below, there were two rebalances, so if we assume you get hit with a similar tax bill each time (which would be an overestimate also because the portfolio size would be less), you'd still come out ahead using the DCA method.  The tax impact would be correspondingly less if you can do at least part of your rebalancing in a tax-advantaged account.

However: note that the difference between the methods will decrease over time, as the size of your contributions relative to the total portfolio decreases.  So there's likely a point where a portfolio gets large enough you're better off with the lagging asset method, especially if you are in a high tax bracket.  I have to say though, if you are in a high tax bracket and have a large portfolio, I should hope that a good chunk of it is in a tax-advantaged account!
start date 1/1/2006, end date 4/1/2012
start with $10,000 portfolio, add $500/month
transaction costs:  $7.95/trade for gold (stock and bond asset trades cost me nothing in either taxable or tax-deferred)
25x4 portfolio, yahoo finance data for TLT used for bonds.  For gold & stock I used data from a gov't online database (blanking on the name right now).  For cash, I used average 6 month CD yields from that same database.  Dividends and interest were included in the simulation.

1) Dollar-cost averaging:  2 rebalance events triggered by the 15/35 bands (late 2008 and 2011).  End balance $69,796.
2) Buy lagging asset:  No rebalance events.  End balance $67,967.
3) Accumulate contributions in cash portion:  8 rebalance events (all out of cash).  End balance $66,682

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Sat Apr 13, 2013 9:27 am
by Bean
I run my PP for my taxable accounts.  So for the PP I always by lagging asset, to avoid taxes.  Also, I have set it up so I have zero commissions with the exception of buying physical gold and its premium.

I think you just have decide what is best for you with the following factors; re-balancing, taxes, fees, always having a lot of cash for emergency, or maximum back tested efficiency,

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Sat Apr 13, 2013 9:47 am
by Tyler
Thanks for the info, Sophie. Very informative.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Sat Apr 13, 2013 4:00 pm
by Kriegsspiel
Sophie, what was the explanation for the better performance of buying all of the assets equally, instead of the lagging one? 

Maybe one reason is that, if you add to the lagging asset, you're preventing yourself from hitting the upper re-balancing band/selling at the highs.  So maybe at the end of the time period you tested, the assets were balancing out, and the lagging asset that you'd been shoveling money into was about to rocket upwards as the other that had been high (and you didn't sell, since you were contributing to the lower asset) is on it's way down?

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Sun Apr 14, 2013 7:57 am
by sophie
Kriegsspiel wrote: Sophie, what was the explanation for the better performance of buying all of the assets equally, instead of the lagging one? 
I played a bit with the test runs at the time to try to figure that out.  I think it's partly due to momentum effects.  Consistently buying into assets that are increasing, as opposed to leaving them alone, helps juice the returns.  With lagging assets you are preferentially buying assets that may be on their way down (like gold now).  There's probably also a benefit to rebalancing (e.g. that paper MG posted with melveyr's excellent Bernoulli RV simulation, where rebalance events turned a zero expected outcome into steady gains).  I wonder what would happen in that simulation if lagging asset contributions were added.

This made me like the PP more and more.  Just set up automatic investments and have the discipline to buy gold regularly regardless of the price, and you get free momentum effects without having to compute 10 month moving averages!

Glad you guys like the avatar.  Muppets are a delightfully snarky bunch, which maybe unintentionally describes us pretty well!

On an unrelated note, I haven't seen any posts from Mom2Boys for a while.  Can one of the moderators check that she hasn't been blocked again?

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Sun Apr 14, 2013 10:10 am
by Kriegsspiel
sophie wrote: I played a bit with the test runs at the time to try to figure that out.  I think it's partly due to momentum effects.  Consistently buying into assets that are increasing, as opposed to leaving them alone, helps juice the returns.  With lagging assets you are preferentially buying assets that may be on their way down (like gold now).
It still doesn't make sense.  If you buy your lagging asset on its way down, you start out getting a "better price" relative to your other investments, and the further it goes down, the better the price will be.  Since our belief is that the 4 investments will produce a profit, isn't buying MORE of them when you can get them for a cheaper price a better way to juice returns than buying into one who's momentum is already being produced by market forces?

Another point about adding to a rocketing asset, is that it will lead you to hit a re-balancing band sooner.  The advantage to hitting a re-balancing band is that you lock in the profits from the upper asset, and then you use that profit to bring up the lower asset.  With adding to the lagging asset, you only miss out on the profits from the upper asset. 

So the upper asset may plateau at the levels it reached, waiting on you to contribute enough to your other lagging assets to offset it, or it may come down to the point where the assets are balanced, and you have "missed out" on capturing those gains, but you have also not added any money that you didn't capture.  Of course, it could also drop from the upper re-balancing band to become the new lagging asset.

This looks like it becomes more of a moot point when the portfolio gets large enough that monthly paychecks don't really carry enough weight to swing percentages substantially.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Sun Apr 14, 2013 11:23 am
by annieB
I like Xan's approach of buying at current percentages.
Otherwise,we're throwing out our belief in the rebalance bands?
Buying the trailing asset when we add money and then when we rebalance is
picking lows twice?I think MG said that profits come from rebllancing out of the gainers,not the losers.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Sun Apr 14, 2013 11:45 am
by AgAuMoney
Kriegsspiel wrote: This looks like it becomes more of a moot point when the portfolio gets large enough that monthly paychecks don't really carry enough weight to swing percentages substantially.
That's my take.

If your regular contributions to the account don't significantly change the percentages, and definitely by the time your entire annual addition would not cause cash to hit a rebalance band unless the other assets drop significantly, it seems there is little difference in any approach.

This is because no matter which asset(s) you add to, it isn't going to significantly change it.

For example, say your entire annual contrib will be 1% of the cash balance (taking it from an ideal 25% to 26%).  If you have an asset significantly lagging, that 1% won't bring it up.  If you have an asset significantly leading, adding 1% isn't going to add significantly to your return.  And no matter how you divide it amongst the four classes the effect will always be smaller than putting it entirely into one class.

And if someone wants to do some pretty pictures to illustrate the concept, that would be great.

Probably most helpful for the forum is if someone could figure the crossover point where monthly and/or annual contributions as a percentage of the portfolio lose their significance.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Sun Apr 14, 2013 1:05 pm
by Kriegsspiel
annieB wrote: I like Xan's approach of buying at current percentages.
Otherwise,we're throwing out our belief in the rebalance bands?
Buying the trailing asset when we add money and then when we rebalance is
picking lows twice?I think MG said that profits come from rebllancing out of the gainers,not the losers.
If you are buying at the current percentages (say, 29% stocks, 27% bonds, 20% gold, 24% cash), and you put $1000 into the PP a month, you'd be buying $290, $270, $200, and $240 into a "monthly unit" of each.  But why would you buy a monthly unit of the stocks or bonds?  From the point where they were all in balance with each other, you get less stocks and bonds for your monthly unit than you used to, but you get more gold, it just doesn't make sense to buy in proportion to the assets, because it's like you're buying MORE of the thing that you get teh LEAST of for your money.

Re: Do I not need to sell outperforming assets if I add money to pp

Posted: Sun Apr 14, 2013 2:01 pm
by Xan
But that's just an argument for constantly rebalancing.  If you want to constantly rebalance, then go for it.  I believe in the rebalancing bands, and so I don't want my contributions to have any effect on when and whether I rebalance.