Everyday Deep Value in Brazil

I. Introduction

1.1 We have completed a much deeper analysis of deep value in Brazil.  After our post Deep Value in Brazil – Further Analysis and New Questions, we decided to test more than one deep value portfolio.  In fact, we tested four different kinds of portfolio, two with the best ranked stocks and two with the worst ranked ones. Also, now we tested the strategy in almost all trading days of the months of June, September and December between 1995 and 2019[1].

1.2 This time we corrected a past omission.  We realized that there was something missing in our prior posts.  We wrote mostly about accounting issues and the mechanics of our research (portfolio size, holding period, etc.), but there was almost nothing that dealt with the reasons that may be behind the results of the deep value strategy.  In short: our texts were missing an introduction. Thus, our goal here is to write the missing introduction to our other posts and summarize the results of this much deeper research.  Additionally, we have a new (and better) explanation of the accounting mechanics of the deep value strategy.  But, if you are a number focused person and do not want much reading, you can go directly to the results of our research in Section VIII below.

1.3 This text has another purpose as well.  It is our thank you to all friends that actually took the time to read our posts (and gave us feedback!).  We will address here the background of the responses to the questions we received and put the rationale behind the deep value investing in a broader context.  The actual responses are in the post Cigars, Diapers and Investments.  This other post is a relaxed approach to most of the topics we cover here.  If you have not read that, click on its title above!

1.4 This post has two annexes.  Annex I is the same one in the post Deep Value in Brazil – Further Analysis and New Questions (since some references are needed for the period covered in our research, we decided to include it here to help the reader to easily find them).  Annex II is a very short explanation about factor investing.  The deep value strategy is a form of factor investing, since it is one of the formats that the value premium may take.  Thus, we decided to give the reader some background on the topic.

II. Value Investing Styles

2.1 I bet that you consider that good investments are the ones in good companies.  A good company is one that is profitable, is in a prosperous sector of the economy and has a bright future ahead.  If you think this way, you are making a surprisingly common confusion: you are taking a good company for a good investment. Both things, in most cases, are different.  If the company is so good, it is very much likely that it is expensive.  All qualities you see on it, you will find in its price as well.  You will pay for the quality.  Can your investment appreciate?  Yes, it can.  But if the company is that good, it may be difficult to make things even better.

2.2 At this point you may think we may be taking you for a fool.  Of course, you know that if a company is good and is in a good moment it will be expensive.  You know that the trick is to find good companies when they are cheap.  Even good companies may face tough times in the market.  From time to time there will be some stress.  Maybe an international crisis will affect the entire stock market for a while.   The business of the company may not suffer, but the stock price will drop.  You know that this is the moment to invest.   You are not an alien, you read a lot about Warren Buffett.  You know that the trick is to find wonderful companies at fair prices.  That is the core of value investing.  You know that the quality of the company is key.

2.3 Now we should go back some years in time, to the teachings of Benjamin Graham, the father of value investing.  In its roots, value investing was all about margin of safety calculated in an objective way.  The quality of the company was not a key element.  The concept of margin of safety was a very conservative shortcut form of valuation to find extremely undervalued stocks[2].  It had a mathematical formula directly derived from the balance sheet.  The acquisition price of a stock could not be more than two thirds of its net-net value.  The net-net was the value of current assets minus all liabilities.  All fixed assets (real estate, equipment, IP, long-term receivables, etc.) were disregarded.  Also, short and long-term liabilities were all taken into consideration.

2.4 It is fair to say that the net-net considered a theoretical liquidation value.  Additionally, it assumed that something could go wrong in the company (requiring the 1/3 value cushion of the price).  In a liquidation scenario, all debts are certain.  Creditors will have to be repaid.  But the value of the assets is uncertain, mainly the one of non-current assets.  It is reasonable to expect that current assets (mainly cash, cash equivalents and inventory) will have a recovery value close to their book value, but the same is not true in the case of non-current assets (mainly equipment, real estate and IP).

2.5 Demanding an acquisition price of 2/3 of the net-net value was a very high threshold.  Finding companies at such prices only happens in times of great stress.  Graham developed this formula after the Great Depression, when stocks were dumped as trash.  Outside times of great financial stress, it is impossible to form portfolios following the original margin of safety metric.  Also, we have to consider that today many valuable assets are not necessarily recorded in the books (such as trademarks) and some companies are asset-light businesses, such as services providers.

2.6 The search for value has evolved over the years and several disciples of Graham followed different paths[3].  The most famous of his disciples is Warren Buffett, who for a long time applied the teachings of Graham in his investment practice[4].  However, it is Warren Buffett who introduced the quality element in the concept of margin of safety, under the influence of his partner Charlie Munger[5].  In short, Buffett started defining margin of safety as the difference between the valuation of a company and the price to acquire such company[6].  In any valuation you must consider future cashflows.  In an approach like this, only the quality of the company may give you some assurance about the quality of the future cashflows.  If in its roots the margin of safety was a relatively simple mathematical formula, it now entered in the realm of the subjectivity. Also, the margin of safety would no longer have as reference the liquidation value of a company.

2.7 The fact that Buffett created his own value investing style does not mean that other styles no longer exist, quite the opposite.  Today, several styles coexist.  In a gross simplification, we can say that value investing can be divided into two schools.  The Graham school, focusing on finding the cheapest stock, without any consideration about quality and the Buffett school, focusing on a combination of good price and good quality.

III. Automatic Value Investing

3.1 The evolution of value investing reached the automatization point[7].  A popular “automatized” version of value investing is the “Magic Formula” of Joel Greenblatt[8].  Greenblatt automatized the Buffett value investing style, mathematically defining the criteria to find good and cheap companies.  To assess the quality of the company, Greenblatt used the Return on Capital (ROC)[9], calculated as EBIT/(Net Working Capital + Net Fixed Assets)[10].  To measure the price, Greenblatt used the Earnings Yield (EY), calculated as EBIT/EV.  Testing the Magic Formula, Greenblatt had a return of 17.9% per year during the period from 1988 to 2004, in a portfolio formed by the best ranking 250 stocks according to the Magic Formula, selected from the 2,500 largest companies in the US market[11].  During the same period, the return of the S&P500 was 12.4% per year[12].

3.2 If Greenblatt made an “automatic Buffet”, we can say that Carlisle made an “automatic Graham”.  Studying the Magic Formula, Carlisle and Wesley Gray replicated the Magic Formula for a longer period (1974 to 2011), and showed that the Magic Formula underperformed the EY alone[13].  In such study, the Magic Formula and the EY had yearly returns of 13.94% and 15.95%, respectively (the S&P500 yearly returned 10.46% in the same period).

3.3 When we decided to replicate famous investment strategies in Brazil, we had a dilemma.  Should we start our work replicating the “automatic Buffett” (Greenblatt) or the “automatic Graham” (Carlisle)?  It was not an easy decision.  We know that in the end we should test both strategies.

3.4 Our choice was to start with the replication of the “automatic Graham” (Carlisle) for three reasons.  Firstly, the rationale that Carlisle offers about the functioning of his strategy is sound.  As we will see in detail below, this is a contrarian strategy in every aspect.  Personally, we like contrarian approaches in so many aspects of life that it could not be different in our investments.  Secondly, the deep value strategy is simpler and it would facilitate our work.  Finally, another issue to be considered was the treatment of goodwill.  In Brazil, goodwill, under some circumstances, was tax deductible for a long period (1997-2017)[14].  In our work as an M&A lawyer, we saw several transactions that have been designed to take advantage of this benefit.  This tax incentive may have polluted some balance sheets and we wanted to start our work with the minimum pollution possible.  The Magic Formula excludes goodwill from the calculation of ROC[15].  If several transactions were structured to create as much goodwill as possible, it means that excluding goodwill from the calculation of ROC could overestimate its value[16].  In any case, testing the Magic Formula in the future is in our pipeline.

IV. Proxies and Relative Valuation

4.1 Before going into the details of the rationale behind the deep value strategy, it is worth mentioning that both, the “automatic Buffett” and the “automatic Graham” rely on proxies to define the actual investments that will be made and are forms of relative valuation.

4.2 To rely on a proxy is to believe that the elected proxy will be a good approximation of the equivalent valuation work.  Since any approximation will contain errors, proxies can only work in the formation of portfolios.  The proper sized portfolio should reduce the consequences of any mistake in the choice of a certain stock.  The “automatic Buffett” believes that the combination of ROC and EY is the best fit to find good investments.  The “automatic Graham” believes that EY alone does a better job.  There is not further valuation of the invested companies.  Some accounting adjustments may be implemented to better capture the value of the stocks, but we have not implemented any of such adjustments in our research.

4.3 A proxy can be used just as a screening method.  In this case, it would filter the universe of investable stocks.  However, it is not advisable to use proxies to pick individual stocks without further valuation for the reasons outlined in the paragraph above.

4.4 It is worth highlighting that both Greenblatt and Carlisle are using methods of relative valuation.  It means that both methods will always be invested in the market, regardless of the actual valuation of the companies in which they invest.  If all stocks in the market are expensive, both methods still will choose stocks to invest.  In this case, the chosen stocks will be the cheapest ones in an overvalued market.  Both strategies can be converted into absolute valuation methods if some sort of threshold is required for the investments, such as a minimum EY of 15%, for example.  In our research, we followed a relative valuation approach and have not included any minimum threshold in the application of the deep value strategy in Brazil.

V. Deep Value Rationale

5.1 We can organize the deep value strategy in five pillars[17]: (i) buy cheap; (ii) behavioral issues and agency conflicts; (iii) mean reversion; (iv) the market finds its ways to unlock value[18]; and (v) the metric.

5.2 Before going into the details of each pillar, we would like to stress a saying that somehow is entwined with a Graham value investing style: “a bird in the hand is worth two in the bush”.  Certainty is good.  It is unnecessary to say that no investment in the stock market can give to the investor any form of certainty, but uncertainty (understood following a commonsense approach to risk) can be minimized. The minimization of uncertainty (not volatility) is key in a Graham approach.  Having it in mind helps to see why margin of safety is the central element of investing for Graham and other investment styles that are more or less derived from his teachings.

Pillar 1 – Buy Cheap

5.3 The concept of margin of safety has been reinterpreted over the years[19].  As we mentioned above, Buffett created his own version of it, taking into consideration subjective aspects of the valuation of a company.  We can say that the margin of safety may have an explicit mathematical form, such as the maximum acquisition price of a stock no higher than 2/3 of the net-net value (the original Graham formula), but it also can be a guiding principle in the formation of a portfolio.   In this sense, we may say that buying the cheapest stocks available (in a reasonable diversified portfolio) is also a form of margin of safety.

5.4 If a stock is cheap, probably something is wrong.   Maybe there is a crisis in the economy or in the sector of that company.  Maybe the company did not meet the market expectations for that quarter.  Or even worse, an accident may have happened, such as the explosion of a factory, with many material and human losses.  Many things may have happened.  In such moments, people sell.  If people sell, stock prices drop.   Regardless of how you calculate the margin of safety, it is very much likely that if stock prices drop, the margin of safety will increase.

5.5 Going for the cheapest stocks, especially if we consider that cheapness is a consequence of something that may be wrong in the general economy, the sector of the company or the company itself, may not be the most intuitive behavior to adopt. In fact, this behavior is exactly the opposite of what normally happens.  If someone shouts “fire”, it is like running into the fire, instead of running from the fire.  At this point, we should take a look at pillar 2 of the deep value strategy, the behavioral issues and agency conflicts.

Pillar 2 – Behavioral Issues and Agency Conflicts

5.6 Maybe, this is the main pillar of the strategy but it is also the most speculative part as well.  We are navigating in the muddy waters of human nature.  One of the main ideas is that people behave in ways that create investment opportunities for the ones that manage to avoid certain behavioral flaws.  It is not our intention to summarize any and all behavioral flaws when dealing with investments.  This is a highly debated topic[20], but we think we can summarize some important ones: (i) fear and fast responses; (ii) need for reasons/stories; and (iii) projection of the present into the future[21].

5.7 It is debated if we are totally in control of our own actions[22].  This debate goes to the point of saying that we have two “operational systems” working at the same time in us.  The conscious and unconscious would be such operational systems[23].  The unconscious system would be inaccessible to us[24], just as the software working in the background of our computers.  We usually think that our conscious mind is a kind of president of our actions and wills[25], but this dual approach speculates that the truth may be radically different.  Maybe our conscious mind is only a form of press secretary of a deliberative mind that operates behind closed doors (outside our consciousness)[26].  Among the actions that would be carried out at closed doors are the fast responses in the view of danger[27].

5.8 A similar approach, citing the existence of more than one operational system in our heads says that we have System 1 and System 2[28].  System 1 has fast responses and is in charge of survival decisions, such as running from a snake (or a stick that looks like a snake).  System 2 has slow but rational responses, such as the decision of buying a house[29].  System 2 should be the one in charge of investment decisions, but is not always the case.  Sometimes System 1 takes charge.  A panic sale of an investment is an example of System 1 working in the field that should be reserved for System 2.   It is the moment when fear takes over.

5.9 We are wired to find reasons to explain what happens around us[30]. There is an ingrained curiosity in the human beings.  This curiosity led to the development of our knowledge and is always testing the limits of science.  That is how knowledge progresses.  However, in several instances our curiosity is satiated with a reason, regardless of the fact such reason is right or wrong.  Superstition and religion are sources of explanations to what happens around us, disregarding any verification of the correctness of the explanations they provide.  When it comes to investments, there may be lots of explanations to what happened during the day.  We always see some explanations in the financial press.  In most cases, it is hard to believe that many of the causes people mention as “reasons” for what is happening is accurate.

5.10 We have the impression that the projection of the present into the future may be the most important element of the behavioral pillar of the deep value strategy[31].  Sometimes, people tend to be very optimistic or pessimistic about the future.  Maybe we are unable to fully access how uncertain the future can be.  It can be better, worse or things can remain the same.  We are just unable to make any correct assessment about it.  We can illustrate this circumstance with the general behavior of soccer fans after their favorite team wins a championship.  It would not be surprising to have several fans saying that their team is the best, is going to win the next championship and is positioned to future records.  Everything is easy and the future of the team seems rosy.  At the same time, if we think about the fans of a soccer team that has just been downgraded to a lower league, we may expect the opposite behavior.  They will see no future for their team, it will be the end of the world and the future is dark.  All fans are projecting the present into the future.

5.11 People tend to think that good companies will always be good and bad companies will always be bad.  It makes them to pay too much for the good ones and dump the bad ones, depressing their stock prices.  But if you think about a competitive market, it is of its essence to equalize things (mean reversion, see below).  If a business is very good, it will attract competition.  Think about Cielo (the payment means company).  It lived a long honeymoon with the market and paid juice dividends for a long time.  Now it has several competitors and is having a hard time in market.  If people perpetuate the present in the future, they may pay too much for some growth that may never exist or drive down the price of stocks that are facing hard times, anticipating that problems will never be solved.

5.12 The issues we mentioned above have a double impact when it comes to professional money managers.  First of all, they are human.  They are not free from behavioral flaws.  However, there is an additional issue.  Money managers are exposed to the moods of their investors.  At this point we have a potential agency conflict.   We should have in mind that in most cases, money managers are remunerated based on the amount of assets under management and a performance fee.  If the manager exceeds the benchmark, the performance fee is paid.  This fee usually is calculated based on an industry benchmark.  In Brazil, the most popular benchmark is the Ibovespa index.  The Ibovespa tracks the performance of the main stocks traded in B3, the Brazilian stock exchange.

5.13 A professional manager that follows a strategy that may be for long periods underperforming the benchmark has to have investors that are genuinely aligned with such strategy.  Otherwise, investors will withdraw their money.  Without investors, the manager will have no management or performance fees.  It is speculated that this kind of career risk is one of the causes that allow value strategies to perform[32].  Otherwise, all professional managers would follow value investing strategies and the value premium would disappear[33].   We also speculate about how investors would react to strategies that on purpose rejects the quality of the companies in which they invest.  We would expect many investors to doubt the merits of such strategies.

Pillar 3 Mean Reversion

5.14 Carlisle provides a broad explanation of the meaning of mean reversion and gives several examples of it[34].  It is fair to say it is a pervasive phenomenon, taking place in several fields.  If we focus on some basic characteristics of a competitive economy, the occurrence of mean reversion becomes very intuitive.  At the company level, if a business is profitable it will attract competition.  The smartphones industry is an example of it.  By the same token, if a business becomes so problematic to the point of bankrupting its participants, it is expected that in a certain moment the remaining companies in that business may start raising prices.  At the market level, from time to time there are situations of over or undervaluation.  When any of them happens, in a certain moment a correction follows.  It is hard to say “why” it happens or “when” it will happen.

Pillar 4 The Market Finds its Ways to Unlock Value

5.15 Carlisle argues in the conclusion of his book[35] that activism and deep value investing go hand in hand.  It is nearly impossible to make a similar affirmation in Brazil, in view of the rare cases in which activism happened in the country.  Despite the development of the Brazilian stock market in the last couple of decades, activism here still is very rare.  However, we can mention at least one recent case that indicates that his conclusion may be appliable in Brazil as well. We refer to the case of Unipar (chemical company), when the minority shareholders refused an offer of the controlling shareholder to make the company private.  The offer was virtually ignored by the market, but increased the public attention on Unipar.  The controlling shareholders initially offered R$4.40 per share in the case of preferred shares class B (ticker: UNIP6), the most liquid paper.  Later, this price was increased to R$7.50 and subsequently adjusted to R$2.46 per share in view of the payment of dividends.  The offer was launched on December, 14, 2015 and withdrawn on August 23, 2017.  At the time the offer was launched, the preferred class B was trading at R$3.88 and when the offer was withdrawn the price of the preferred class B was R$11.04.  Within one year from the cancelation of the offer, the price of the preferred class B of Unipar reached R$43.50.  Luiz Barsi, a legendary Brazilian investor, was leading the minority shareholders at that moment.

5.16 If on the one hand we rarely see activism in Brazil, it does not mean that we do not have other mechanisms to unlock the value of deep value stocks.   Several situations may unlock value, such as the acquisition of the undervalued company by a competitor.  One example is the tender offer of Cosan to acquire the control of Comgas (São Paulo gas company), paying a premium of 23,31% on the preferred class A (ticker: CGAS5) in January of 2019[36].

5.17 We also speculate that discretionary value investors contribute to unlocking value of undervalued companies, in a kind of self-fulfilling prophecy.  If many discretionary value investors start buying the same stocks, at a certain point such stocks will appreciate.

 Pillar 5 – The Metric

5.18 The metric is the heart of the deep value strategy.  Gray and Carlisle tested several metrics and covered the academic research on many metrics as well[37]. In the end, the chosen metric was the earnings yield (EY).  EY takes in its calculation EBIT and Enterprise Value (EV).  EY=EBIT/EV.  We can better understand it if we analyze Enterprise Value (EV) and EBIT separately.

5.19 EV is the sum of Net Debt and Market Capitalization of the company (market value of each stock multiplied by the number of outstanding shares).  Before defining its elements, we have to go back to the basics.  Results are produced by real assets.  At this point, please do not think about how the assets are financed (debt or equity).

5.20 Knowing the value of an asset can be tricky.  If we put a bunch of assets together to run a business, knowing the value of this bunch of assets bundled together is even trickier.  Think about a bakery.  You have an oven, tables, an expresso machine, a counter and so on and so forth.  All assets together form the bakery.  In normal circumstances, the bundled assets should have a value that exceeds the value of sum of each part.  The ingenuity of the entrepreneur must have some value.

5.21 It is the dream of every investor to know the value of a company.  We dream about learning how to calculate it every night and daydream about it too.  But that is not the point.  We will never know the precise value of any asset and, more importantly, we do not need precision.  But we need to assess the value of the business, even if imprecisely.

5.22 Accounting will help us to find the value of the business.   If we think about a balance sheet, results are produced by the asset side.  Not coincidently, the asset side of the balance sheet has the real assets (mostly) plus cash and cash equivalents.   Here, for the sake of simplification, we can assume that the asset side of the balance sheet has only real assets plus cash and cash equivalents.   We will call the real assets plus cash and cash equivalents the “Business”.  Also, still for the sake of simplification, we can assume that the liability side of the balance sheet we will have only financial debts (i.e., no liabilities related to accounts payable, salaries or taxes).

5.23 Taking a step further, one should bear in mind the fundamental accounting equation:  Assets = Liabilities + Net Equity.  Voilà, the value of the Business is equal to the value of the liabilities + net equity.  We have a proxy to know the value of the Business.  Here we can improve our terminology.  When we assumed that there were no liabilities related to accounts payable, salaries or taxes in the liability side of the balance sheet, in fact we made the liabilities equal to the Total Debt. We need to know the value of the liabilities (total debt) and the net equity.  If we stop our work in the fundamental accounting equation, we will get a form of book value of the Business.  The book value may be extremely helpful, but we want something more precise, we want the market value of the Business.

5.24 How can we know the value of the liabilities (total debt)?  We will have to rely on the financial statements to know it.  At this point, it is worth mentioning that accounting is guided by the conservatism principle, mainly when it comes to liabilities.  Thus, putting aside accounting magics, we can expect that liabilities will be conservatively recognized.  How about the value of the net equity?  Since we are talking about publicly traded companies, the value of the net equity for our purposes is given by the market.  If you sum the book value of the liabilities (total debt) to the market value of the company, you will find the value of the Business.

5.25 The exercise above gives you a rudimentary value of the Business that can be improved.  You remember that the Business includes the value of cash and cash equivalents.  Since the value of USD1.00 in cash is USD1.00, it is easy to assess the value of cash and cash equivalents.  Thus, we exclude the value of cash and cash equivalents from the value of the Business.  Let’s assume that the value of cash and cash equivalents is used to pay part of the debt.  This allow you to know the value of the part of the Business that is much harder to assess.  You just found, the Enterprise Value, explained in a somewhat different way.  In the dictionary, “business” and “enterprise” are synonyms, but the word commonly used in finance to define the value of the entire business is “enterprise”.  We followed this path to calculate the Enterprise Value to be closer to the accounting mechanics.

5.26 You may have heard that the Enterprise Value is the value necessary to acquire the entire business.  It is correct, but ordinary people may have some difficult to get it.  They will think about the Market Capitalization value as the value necessary to acquire the entire business[38].  They will think that you acquire the business if you acquire all capital stock.  But if you acquire all capital stock of a company, you still have to repay the debt.  The Enterprise Value is the value that allows you acquire all capital stock and repay all debt.

5.27 See below the calculations, starting from the fundamental accounting equation:

Starting point: original equation

Assets = Liabilities + Net Equity

Step 1: assume that (i) only real assets and cash and cash equivalents are in the asset side and rename such assets to Business (ii) the liabilities side of the balance sheet has no liabilities related to accounts payable, salaries or taxes and rename it to Total Debt; now the equation reads as follows,

Business Value = Total Debt + Net Equity

Step 2: since the company is publicly traded, use Market Capitalization in the place of Net Equity

Business Value = Total Debt + Market Capitalization

Step 3: exclude cash and cash equivalents from Total Debt

Business Value = Total Debt – (Cash + Cash Equivalents) + Market Capitalization

Step 4: substitute “Total Debt – (Cash + Cash Equivalents)” by “Net Debt” and “Business” by Enterprise

Enterprise Value (EV) = Net Debt + Market Capitalization

5.28 You can see that in the case of public companies, EV has a stable component, the value of the Net Debt (released quarterly) and a floating component, the Market Capitalization (that varies each day).   Thus, in any backtesting it is important to match the market value in a certain moment with the then available financial statements.  A parochial, but tricky subject, as we will see later.

5.29 We do not ignore that restricting the asset side of a balance sheet to the Business (real assets) is a form of oversimplification.  In addition to real assets and cash and cash equivalents, the asset side has tax credits and goodwill, for example.  To define the value of such assets can be extremely hard.  One can even question whether goodwill has any value or if it is only an accounting plug (an accounting number to make the asset and liability sides of the balance sheet equal).  Also, we know that the liability side of the balance sheet can be maneuvered as well.  The accounting of leases is an example.

5.30 Now, we will explain EBIT.  We can say it is the less famous cousin of EBITDA.  We take it as proxy for free cash flow, meaning the cash available to the company after saving some money to keep the business running.  Think about the bakery again.  Some money has to be spent to maintain the equipment running.  Since some cash has to be reinvested in the business, not all money a company generates is free cash flow.  It is the reason that we use EBIT as a proxy for free cash flow instead of EBITDA. EBIT is a number lower than EBITDA because the D (depreciation) and A (amortization) have been deducted from it.  As EBIT is a proxy for free cash flow, the depreciation and the amortization are proxies for the monies reinvested in the business to keep it running.

5.31 We also know that EBIT can be – and is – heavily managed.  There is a lot of room to define the moment in which a company recognizes its sales, for example.  More importantly, in the real world, depreciation and amortization are defined by tax reasons.  EBIT as a proxy for free cash flow can be highly imprecise.  In any case, we understand that from a conceptual point of view it still does a better job than EBITDA, since at least the values of depreciation and amortization are considered in its calculation.

5.32 As in the case of EV, knowing the proper EBIT in place in the past is essential to have a reliable backtest.

5.33 Putting everything together, the higher the ratio EBIT/EV, the better.  Thus, a company whose market value is sinking may be a good investment, for example.  Remember that EBIT/(Net Debt + Market Capitalization), the lower the Market Capitalization, the higher is the ratio EBIT/EV.

5.34 If we invert the equation EBIT/EV and use EV/EBIT instead, we will be using a multiple.  It means that the value of a certain company is “X” times its EBIT.  This multiple (to be more precise the one using EBITDA in the place of EBIT) is one of the most common measure to define price in private deals.  Carlisle call it the Acquirer’s Multiple[39].  In any case, we prefer to mention EBIT/EV because we understand that it is a better way of showing the yield of the assets of a company.

VI. New Research – Methodology

What is New and Selection Criteria

6.1 This time our research was broader.  Our prior research was limited to the best ranked stocks.  At that time, our tested portfolios included only the top 20 best ranked stocks.  This time, we wanted to see the performance of other portfolios as well.  We wanted to see the performance of the best ranked stocks and the worst ranked ones.  Also, the 2019 research was limited to portfolios formed in one day per month in June, September and December.  Now, we decided to form portfolios in virtually all business days of such months.

6.2 To test differently ranked portfolios, we had a practical problem: the limited number of stocks publicly traded in Brazil.  To give you the dimension of the problem, after applying some hygiene filters, we ended up with only 36 stocks in the 2nd business day of September 1995 and  33 stocks in the 2nd business day of September 1996 in the investable universe of our large cap portfolio, for example (fortunately, such a limited number of stocks happened in very few years only).  It would be impossible to organize the stocks in deciles and keep some decent diversification.   Thus, we divided the stocks into two categories of 40 stocks each (for most of the time)[40], and formed 2 portfolios of stocks in each one.  The top-ranking stocks are the “Top” category, divided into group 1 (“G1”) and group 2 (“G2”).  The bottom-ranking stocks are the “Bottom” category, divided into group 1 (“G1”) and group 2 (“G2”)[41].

6.3 The choice of June, September and December was a consequence of the dates in which financial statements were released in Brazil until 2011 (see Annex II for details).  In such months we know that the last available financial statements were the ones for the first, second and third quarters, respectively, for all publicly traded companies during all period of our research.  Thus, we are able to match the stock price with the last available accounting information at the time (this information is necessary to define EV, since market capitalization is one of its elements, and the purchase price of the stocks in our theoretical portfolios).  This choice of months was our attempt to be as precise as possible in the matching of the market value of a company and the stock price to the them available accounting information.  Our database does not offer a “point in time” option, meaning that we cannot rule out that part of our results is influenced by the restatement of some financial statements.  Also, we have to consider that some companies may have fiscal years that do not perfectly match the quarter endings, what adds some uncertainty about the proper date matching.  In any case, it is common practice in Brazil that the fiscal year follows the calendar year.

6.4 In this new research, instead of having a single portfolio in June, September and December, we formed portfolios in each of eighteen business days in June and September and seventeen business days in December (the year-end holidays made December a shorter month).  The number of days selected was not arbitrary.  Since the number of business days in each month varies over the years, we formed portfolios in a number of business days that allowed perfect matching between a certain business day of a month with the same business day of that month in the subsequent year.   We wanted to have one-year portfolios in all of each of the selected days.  It means that a portfolio formed, for example, in the third business day of June of a certain year was held for one year, until the third business day of June of the next year.  In Section VII we compare the performance of our theoretical portfolios to the performance of a theoretical Ibovespa, as well.  Since each presented theoretical portfolio is an average of other portfolios, we had to make an average of the Ibovespa.  We compared each of the individual portfolios we formed to the Ibovespa and them made an average of all of performances of the Ibovespa, just to keep uniformity.

6.5 We had to exclude the first business day of each of the chosen months.  In Brazil, it is very common to have companies with more than one class of shares publicly traded.  To deal with it and keep the defined diversification, we selected only the most liquid class of traded shares in the first day of the month.  To have the liquidity requirement also reflecting the new financial information, the first business day of the month was not taken into consideration in the portfolio formation.

6.6 Below, we present more details of the methodology of our research: (i) top and bottom categories and their groups; (ii) base and average portfolios; (iii) number of stocks in portfolios; (iv) weight per stock in each portfolio; (v) holding period; (vi) market cap; (vii) required liquidity; (viii) presence in trading days; (ix) excluded sectors; and (x) hygiene filters.

Top and Bottom Categories and Their Groups

6.7 We ranked the stocks following their EY and formed two portfolios with 20 stocks in each of the Top and Bottom categories mentioned above.  The Top G1 portfolios were formed with the best 20 ranked stocks and the Top G2 portfolios were formed with the next 20 best ranked stocks.  The Bottom G1 portfolios were formed with the worst 20 ranked stocks and the Bottom G2 portfolios were formed with the next 20 worst ranked stocks[42].

Base and Average Portfolios

6.8 The portfolios formed in each of the selected business days of June, September and December we called base portfolios.  In view of the number of base portfolios formed (18 in June and September and 17 in December for each of the 4 groups (Top G1, Top G2, Bottom G1 and Bottom G2)) we formed an average portfolio for each of such months for each of the categories mentioned above.  Thus, the charts and graphics below show the average of the base portfolios.

Number of Stocks and Weight Per Stock in each Base Portfolio

6.9 In the studies we are posting here, the rule was to form portfolios with 20 stocks, each representing 5% of the total of the respective portfolio[43].  One alternative to weight the stocks would be to market weight them in the portfolio formation (following the market cap of each invested company).  This alternative would further complicate our work at this stage.  Also, equal weight in the portfolio formation looks more intuitive.

6.10 Choosing 20 stocks was an arbitrary decision guided by our most fundamental (and rudimentary) form of margin of safety: diversification.  If one company goes bust, we lose only 5% of our investment.  We do not like the idea of further diversification in view of the size of the Brazilian market.   Just to give you an example, on April 6, 2020, only 353 companies were active in the Brazilian stock market[44].  A portfolio of 20 stocks already represents approximately 5.70% of the Brazilian market in number of companies.  A portfolio of 30 companies would mean investing in almost 8.50% of the Brazilian market.

Holding Period and Portfolio Dates

6.11 We followed a one year holding period.   This was the holding period of the portfolios formed by Carlisle[45], Greenbaltt[46], and O’Shaugnhnessy[47] in their studies.  Additionally, from a practical point of view, this holding period is a way to minimize trading costs and the time spent on the portfolio management.

6.12 As mentioned in an earlier post (Deep Value in Brazil – The Novel), the chosen starting year was 1995 (first full year after the end of our chronic inflation).  As mentioned above, we formed portfolios in the months of June, September and December in view of the dates in which financial statements were disclosed until 2011 (see Annex II for more details about this issue).

Market Cap

6.13 It is impossible to define precisely the meaning of small, mid and large caps.  To avoid arbitrary definitions, such as an absolute number of market capitalization (for example, R$100 million), we decided to create portfolios based on relative market capitalization.  We created 3 categories of portfolios, DVA, DVB, and DVC (DV meaning Deep Value).  DVA has no minimum market cap requirement.  Thus, DVA portfolios can invest in small, mid or large caps.  DVB excludes the 1/3 of the smallest companies in the market.  We used this choice of the remaining 2/3 of the companies as a proxy selection for mid and large caps.  Thus, DVB portfolios can invest only in mid and large caps.  Finally, DVC excludes the 2/3 of the smallest companies in the market.  We used this choice of the remaining 1/3 of the companies as a proxy selection for large caps.  Thus, DVC portfolios can invest only in large caps.  Table 1 shows the minimum market cap for DVA, DVB, and DVC portfolios, respectively[48].  Such values have been collected on June 1st of each year.

Table 1

Minimum Market Capitalization (in thousands)
Year DVA DVB DVC
1995 N/A  R$41,160  R$201,769
1996 N/A  R$28,635  R$189,155
1997 N/A  R$42,754  R$334,072
1998 N/A  R$43,013  R$294,149
1999 N/A  R$47,436  R$324,145
2000 N/A  R$59,197  R$444,561
2001 N/A  R$64,458  R$502,989
2002 N/A  R$55,853  R$518,903
2003 N/A  R$65,078  R$600,206
2004 N/A  R$92,282  R$859,690
2005 N/A  R$139,418  R$1,174,608
2006 N/A  R$188,430  R$1,686,403
2007 N/A  R$322,970  R$2,420,074
2008 N/A  R$481,522  R$2,357,514
2009 N/A  R$266,276  R$1,589,000
2010 N/A  R$482,367  R$2,334,060
2011 N/A  R$472,884  R$2,800,483
2012 N/A  R$459,085  R$2,514,586
2013 N/A  R$508,373  R$3,353,213
2014 N/A  R$405,864  R$2,860,525
2015 N/A  R$301,976  R$2,758,578
2016 N/A  R$263,804  R$2,480,484
2017 N/A  R$360,559  R$3,870,917
2018 N/A  R$367,604  R$4,418,340
2019 N/A  R$415,925  R$5,401,650

Required Liquidity

6.14 Since a DVA portfolio may include companies that are very small, to avoid the creation of a portfolio that is impossible to replicate in the real world we required a minimum daily liquidity from each stock.  The rationale behind it was to allow a small investor (but not very small one) to liquidate, in a stress situation, all of his position in a certain stock in a single trading day or over the course of few days.  The individual liquidity requirement was arbitrarily defined as R$100,000 per day in January 1995 and corrected by the IPCA (official inflation rate) annually since then.  Table 2 shows the values for each year.  For the sake of simplicity, we correct the value annually only[49].

6.15 To assure uniformity in the formation of all portfolios, we also used the liquidity requirement above mentioned in DVB and DVC portfolios.

Table 2

Year Daily Average Liquidity
1995  R$ 100,000.00
1996  R$ 117,490.00
1997  R$ 126,548.48
1998  R$ 131,547.14
1999  R$ 135,677.72
2000  R$ 144,456.07
2001  R$ 154,625.78
2002  R$ 166,640.20
2003  R$ 195,368,97
2004  R$ 205,430.48
2005  R$ 221,967.63
2006  R$ 231,356.86
2007  R$ 238,714.01
2008  R$ 252,034.25
2009  R$ 265,140.03
2010  R$ 278,980.34
2011  R$ 297,253.55
2012  R$ 312,086.51
2013  R$ 332,372.13
2014  R$ 353,544.23
2015  R$ 383,489.43
2016  R$ 419,230.65
2017  R$ 434,322.95
2018  R$ 446,744.59
2019  R$ 467,562.88

Presence in Trading Days

6.16 In view of the fact that the minimum liquidity is an average of the value traded over the year, it can be misleading.  A single day in which high amounts were traded can mislead the numbers.  Thus, to mitigate this risk we also required a minimum trading presence in all trading days of 80%.  It means that a certain stock had to be traded in 80% of all trading days.  If this solution did not eliminate distortions in the minimum liquidity, it at least reduced it.

Excluded Sectors

6.17 We excluded from our investable universe financial institutions and insurance companies.  They have different balance sheets (if compared to other companies) and we cannot apply the ratio EBIT/EV to them.

6.18 We did not exclude utilities from our investable universe.  The number of public companies in Brazil is too small and if we did that, we would have an even smaller investable universe.

Hygiene Filters

6.19 We had two requirements that we call numeric “hygiene filters”: (i) negative net equity and (ii) negative EBIT.  We excluded from our investable universe companies that presented any of them.  The rationale behind the exclusion of negative net equity was to avoid companies that could be in a very bad situation.  We know that we risk losing good opportunities, but this exclusion was a form of margin of safety too.  As to the rationale to exclude companies with negative EBIT, we did that to avoid to be fooled by the numbers.  We wanted to eliminate from our investable universe a mathematical mistake.   A negative EBIT divided by a negative EV would be a positive number.

VII. Our Results

Portfolios DVA

7.1 The DVA Portfolios are the ones in which our investable universe has no minimum market capitalization requirement.  As all portfolios, they contain 20 stocks each[50] and the holding period is one year. We rank all stocks of our investable universe using the EBIT/EV ratio.  The best 20 ranked stocks are in the DVA-Top-G1 and next 20 better ranked are in the DVA-Top-G2.  The 20 worst ranked stocks are in the DVA-Bottom-G1 and the next 20 worst stocks are in the DVA-Bottom-G2.   The June, September and December portfolios presented below are averages of our theoretical portfolios formed in each of eighteen business of June and September and seventeen business days of December (see item 6.4 above), respectively.  The performance of the Ibovespa presented is also an average of the performance for the same period of the DVA Portfolios (see item 6.4 above).  Table 3, Table 4 and Table 5 show the performance of each all DVA Portfolios and the Ibovespa.   Also, Chart 1, Chart 2 and Chart 3 compare each of such portfolios to the Ibovespa.

Table 3 – Performance Portfolio DVA – June

DVA June Performance

DVA-Top-G1 DVA-Top-G2 DVA-Bottom-G2 DVA-Bottom-G1

IBOV

Initial Value

100.00

100.00 100.00 100.00

100.00

Final Value

10,796.74

10,956.12

6,537.01  1,894.57

2,755.78

Initial Month

Jun/95

Jun /95 Jun/95 Jun/95

Jun/95

Final Month

Oct/19

Oct/19 Oct/19 Oct/19

Oct/19

CGAR

21.20%

21.27% 18.73% 12.84%

14.59%

Positive Months

 183

189 186 178

174

Negative Months

109 103 106 114

118

Best Month

19.72%

74.94% 32.11% 79.97%

26.41%

Worst Month

-26.16%

-28.58% -32.00% -32.64%

-26.53%

Total Return

10,697.00%

10,856.00% 6,437.00% 1,795.00%

2,656.00%

Source: Proprietary research, based on information provided by Economatica

 Chart 1 – DVA All Portfolios – June x Ibovespa

Chart 1

 

Source: Proprietary research, based on information provided by Economatica

 Table 4 – Performance DVA – September

DVA September Performance

DVA-Top-G1

DVA-Top-G2 DVA-Bottom-G2 DVA-Bottom-G1

IBOV

Initial Value

100.00

100.00 100.00 100.00

100.00

Final Value

22,649.60

9,361.25 6,720.80 1,239.33

2,420.58

Initial Month

Sep/95

Sep/95 Sep/95 Sep/95 Sep/95
Final Month

Nov/19

Nov/19 Nov/19 Nov/19

Nov/19

CGAR

25.14%

20.65% 19.00% 10.97%

14.08%

Positive Months

181

190  184 175

173

Negative Months

 109

100  106 115

117

Best Month

22.92%

22.11% 20.53% 74.36%

26.41%

Worst Month

-26.29%

-27.56% -28.98% -34.87%

-26.53%

Total Return

22,550.00%

9,261.00% 6,621.00% 1,139.00%

2,321.00%

Source: Proprietary research, based on information provided by Economatica

Chart 2 – DVA All Portfolios – September x Ibovespa

Chart 2

Source: Proprietary research, based on information provided by Economatica

 Table 5 – Performance DVA – December

DVA December Performance

DVA-TOP-G1

DVA-Top-G2 DVA-Bottom-G2 DVA-Bottom-G1

IBOV

Initial Value

100.00

100.00 100.00 100.00

100.00

Final Value

 30,703.72

13,487.21  3,496.45 2,574.83

2,594.82

Initial Month

Dec/95

Dec/95 Dec/95 Dec/95 Dec/95
Final Month

Dec/19

Dec/19 Dec/19 Dec/19

Dec/19

CGAR

26.93%

22.66% 15.95% 14.48%

14.52%

Positive Months

 180

  184  173  183

 173

Negative Months

108 104 115 105

115

Best Month

26.19%

21.16% 21.44% 66.91% 27.37%

Worst Month

-26.94%

-29.40% -29.28% -30.84%

-27.66%

Total Return

30,604.00% 13,387.00% 3,396.00% 2,475.00%

2,495.00%

Source: Proprietary research, based on information provided by Economatica

Chart 3 – DVA All Portfolios – December x Ibovespa

Chart 3

Source: Proprietary research, based on information provided by Economatica

Portfolios DVB

7.2 The DVB Portfolios are the ones in which our investable universe is restricted to mid and large caps.  As all portfolios, they contain 20 stocks each[51] and the holding period is one year. We rank all stocks of our investable universe using the EBIT/EV ratio.  The best 20 ranked stocks are in the DVB-Top-G1 and next 20 better ranked are in the DVB-Top-G2.  The 20 worst ranked stocks are in the DVB-Bottom-G1 and the next 20 worst stocks are in the DVB-Bottom-G2.   The June, September and December portfolios presented below are averages of our theoretical portfolios formed in each of eighteen business of June and September and seventeen business days of December (see item 6.4 above), respectively.  The performance of the Ibovespa presented is also an average of the performance for the same period of the DVB Portfolios (see item 6.4 above).  Table 6, Table 7 and Table 8 show the performance of each all DVB Portfolios and the Ibovespa.   Also, Chart 4, Chart 5 and Chart 6 compare each of such portfolios to the Ibovespa.

Table 6 – Performance DVB – June

DVB June Performance

DVB-Top-G1

DVB-Top-G2 DVB-Bottom-G2 DVB-Bottom-G1

IBOV

Initial Value

100.00

100.00 100.00 100.00

100.00

Final Value

 13,311.54

13,159.06 6,898.91 1,903.97

2,755.78

Initial Month

Jun/95

Jun/95 Jun/95 Jun/95

Jun/95

Final Month

Oct/19

Oct/19 Oct/19 Oct/19

Oct/19

CGAR

22.25%

22.19% 18.99% 12.86%

14.59%

Positive Months

182

193  186 178

174

Negative Months

 110

99 106 114

118

Best Month

20.23%

72.99% 39.29% 21.03%

26.41%

Worst Month

-26.34%

-28.41% -33.26% -32.04%

-26.53%

Total Return

13,212.00%

13,059.00% 6,799.00% 1,804.00%

2,656.00%

Source: Proprietary research, based on information provided by Economatica

 Chart 4 – DVB All Portfolios – June x Ibovespa

Chart 4

Source: Proprietary research, based on information provided by Economatica

 Table 7 – Performance DVB – September

DVB September Performance

DVB-Top-G1

DVB-Top-G2 DVB-Bottom-G2 DVB-Bottom-G1

IBOV

Initial Value

100.00 100.00 100.00 100.00

100.00

Final Value

20,654.43

10,280.01  7,384.80  1,549.83

2,420.58

Initial Month

Sep/95

Sep/95 Sep/95 Sep/95

Sep/95

Final Month

Nov/19

Nov/19 Nov/19 Nov/19

Nov19

CGAR

24.66%

21.11% 19.47% 12.00%

14.08%

Positive Months

 181

 187    185 176

 173

Negative Months

109

103  105  114

117

Best Month

22.94% 22.15% 20.09% 22.37% 26.41%
Worst Month

-27.36%

-27.24% -28.96% -33.10%

-26.53%

Total Return

20,554.00% 10,180.00% 7,285.00% 1,450.00%

2,321.00%

Source: Proprietary research, based on information provided by Economatica

 Chart 5 – DVB All Portfolios – September x Ibovespa

Chart 5

Source: Proprietary research, based on information provided by Economatica

 Table 8 – Performance DVB – December

DVB December Performance

DVB-Top-G1

DVB-Top-G2 DVB-Bottom-G2 DVB-Bottom-G1

IBOV

Initial Value

 100.00

100.00 100.00 100.00

100.00

Final Value

22,159.76  10,544.33 3,844.29 2,241.46

2,594.82

Initial Month

Dec/95

Dec/95 Dec/95 Dec/95

Dec/95

Final Month

Dec/19

Dec/19 Dec/19 Dec/19

Dec/19

CGAR

25.22%

21.40% 16.41% 13.82%

14.52%

Positive Months

177

184 174 182

173

Negative Months

111

104 114 106

115

Best Month

21.22%

21.44% 21.21% 26.33%

27.37%

Worst Month

-27.02%

-29.37% -29.42% -30.86%

-27.66%

Total Return

22,060.00%

10,444.00% 3,744.00% 2,141.00%

2,495.00%

Source: Proprietary research, based on information provided by Economatica

 Chart 6 – DVB All Portfolios – December x Ibovespa

Chart 6

Source: Proprietary research, based on information provided by Economatica

 Portfolios DVC

7.3 The DVC Portfolios are the ones in which our investable universe is restricted to large caps.  As all portfolios, they contain 20 stocks each[52] and the holding period is one year. We rank all stocks of our investable universe using the EBIT/EV ratio.  The best 20 ranked stocks are in the DVC-Top-G1 and next 20 better ranked are in the DVC-Top-G2.  The 20 worst ranked stocks are in the DVC-Bottom-G1 and the next 20 worst stocks are in the DVC-Bottom-G2.  The June, September and December portfolios presented below are averages of our theoretical portfolios formed in each of eighteen business of June and September and seventeen business days of December (see item 6.4 above), respectively.  The performance of the Ibovespa presented is also an average of the performance for the same period of the DVC Portfolios (see item 6.4 above).  Table 9, Table 10 and Table 11 show the performance of each all DVC Portfolios and the Ibovespa.   Also, Chart 7, Chart 8 and Chart 9 compare each of such portfolios to the Ibovespa.

Table 9 – Performance DVC – June

DVC June Performance

DVC-Top-G1

DVC-Top-G2 DVC-Bottom-G2 DVC-Bottom-G1

IBOV

Initial Value

100.00

100.00  100.00 100.00

100.00

Final Value

11,053.03

4,645.18 5,498.24 2,619.03

2,755.78

Initial Month

Jun/95

Jun/95 Jun/95 Jun/95

Jun/95

Final Month

Oct/19

Oct/19 Oct/19 Oct/19

Oct/19

CGAR

21.32%

17.07% 17.89% 14.35%

14.59%

Positive Months

186 189 177 180

174

Negative Months

106

103 115 112

118

Best Month

61.15%

33.76% 67.99% 21.61%

26.41%

Worst Month

-29.46%

-27.12% 28.68% -28.90%

26.53%

Total Return

10,953.00%

4,545.00% 5,398.00% 2,519.00%

2,656.00%

Source: Proprietary research, based on information provided by Economatica

 Chart 7 – DVC All Portfolios – June x Ibovespa

Chart 7

Source: Proprietary research, based on information provided by Economatica

 Table 10 – Performance DVC – September

DVC September Performance
 

DVC-Top-G1

DVC-Top-G2 DVC-Bottom-G2 DVC-Bottom-G1

IBOV

Initial Value

100.00

100.00  100.00 100.00

100.00

Final Value   10,084.84 4,410.37 7,580.19 2,017.82 2,420.58
Initial Month

Sep/95

Sep/95 Sep/95 Sep/95

Sep/95

Final Month

Nov/19

Nov/19 Nov/19 Nov/19

Nov/19

CGAR

21.02%

16.95% 19.60% 13.23%

14.08%

Positive Months

187

191 183 183

173

Negative Months

103

 99 107 107

117

Best Month 22.54% 23.19% 22.60% 23.05% 26.41%
Worst Month

-31.40%

-26.53% -30.40% -30.81%

-26.53%

Total Return

9,985.00%

4,310.00% 7,480.00% 1,918.00%

2,321.00%

Source: Proprietary research, based on information provided by Economatica

 Chart 8 – DVC All Portfolios – September x Ibovespa

Chart 8

Source: Proprietary research, based on information provided by Economatica

 Table 11 – Performance DVC – December

DVC December Performance

DVC-Top-G1

DVC-Top-G2 DVC-Bottom-G2 DVC-Bottom-G1

IBOV

Initial Value

100.00

 100.00  100.00 100.00

100.00

Final Value

12,361.42

4,657.49  6,447.66 1,706.87

2,594.82

Initial Month

Dec/95

Dec/95 Dec/95 Dec/95

Dec/95

Final Month

Dec/19

Dec/19 Dec/19 Dec/19

Dec/19

CGAR

22.21%

17.34% 18.94% 12.54%

14.52%

Positive Months

184

186 181 181

173

Negative Months

104

102 107 107

115

Best Month

19.59%

29.33% 19.64% 25.40%

27.37%

Worst Month

-28.58%

-29.27% -28.61% -30.13%

-27.66%

Total Return

12,261.00% 4,557.00% 6,348.00% 1,607.00%

2,495.00%

Source: Proprietary research, based on information provided by Economatica

 Chart 9 – DVC All Portfolios – December x Ibovespa

Chart 9

Source: Proprietary research, based on information provided by Economatica

Performance Comparison

7.4 It is worth organizing the results in some short tables, to better visualize the order of performance of our theoretical portfolios according to the EY ranking (Top-G1, Top-G2, Bottom-G2 and Bottom-G1).  Table 12, Table 13 and Table 14 show the order of performance (from 1 to 4, 1 meaning the best performance and 4 the worst one) in June, September and December for the portfolios DVA, DVB and DVC, respectively.

Table 12

DVA
Month Top-G1 Top-G2 Bottom-G2 Bottom-G1
June 2 1 3 4
September 1 2 3 4
December 1 2 3 4

Table 13

DVB
Month Top-G1 Top-G2 Bottom-G2 Bottom-G1
June 1 2 3 4
September 1 2 3 4
December 1 2 3 4

Table 14

DVC
Month Top-G1 Top-G2 Bottom-G2 Bottom-G1
June 1 2 3 4
September 1 3 2 4
December 1 3 2 4

VIII. Final Remarks

8.1 In this research we covered 25-year portfolios (1995-2019) in virtually all business days of June, September and December.  During such period, we followed 3 different variations of the deep value strategy, according to the market capitalization of the invested companies.  For each of such variations, we formed 4 portfolios.  In total, we formed 15,900 portfolios, if we consider each individual portfolio formed each day.  Putting this number in context, the number is not that huge.  In the end, we covered few variations (market capitalization and EY ranking) of the same strategy.  We still need to test such results statistically.

8.2 We think that some of the findings of our research are worth highlighting: (i) considerable difference of performance among the portfolios according to the EY ranking of the underlying shares; (ii) relative order of performance of the portfolios according to the EY ranking of the underlying shares; and (iii) poor performance of the portfolios formed with the most expensive stocks.

8.3 If we take into consideration the average 36 portfolios that we formed, in all of them there was a considerable performance difference between the ones formed with the cheapest stocks (Top-G1) and the ones formed with most expensive stocks (Bottom-G1).  In all cases, the Top-G1 portfolios performed much better than the Bottom-G1 ones.  Such portfolios represent both extremes of our universe, the best ones on the one side and the worst ones on the other side.  In this case, the performance difference was always large and in favor of the cheapest stocks.

8.4. Also, in most cases, the performance of the portfolios followed the EY ranking of the underlying stocks.  It means that Top-G1 portfolios (cheapest stocks) performed better than Top-G2 ones, that performed better than Bottom-G2 portfolios, that in the end performed better than Bottom-G1 ones (most expensive stocks). There were 3 exceptions to this order.  One of such exceptions were the DVA June portfolios, in which the Top-G2 performed better that Top-G1.  The other exceptions were in the DVC September and December portfolios in which the Bottom-G2 performed better than Top-G2.  We see no explanation for such performance difference.  Maybe the overlap of some stocks between the Top-G2 and Bottom-G2 in some years in the DVC portfolios influenced this result.

8.5 All Bottom-G1 portfolios (most expensive stocks) underperformed if compared to the Ibovespa.  In some cases, the result was almost the same (DVA December, DVB June and DVC June), but still with some underperformance.

8.6 In our opinion, the results above, with all their limitations, are encouraging.  Firstly, we covered a reasonable large period of time in our research.  Secondly, similar researches in other markets showed that the value premium is a pervasive phenomenon, thus it makes sense to find it in Brazil as well.  Finally, and more importantly, if you chose a common sense approach, it makes a lot of sense to buy cheap and keep some level of diversification.  If you buy cheap, you minimize your chances of error.  If you keep some lever of diversification, you have a second layer of error minimization.  Sometimes, better than being right, is not being wrong.  It is the right thing to do.

 IX. Bibliography

Berkin, Andrew L. and Swedroe, Larry E., Your Complete Guide to Factor-Based Investing (Saint Louis: Bam Alliance Press, 2016).

Buffett, Warren, The Superinvestors of Graham-and-Doddsville, text found at https://www8.gsb.columbia.edu/sites/valueinvesting/files/files/Buffett1984.pdf.

Carlisle, Tobias E., The Acquirer’s Multiple (San Bernardino: Ballymore Publishing, 2017).

Carlisle, Tobias E., Deep Value (New York: Wiley, 2014).

Gray, Wesley R., “Factor Investing is More Art, and Less Science”, in The Best Investment Writing, Volume 2, edited by Meb Faber, Harriman House, 2018.

Gray, Wesley R. and Vogel, Jack, Quantitative Momentum (New York: Wiley, 2016).

Gray, Wesley R., Vogel, Jack R. and Foulke, David P., DIY Financial Advisor (New York: Wiley, 2015).

Greenblatt, Joel, The Little Book that Still Beats the Market (New York: Wiley, 2010).

O’Shaughnessy, James P., What Works on Wall Street, Fourth Edition (New York: McGraw Hill, 2012).

Wilson, Timothy D., Strangers to Ourselves (Cambridge: The Belknap Press of Harvard University Press, 2002).

Annex I

Disclosure of Financial Information in Brazil – Dates

1. In January 1995, the rule in force regulating the disclose of financial information by public companies was Instruction No. 202/93, of the Brazilian Securities Exchange Commission (Comissão de Valores Mobiliários – “CVM”) (“ICVM 202”). Without going into details, at that time we had the Standard Financial Statements (Demonstração Financeira Padronizada-DFP) and the Quarterly Information (Informações Trimestrais-ITR).  ITR contains financial statements as well.

2. The DFP had to be made available at least 30 days prior to the Annual Shareholder’s Meeting (required to take place until April, 30 of every year, as per Article 132 of the Brazilian Corporations Law) (Article 16, II of ICVM 202). On its turn, the ITR had to be made available within 45 days after the end of each quarter, except the last one.

3. One exception to the rule above made the matching of market price and financial statements a little bit messy. In the beginning of 1996, CVM issued Instruction No. 245/96, that allowed companies with turnover lower than R$100 Million to release the ITR within 60 days after the end of each quarter, except the last one.

4. In late 2009, ICVM 480/09 revoked ICVM 202 and ICVM 245 and created an unified period of 30 days for the disclose of quarterly information (Article 29, II). This regime remained in force for less than two years and in late 2011, ICVM 511/11 created an unified period of 45 days for the disclosure of quarterly information.

5. In practice it meant that we had two main different deadlines to issue quarterly information, based on the turnover of a company. Figure 1 and Figure 2 below better explain this.

6. Under the ICVM 202 regime (and ICVM 511 as well) we could form 6 monthly portfolios matching past market prices with the then available financial statements (May, June, August, September, November, and December). On its turn, under the ICVM245 regime we could form only 3 monthly portfolios following the proper matching (June, September, and December). In view of the limited number of public companies in Brazil and to have a longer covered period, we decided to form only 3 monthly portfolios, in June, September, and December.  It means that we can cover the entire Brazilian market without distinction between the disclosure regimes and keep uniformity in our portfolio formation.

Figure 1 – ICVM 245 Regime:

Tabela- Post Deep Value Further Analysis and New Questions - ENG

Period A: during this period, we do not know if the then last financial statements publicly available were the ones from the 3Q or 4Q of the prior year.  It happens because the DFP might have been disclosed in any day from January 1 until March 31.  Thus, the risk of a mismatch between the financial statements then available to the public and the market price of a backtest is high.

Period B: during this period, we do not know if the then last financial statements publicly available were the ones from the 4Q of the prior year or 1Q of the current year.  It happens because the 1Q financial statements might have been disclosed in any day between March 31 until May 30.  Thus, the risk of a mismatch between the financial statements then available to the public and the market price of a backtest is high.

Period C: during this period, the then last financial statements publicly available were the ones of 1Q.  It happens because the date to present such financial statements had already expired (it was May 30) and 2Q is not over yet.

Period D: during this period, we do not know if the then last financial statements publicly available were the ones from the 1Q or 2Q of the current year.  It happens because the 2Q financial statements might have been disclosed in any day from June 30 until August 30.  Thus, the risk of a mismatch between the financial statements then available to the public and the market price of a backtest is high.

Period E: during this period, the then last financial statements publicly available were the ones of 2Q.  It happens because the date to present such financial statements had already expired (it was August 30) and 3Q is not over yet.

Period F: during this period, we do not know if the then last financial statements publicly available were the ones from the 2Q or 3Q of the current year.  It happens because the 3Q financial statements might have been disclosed in any day from September 30 until November 30.  Thus, the risk of a mismatch between the financial statements then available to the public and the market price of a backtest is high.

Period G: during this period, the then last financial statements publicly available were the ones of 3Q.  It happens because the date to present such financial statements had already expired (it was November 30) and 4Q is not over yet.

Figure 2 – ICVM 202 and ICVM 511 Regime:

Tabela- Post Deep Value Further Analysis and New Questions - ENG2

Period A: during this period, we do not know if the then last financial statements publicly available were the ones from the 3Q or 4Q of the prior year.  It happens because the DFP might have been disclosed in any day from January 1 until March 31.  Thus, the risk of a mismatch between the financial statements then available to the public and the market price of a backtest is high.

Period B: during this period, we do not know if the then last financial statements publicly available were the ones from the 4Q of the prior year or 1Q of the current year.  It happens because the 1Q financial statements might have been disclosed in any day between March 31 until May 15.  Thus, the risk of a mismatch between the financial statements then available to the public and the market price of a backtest is high.

Period C: during this period, the then last financial statements publicly available were the ones of 1Q.  It happens because the date to present such financial statements had already expired (it was May 15) and 2Q is not over yet.

Period D: during this period, we do not know if the then last financial statements publicly available were the ones from the 1Q or 2Q of the current year.  It happens because the 2Q financial statements might have been disclosed in any day from June 30 until August 14.  Thus, the risk of a mismatch between the financial statements then available to the public and the market price of a backtest is high.

Period E: during this period, the then last financial statements publicly available were the ones of 2Q.  It happens because the date to present such financial statements had already expired (it was August 14) and 3Q is not over yet.

Period F: during this period, we do not know if the then last financial statements publicly available were the ones from the 2Q or 3Q of the current year.  It happens because the 3Q financial statements might have been disclosed in any day from September 30 until November 14.  Thus, the risk of a mismatch between the financial statements then available to the public and the market price of a backtest is high.

Period G: during this period, the then last financial statements publicly available were the ones of 3Q.  It happens because the date to present such financial statements had already expired (it was November 14) and 4Q is not over yet.

Annex II

Factor Investing – Extremely Short Summary

1. Our aim here is very limited. We just want to give the readers a taste of what is a factor and the meaning of factor investing[53], to allow them to better understand what we meant when we said that the deep value strategy can be seen as a form of factor investing.

2. Maybe a good analogy is to think about your favorite fruit. Is it orange, lemon or strawberry? It doesn’t matter.  Let’s say it is lemon.  Think about the chemical research that managed to isolate the lemon flavor.  Probably, you already read a label of a lemon ice cream that said “it contains lemon flavor”.  It means that the chemical composition of the lemon flavor was included in the ice cream.

3. Oversimplifying things a lot, a factor is similar to a flavor. A lot of financial research took place to “isolate” the elements of different investment styles. An isolated element of an investment style is a factor. Basically, the research consists on forming portfolios with assets that have certain characteristics (small cap stocks, for example), and observing the performance and risk of such portfolios.  The behavior of such theoretical portfolios is the behavior of the factor.

4. Once a factor is identified, following that factor with real investments is factor investing. A factor may be compared to a flavor, but it is far from that. In the case of a flavor, with isolated chemical elements one can achieve the flavor, but in the case of a factor, one need actual investments to achieve the factor.  Here lays the most import use of a factor: it is a shortcut to form a portfolio.  Outside a factor approach, the formation of a portfolio involves a bottom-up research about each stock.  Using a factor approach, the formation is simplified, allowing a top-down selection of stocks.

5. The process to isolate factors created a form of financial golden rush to find the perfect factor. Indeed, several factors have been found, but very few of them proved to be real factors. A lot of research found pyrite (fool´s gold) instead of real gold.  Swedroe and Berkin summarize lots of researches about factor investing and identified the “true” factors out there[54].  They say a reliable factor has five characteristics.  The first is persistence, meaning the that factor has to persist for a long time and in different economic environments.  The second is pervasiveness, meaning that factor has to hold in different countries, sectors and assets.  The third is robustness, meaning that different versions of the factor have similar performance (such as value investing defined using price-to-book, earnings or cash flow). The fourth is to be investable, that means an investor has to be able to invest in it in practice.  The fifth is intuitiveness, meaning that the factor has to have a risk or behavioral explanation.

6. One of the real factors identified is the value factor[55]. The authors do not mention the deep value strategy in particular, but they covered several other variations of the value factor. Taking into consideration that the deep value strategy is variation of the value factor, we believe that pursuing the deep value in Brazil makes a lot of sense.  Also, the very existence of the value factor elsewhere gives us some comfort to actually implement the deep value strategy in Brazil.

End Notes

[1] The 2020 crisis is not covered here.  The effects it had in our real portfolio is found in the post Cigars, Diapers and Investments.  Another effect it had on us was to postpone the finalization of this post.

[2] You find a discussion of the margin of safety and research on it in Tobias E. Carlisle, Deep Value (New York: Wiley, 2014), pp. 20 to 28.

[3] See Warren Buffett, The Superinvestors of Graham-and-Doddsville, about a short description of the disciples of Benjamin Graham and their investment styles.  Text found at https://www8.gsb.columbia.edu/sites/valueinvesting/files/files/Buffett1984.pdf

[4] See Tobias E. Carlisle, Deep Value, pp. 35 to 51 about the evolution of the investment style of Warren Buffett.

[5] See Tobias E. Carlisle, Deep Value, p. 38.

[6] See Tobias E. Carlisle, Deep Value, pp. 43 to 51.

[7] See Annex II about factor investing.  In short, the “automatization” of the selection of stocks following a certain characteristic is a form of factor investing.

[8] Joel Greenblatt, The Little Book that Still Beats the Market (New York: Wiley, 2010).

[9] Joel Greenblatt, The Little Book that Still Beats the Market, pp. 166 to 168.

[10] Joel Greenblatt, The Little Book that Still Beats the Market, pp. 169 to 172.

[11] Joel Greenblatt, The Little Book that Still Beats the Market, p. 67.

[12] Joel Greenblatt, The Little Book that Still Beats the Market, p. 64.

[13] See Tobias E. Carlisle, Deep Value, p. 61.

[14] The tax incentive was created by Law No. 9,532/97 and extinguished by Law No. 12,973/14.

[15] Joel Greenblatt, The Little Book that Still Beats the Market, p. 168.

[16] We could calculate the Brazilian version of the Magic Formula keeping the goodwill value (and it looks like we will do that in the future), but we wanted to be as close as possible to the original formulas at least in the beginning of our work.

[17] See Tobias E. Carlisle, Deep Value.  Such concepts are developed in the entire book, a very short summary of them you find in pp. XI to XIII.

[18] See Tobias E. Carlisle, Deep Value.  The book has several examples of market action unlocking value of companies. In pp. 1 to 17 you find a short description of some transaction of Carl Ichan.  A summary on the subject you can find in pp. 205 to 213.

[19] See Tobias E. Carlisle, Deep Value, pp. 19 to 33 about different ways of seeing the meaning of margin of safety.

[20] James P. O’Shaughnessy, What Works on Wall Street, Fourth Edition (New York: McGraw Hill, 2012), pp. 30 to 30 and Wesley R. Gray, Jack R. Vogel and David P. Foulke, DIY Financial Advisor (New York: Wiley, 2015), pp. 31 to 47.

[21] Tobias E. Carlisle, Deep Value, p. 90.

[22] Timothy D. Wilson, Strangers to Ourselves (Cambridge: The Belknap Press of Harvard University Press, 2002), pp. 43 to 66.

[23] Timothy D. Wilson, Strangers to Ourselves, pp. 43 to 66.

[24] Timothy D. Wilson, Strangers to Ourselves, p. 23.

[25] Timothy D. Wilson, Strangers to Ourselves, pp. 46 to 48.

[26] Timothy D. Wilson, Strangers to Ourselves, pp. 46 to 48.

[27] Timothy D. Wilson, Strangers to Ourselves, p. 50.

[28] Wesley R. Gray and Jack Vogel, Quantitative Momentum (New York: Wiley, 2016), pp. 19 and 20, citing the work of Daniel Kahneman in the book Thinking Fast and Slow.

[29] Ibid.

[30] Wesley R. Gray, Jack R. Vogel and David P. Foulke, DIY Financial Advisor, p. 28 and pp. 49 to 53, James P. O’Shaughnessy, What Works on Wall Street, pp. 38 to 40, and Timothy D. Wilson, Strangers to Ourselves, pp. 93 to 115.

[31] Tobias E. Carlisle, Deep Value, pp. 77 to 97, p. 127, and p. 211.

[32] Wesley R. Gray and Jack Vogel, Quantitative Momentum, p. 24.

[33] Putting aside other issues, such as the liquidity of the invested stocks and the small market cap of some investment opportunities.

[34] Tobias E. Carlisle, Deep Value, pp. 77 to 97.

[35] Tobias E. Carlisle, Deep Value, pp. 210 to 213.

[36] The offer price was R$82.00 per share and the closing price of CGAS5 in the day in which the offer was launched was R$65.50.

[37] Wesley R. Gray and Tobias E. Carlisle, Quantitative Value (New York: Wiley, 2016), pp. 129 to 163.  Tobias E. Carlisle, Deep Value, pp. 53 to 75.

[38] Tobias E. Carlisle, The Acquirer’s Multiple (San Bernardino: Ballymore Publishing, 2017), p.69.

[39] Tobias E. Carlisle, The Acquirer’s Multiple, p.66.

[40] In some cases, when we were unable to have 20 stocks in a certain portfolio in a year, we included the maximum number of available stocks in such portfolio, increasing its concentration.  The minimum number of stocks in a portfolio we had was 13.  Also, in some cases there was some overlap of stocks in portfolios between the Top-G2 and Bottom-G2, mainly in the DVC portfolios. In a total of 15,900 annual portfolios the vast majority of them had 20 shares (15,462).  122 portfolios had 19 shares, 46 had 18 shares, 104 had 17 shares, 64 had 16 shares, 22 had 15 shares, 44 had 14 shares and 36 had 13 shares.

[41] In some cases, in view of the limited number of investible shares, there is some overlap between the Top-G2 and Bottom-G2.  The minimum number of investable stocks was 33.

[42] See notes 40 and 41 above.

[43] See notes 40 and 41 above.

[44] According to information available at Economatica, in a search for active stocks in Brazil.  Note that only 294 stocks had been effectively traded in the month preceding our search.

[45] Tobias E. Carlisle, Deep Value, p. 66.

[46] Joel Greenblatt, The Little Book that Still Beats the Market, p. 65.

[47] James P. O’Shaughnessy, What Works on Wall Street, p. 54.

[48] The values of market capitalization in this research are different from the ones in the 2019 research because now we verified the market capitalization in the first business day of June of each year.  In the 2019 research, we calculated the market capitalization in each May 20 of each year.

[49] This table correct a mistake in the values in the 2019 research.

[50] See notes 40 and 41 above.

[51] See notes 40 and 41 above.

[52] See notes 40 and 41 above.

[53] You can find a very good and accessible summary of factor investing in Wesley R. Gray, “Factor Investing is More Art, and Less Science”, in The Best Investment Writing, Volume 2, edited by Meb Faber, Harriman House, 2018.

[54] Andrew L. Berkin and Larry E. Swedroe, Your Complete Guide to Factor-Based Investing (Saint Louis: Bam Alliance Press, 2016).

[55] Andrew L. Berkin and Larry E. Swedroe, Your Complete Guide to Factor-Based Investing, pp. 61 to 86.

4 comments

  1. Que baita artigo! Durante essas quarentena fiz alguns testes, mas a base que eu tinha (comdinheiro) nem chega perto da Economática.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.