PP Performance As A Range (Due to Uncertainty in Rebalancing Events)

[TripleB]

I frequently see the PP being touted as doing X% YTD or Y% for the last year. However, that assumes one had a perfect 4x25 split on the starting point of that metric.

Since it's possible a person was 16/16/34/34, and didn't hit a rebalance band, and from this we can extrapolate they were 16/16/34/34 such that the Best Two performing assets were 34s while the worst 2 performing assets were 16s OR - Vice Versa and they had 34% of the worst two assets and only 16% of the best two.

I ran some numbers from January 3rd 2012 until the close of the market on Friday, using Yahoo Finance "Adjusted" Close price on Jan 3rd, which purportedly re-adds back in dividends (or technically subtracts, so that the correct gain is calculated by subtracting Today's Value from the Adjusted Value on Jan 3).

I found the PP returned 5.3% over this approximate 1 year time period. The best two assets were Stocks and Bonds.

If someone were 34/34/16/16 with the 34s being in Stocks and Bonds (the best two performers), the total return is 6.6% which is 1.3% higher than the return of a 4x25 portfolio on the start date.

If someone were 16/16/34/34 their return is 1.3% lower, which would be 4%. Thus, the total spread is 2.6% over the last year. This ignores the possibility of hitting any rebalancing bands during the year because that's a significantly more complex calculation.

If I were better at statistics, I could calculate the likelihood of where you'd fall in that 2.6% spread, however because I have an MBA and not a PhD in Stat, I'll just estimate it and say you're probably most likely to be within 0.75% and only in extreme cases wind up towards that 1.3%.

In other words, it doesn't really matter, or at least it didn't matter much during 2012. As the spread between individual assets widens, it will make a bigger difference, however as that spread widens, if you were 34/34/16/16 then you'd hit a rebalancing event mid-year.

TL;DR - TripleB had a hypothesis, ran some round numbers, and determined further investigation is not warranted.

[TripleB]

ha. if you were going to put TL;DR in there it should have been at the top, not the bottom.
by the time I saw that I'd already read the whole post 

That would be an executive summary, not a TL;DR

[Pointedstick]

I'd probably rebalance if I was at 16/16/34/34. Close enough for me!

[edsanville]

Interesting that the best/worst case scenarios only differ by 2.6%.  I wonder what the spread is for other years, particularly the 2 or 3 down years, when the losing asset dropped like a rock.

[mmaurice]

Isn't the rule to rebalance once a year back to 4x25 and then any time when any asset hits an upper or lower band?

[rickb]

Isn't the rule to rebalance once a year back to 4x25 and then any time when any asset hits an upper or lower band?

The completely orthodox method is to rebalance only when the bands are hit.  This tends to reduce the frequency of rebalancing, minimizing taxable events and trading costs.  Whether this has performed better or worse in the past than annual rebalancing plus band-based rebalancing depends on what interval you backtest - and going forward it's simply a crapshoot.  Although it definitely sounds like a cop out, I think the bottom line is either way is fine and you should do what is most comfortable for you.