Grahamites

# Why It's Almost Always a Mistake to Sell Compounding Machines

## Math shows it is better to hold true compounding machines than dancing in and out every few years

Chuck Akre (Trades, Portfolio) was the keynote speaker of this year’s GuruFocus Value Conference. During the Q&A session, somebody asked Akre what investor should do if the valuation of a great compounding machine becomes excessively high. Akre’s answer inspired me to write this belated article. In his experiences, it has almost always been a mistake to sell the great compounding machines, even when the valuation seems high.

Why is that?

Everybody may come up with different answers. I’m not sure I have found the perfect answer but I did a little exercise myself, which showed mathematically and logically why it makes sense to hold the truly great compounders for a long time.

In this simple exercise, I calculated the after-tax compounded rate of return under a few different scenarios using the price, valuation and fundamental data of Markel (NYSE:MKL) for the past 15 years. Undoubtedly, there are limitations and drawbacks of the study. However, I think it provides me with a good mathematical framework to figure out the issue at hand.

Scenario 1: Buy Markel on the last day of 2002 and hold it until the last day of 2017 (buy and hold for 15 years).

Scenario 2: Buy Markel on the last day of 2002, sell at the exact high of 2007. Buy again at the exact low of 2009, and sell it at the exact high of 2017 (perfect timing).

Scenario 3: Buy Markel on the last day of 2002 and trade in and out every five years by selling at 10% below the high price of the fifth year and using the proceeds to buy Markel at 10% above the low price of the next year. The transactions under this scenario are as follows:

• Buy Markel on Dec. 31, 2002, at \$205.50 per share.
• Sell Markel at 10% below the 2007 high of \$545.50, or \$491 per share, and pay long-term capital gain of 15%.
• Use the net proceeds to buy Markel at 10% above the low of 2008 of \$245.30, or 269.80 per share.
• Sell Markel 10% below the 2012 high of \$500.70, or \$460.60 per share, and pay long-term capital gain of 15%.
• Use the net proceeds to buy Markel at 10% above the low of 2013 of \$433.40, or 476.80 per share.
• Sell Markel 10% below the 2017 high of \$1147.10, or \$1032.40 per share, and pay long-term capital gain of 15%.

Scenario 4: Buy Markel on the last day of 2002 and trade in and out every five years by selling at 15% below the high price, using the proceeds to buy Markel at 15% above the low price of the next year.

The transactions under this scenario are as follows:

• Buy Markel on Dec. 31, 2002, at \$205.50 per share.
• Sell Markel at 15% below the 2007 high of \$545.50, or \$463.70 per share, and pay long-term capital gain of 15%.
• Use the net proceeds to buy Markel at 15% above the low of 2008 of \$245.30, or \$282 per share.
• Sell Markel 15% below the 2012 high of \$500.7, or \$425.6 per share, and pay long term capital gain of 15%.
• Use the net proceeds to buy Markel at 15% above the low of 2013 of \$433.40, or 498.40 per share.
• Sell Markel 15% below the 2017 high of \$1147.10, or \$975 per share, and pay long-term capital gain of 15%.

I will go straight to the results:

Scenario 1: 12.1% CAGR.

Scenario 2: 17.8% CAGR.

Scenario 3: 13.8% CAGR.

Scenario 4: 12.05% CAGR.

Obviously if you dance in and out at the exact low and high of 2009 and 2017, you had better to that. But we know that is almost impossible.

If you are very good at timing the low and high, as in Scenario 3, you will also beat buy and hold.

If you are relatively good at timing the low and high, as in Scenario 4, you are better off not dancing in and out.

You may think that it looks like it’s worth trying to time the low and high.

There is one major caveat: There is no dependable yardstick to tie the low and high to a specific valuation level. In other words, historic valuation level offers little guidance. Prior to 2007, the low price-book ratio for Markel was 1.38 times. Between 2008 and 2014, the low price-book ratio was between 0.91 and 0.99 times (except in the exactly low of 2009 when it got to 0.75 times). And during the recent two years it has been 1.3 times.

Similarly, the high price-book ratio was roughly 2.0, 1.2 and 1.65 during these three time periods. Again, there is no dependable yardstick, which is understandable given there are so many factors influencing short-term and medium-term valuation levels. Notice the high price-book ratio during the second period is lower than the low price-book ratio level during the first period.

What are the likely consequences if you apply a historic valuation method, which is commonly used among value investors? You may have bought Markel at 1.38 times book value during 2003 to 2007, only to see it go down to less than 1 time book. And then you wanted to buy it at one time book value during 2015 to 2017, only to miss out on the compounding.

And the psychological biases are not working in your favor. Once you have sold something at X, it is harder to buy it back at prices higher than X due to anchoring and contrast (how many investors sold Berkshire’s A shares in the early years for a few thousand dollars a share and never bought them back?).

In reality, most people may have done worse but because they may have played the valuation games multiple times during a 15-year period, somehow they may feel like they have achieved betters returns trading in and out every few years compared to buy and hold.

Considering all of the above, I think the odds for anybody to successfully play the valuation fluctuation game on Markel are incredibly unfavorable. However, if valuation reaches extreme levels either on the high end or the low end, it may make sense to either trim a little or buy a little more. By definition, extreme valuations are a rare occurrence.

Disclosure: Long Markel.

Grahamites
A global value investor constantly seeking to acquire worldly wisdom. My investment philosophy has been inspired by Warren Buffett, Charlie Munger, Howard Marks, Chuck Akre, Li Lu, Zhang Lei and Peter Lynch.

 Currently 5.00/512345 Rating: 5.0/5 (7 votes) Voters:

Stephenbaker - 1 year ago    Report SPAM

Generally agree with this but why not use basic valuation principles, such as buy at a cheap PB (or other valuation metric) and sell when the valuation gets excessive? The valuation metric may be a function of price but price, itself is not the defining variable as to when to buy or sell.

Dr. Paul Price - 1 year ago    Report SPAM

I think you have missed a salient point.

There is no mention of the ultimate LT cap. gains tax due in scenario one if you cash out.

That would nick the returns as it did for the other techniques. Sure, if you never sell you never pay tax... but then you'd never get to use or enjoy the money, making it worthless except to your heirs.

Shb600 - 1 year ago

Identifying the future compounding machines from the hundreds/thousands of stocks that only appear to be compunding machines is the the hard part. Not selling is easy. That's why there is one Buffett and millions of pretenders.

Thomas Macpherson - 1 year ago

Anchoring and confirmation bias can make it very difficult to admit that a long term and successful investment is no longer the compounding machine you thought it once was. It took me losing 25% of my profits and three go arounds with my DCF models, risk assessments, etc. before I could admit Ansys (ANSS) wasn’t the juggernaut I had invested in 13 years earlier. Just my \$.02 worth. Thanks for a great article. Best - Tom

Grahamites - 1 year ago    Report SPAM

Stephenbaker- Thanks for commenting. The issue with using valuation metrics such as P/B to trade in adn out, in my opinion is that there are so many factors affecting valuation metrics (interest rates, inflation rates, market conditions, investor preference, business cycles etc) that it's not practical to tell when valuations are extremely cheap or expensive. For instance, between 2002 and 2007, a P/B of 1.4 is considered cheap for Markel. But for the next few years, a P/B of 1.4 is actually "expensive" for Markel. The question then becomes, how do you consistently make accurate assessment of "cheap" and "expensive" multiples when there's no dependable yardstick?

Grahamites - 1 year ago    Report SPAM

Dr. Paul Price: You raise a valid point. I agree if one's goal is to enjoy the compound early on, then buy and hold may not meet the investment objective, in which case a different strategy may better serve one's need. With regards to LTCG, yes ultimately when you sell you have to pay the tax, but the longer you hold Markle, the less the LT Capital Gain affect your CAGR compared to a "trading strategy."

Grahamites - 1 year ago    Report SPAM

Shb600: Yes, identifying the compounders is much harder. As Tom pointed out, sometimes the business fundamentals will change and the compounding machine may stop. It's hard to recognize when that happens too. Thanks for raising this point.

Grahamites - 1 year ago    Report SPAM

Tom: First of all, hats-off for holding Ansys for 13 years. I think it's still very respectable even with losing 25% of the profit but recognizing the change in compounding power and moving on. I completely agree that anchoring and confirmation biases can make it hard to reassess the compounding power of a previously-confirmed compounding machine that may no longer keep comounding at the rate it used to.Great point!

Snowballbuilder - 1 year ago    Report SPAM

Hi grahamites interesting article

A relete point is that the entry price matters (the ability to buy great company when they are relative cheap)

akre itself in his great talk "an investor's odissey the search for outstanding investments")

..." So an important observation to us is that price matters enormously. The starting price has everything to do with your compound return, and here we see that the difference between buying the shares in February 1998, March of 2000, October of 2002, and January of 2003, and if you can’t see the chart very well, from February of ’98 to present, there’s an 11% CAGR. From March of 2000 to present, it’s a 3/10% cagger, from January of 2003 to present, it’s a 38 and a half percent cagger, and by the way, from the market bottom, October 9th, 2002, it’s a 66% cagger. So that is indeed the Davis double play."

https://www.gurufocus.com/news/150194/chuck-akre-value-investing-conference-talk-an-investors-odyssey-the-search-for-outstanding-investments

for a really long term owner buying the right company is the single most important factor but buying cheap can boost the return

Have a good day

best snow

Brianbook - 1 year ago

Sell calls when the price is high & puts when the price is low. If the puts are called, better entry price.

Batbeer2 - 1 year ago

Hi Grahamites,

Thanks for your article. It is a very painful reminder of my most expensive mistake so far. A few times a year I review my past transactions and every time I do that, I come across Ingersoll Rand. I bough that one in spring of 09. I probably paid \$12 or thereabouts. I even took time to write about it here: https://www.gurufocus.com/news/51334/ingersoll-rand

At the time IR had just acquired Trane (Trane sells and maintain refrigeration systems) but was percieved to be a company that sold air compressors for building and mining. That was not a booming business. I could see that the acquisition of Trane had transformed IR into a vastly superior franchise. There are three companies (perhaps there are a couple more that I don't know of) with a global maintenance network for refrigerartion systems. You just don't want to write off 30 tons of half-fermented pineappels because the compressor on your reefer failed half way between Hawaii and Hamburg....

There are obvious efficiencies of scale and barriers to entry in that business. I knew Trane and by extension IR would be generating very high returns on equity for a very long time. I figured IR was trading at 7x owner earnings and much lower on FCF. Many companies were cheap back then but I could see this one was of much better quality.

So I bough < 16 and sold some months later @ 24 or something and felt pretty smart.

Now each time I see the stock of IR it pains me. It is my most expensive mistake by far and a very good reminder of recognising the value not only of what you buy but also of what you sell.

In the short run (1-3 years) you can make good money with P/E expansion but in the long run (5-15 years) high ROE matters most. I have learnt that lesson the hard way.

Just some thoughts.

.... Unless it proves to be something else than I think it is, I will be a long term holder (>> 5 years).

!#*&\$^mn!

In hindsight I can only explain my actions because my ego could not resist taking a 100% profit within a year. In the end it turned out to be the compounder I thought it would precisely for the reasons I thought. Ah well, I give my old self a A for analysis and a C for capital allocation.