Folio Management | Optimal Rebalancing [Selling/Exit] Frequency - How Do We Conclude?

2nd Jan '16 by mpat89

4 comments 5608 reads

A thread to discuss the topic of rebalancing but in particular taking off the stocks that have fallen out of the top decile to be replaced with those currently in the top decile.

When I first joined this site about 18 months ago I was selling stocks the very day/week their ranks changed. I realised how ridiculous this was and it became quarterly. Results in quarterly exits have been ok, but I notice the older portfolio that I left for 11 months without rebalancing did much better over the same period.

Ignoring transaction costs for a minute, the issue becomes more about this rotation out of more volatile smaller cap stocks that Ed discussed in his webinar. i.e. you can inadvertently sell out of a micro cap that has swung down quite a bit and straight into another micro cap that happens to do the same thing after you enter -- less risk of this with longer rebalancing or by targeting larger caps.

So is there any conclusive evidence or arguments that suggest what we should be doing here?

Maybe a tiered system, i.e. large caps can be exited any quarter, mid caps can be exited provided held for 6 months, small/micro caps can be exited provided held for 12 months. They don't necessarily have to be recycled into stocks of the same size though.

With regards to transaction costs, I don't really see the need to discuss them in this thread as I have found plenty of resource on the issue on this site.

Similarly for spread, as presumably most of us have been taught by now about the impact of this as well as shown how to avoid trading stocks with high spreads.

Contributions welcomed.

Unlock with Email

Unlock with Google

Already have an account?
Login here

Noodle Hat

2nd Jan '16

Obviously it depends on your overall strategy etc. However from what i have read/ researched in beginning my investment journey and from what Ed has written regarding the Naps, I think once a year is really the most frequent you should re balance, unless say you have trailing stop loss alerts etc ( not too close that general volatility triggers it thou! i giuess these should be set by looking at historically how volatile the stock is?) . Quarterly is far too often and in my mind is almost like day trading!

I know Greenblatt suggests once a year for tax reason but suggests one of the reasons the Magic Formula doesn't work every year is because it sometimes takes 1-3 years for a stock to be fully valued by the market etc. A glance at some guru screeners show stocks are still on them that appeared over a year ago, wither this is because the stock is still a good buy, or was never a good buy is another matter!?

pka

3rd Jan '16

"With regards to transaction costs, I don't really see the need to discuss them in this thread as I have found plenty of resource on the issue on this site. Similarly for spread, as presumably most of us have been taught by now about the impact of this as well as shown how to avoid trading stocks with high spreads."

Actually spreads and transaction costs are crucially important in deciding on a rebalancing frequency. If we look at the past performance graphs on this website here:

http://www.stockopedia.com/stockranks/performance/

We would find that the cumulative performances of UK stocks > £10 miliion market capitalisation with QVM StockRanks in the 90-100 band rebalanced quarterly from 20 April 2013 to 31 December 2015 was +73.78%. The figure for the same stocks rebalanced annually was +68.04%. So there was a cumulative outperformance due to quarterly rebalancing of +5.7% over that 2.7 year performance history. However, those figures ignore spreads and transaction costs. (They also ignore reinvested dividends, but that is not relevant to this discussion.) Let us assume that the average bid-offer spread is 1% for all those stocks (which is an optimistic assumption for smaller companies), that one pays £10 stockbroker's commission for each trade, and that the average holding of each stock in one's portfolio is £5,000 (smaller holdings than that would mean higher average commissions when expressed as a percentage of one's average holding) . If ALL the stocks in one's portfolio are replaced at each rebalancing, then after each rebalancing one will lose about 1% (due to the bid-offer spread), plus 0.2% (= 100% x £10 / £5,000) stockbroker's commission on each sale and another 0.2% on each purchase, plus 0.5% Stamp Duty on each purchase of a UK stock, making a total loss at each complete rebalancing of about -1.9%. This is equivalent to a multiplying factor of 0.981 on the total portfolio value after each complete rebalancing. With quarterly rebalancing there are 3 more rebalancing events each year than with annual rebalancing, so the relative total multiplying factor of quarterly rebalancing compared with annual rebalancing is 0.981 x 0.981 x 0.981 = 0.944, which represents a relative annual loss from quarterly rebalancing of -5.6%! Over the 2.7 year performance history of the StockRanks, there would have been 10 rebalancing events with quarterly rebalancing and just 2 with annual rebalancing (i.e. 8 less), so the relative multiplying factor over that 2.7 year period would have been 0.981 to the power of 8 which is 0.8577 or a relative total loss of -14.3%! That swamps the relative outperformance of +5.7% ignoring transaction costs over that period. To be mathematically correct, one should multiply the factor 0.8577 due to cumulative transaction costs by the 1.057 multiplying factor due to the 5.7% outperformance without transaction costs, which gives a combined factor of 0.907 or a relative cumulative loss of -9.3% due to quarterly rebalancing compared with annual rebalancing!

Of course, the figures above assume that one replaces ALL the stocks in the portfolio at each rebalancing. If one only replaced those stocks whose StockRanks had fallen by a significant amount at each rebalancing, then the cumulative transaction costs would be significantly less, but that is true whatever is the frequency of rebalancing.

Therefore from the above figures, I suggest that one should not do a complete rebalance of one's portfolio more often than once a year.

I've just realised that I made a mistake in my previous thread in the figure for the cumulative outperformance WITHOUT transaction costs of quarterly rebalancing over annual rebalancing a 90-100 StockRanks portfolio from 20 April 2013 to 31 December 2015 . I incorrectly wrote the figure was +5.74%, which I obtained by subtracting +68.04% from +73.78%. What I should have done was to divide 1.7378 by 1.6804, which gives a figure of 1.034 or +3.4% over that 2.7 year period.

That means I should have multiplied the factor 0.8577 due to cumulative transaction costs by the 1.034 multiplying factor due to the 3.4% outperformance without transaction costs, which gives a combined factor of 0.887 or a relative cumulative loss of about -11.3% due to quarterly rebalancing compared with annual rebalancing!

mpat89

Yep I didn't say they weren't crucial just that they don't need to be discussed because there is already plenty of info about those issues on this site. Just looking to isolate the factors discussed in the original post.

Optimal Rebalancing [Selling/Exit] Frequency - how do we conclude?

Unlock the rest of this article with a 14 day trial

About the Author

Mpat89

4 comments

Unlock the metrics for all these stocks and more...

Unlock the metrics for all these stocks and more...