9 out of 10 asset managers don’t beat the market. This is driving the ‘Do It Yourself ‘(DIY) rush to passive. Below I explain how what should have been a 50% chance to beat the market became a poor hand for the asset managers and how the intelligent future is going to break the game.
Selection (S), Weight (W), and Holding (H)
The market is a basket, which includes everything investable. An asset manager's job is to look into the basket and make a selection and make a smaller basket also called a portfolio, mutual fund, hedge fund, etc. Once the selection is done, the asset manager has to assign a weight i.e. how much money should be allocated to the certain selection, 1% or 10%, etc. Once selection and allocation are done, the asset manager has to hold the selection for a certain time, so that he/she can book the profit (gain) on the respective selection. For this human skill i.e. SWH [1], an asset manager charges a 1 to 3% fee for an implicit understanding that his/her respective skill will work and the portfolio with the SWH will perform better than the overall universe, or a similar basket without the skill. Historically, since the early 1950s it has been observed that 9 out of 10 asset managers are unable to showcase the skill, and buying the basket without the asset managers help would have fared better and would have also have saved the 1-3% paid fees, and could incrementally add up to 50% additional gain [2] over 10 years.
Is this a generational problem?
Yes, it is a generational problem and it is worsening because humans are living longer, and interest rates don’t have much lower to go. We have reached peak credit and there is a limit to credit-induced growth. The asset owners who rely on asset managers to do a good job feel the pressure and the shift to alternative assets to seek illiquidity premiums, and search for negatively correlated assets is forcing asset owners to orchestrate their business and renegotiate fees and demand that they simply pay for alpha [3].
Why did you not hear about it?
If as an investor you are not aware of the statistic, now you know it and now you also know why there is a rush to crypto investing and generally investors taking charge of their investments, the whole rush to Robo-advisors is primarily driven by investors refusing to pay a fee for lack of skill. Not everybody is young, capable, competent, and has the luxury of time to manage his/her investments, and a relationship with an advisor may seem to prevail despite odds, but this could partially be also a function of the 14-year bull market. The complacency that accompanies secular positivity is not difficult to quantify. Any large drawdown or longer recovery can bring many advisory-asset management relationships under test, adding to this the ever-increasing compliance. We are talking about crunch points, stacked against the asset managers. The passive, zero fees, DIY passive investing is overwhelming the industry. In the era of intelligence, the search for alpha is critical for survival.
Where are the questions?
Why is this happening for more than 50 years of documented history? Why is this happening forever? Why is selection skill so hard? Even if markets were efficient, the skill should be on average near a flip of a coin, at ½ a chance? What are the asset managers distinctly doing wrong, to persistently underperform? Why is it so hard to see the mistake? How do you select consistently to lose? It seems like a game the market is playing against the asset managers.
What are we comparing against?
In the case of equity, the most popular asset class, the asset managers in the U.S. are generally looking at the broad market of top 500 stocks. The benchmark that tracks these components is the S&P 500. The world’s most popular index. Any basket construction is a function of its weightage methodology. How does it weigh its components? What measure does it use? The S&P 500 and most global benchmarks use the market capitalization weighting methodology, which gives the most weight to the biggest company, which would mean Apple at a weightage of 7% in the S&P 500 today. Another way to explain the market capitalization methodology is using the rich get the richer analogy. The more a company rises in price, the more it becomes richer, and the more the richer it becomes, the more it becomes important and weighty in the index. Hence rich get richer.
Why is it hard to beat, the rich get richer?
Or in other words, how can all the human intelligence in the world, with access to all the information in the world, and all the experience of 300 years of listed markets, fail the human asset managers? As I explained before, it is not a flip of a coin failure, it is an utter failure. The probabilities are stacked against the human asset managers. How did this happen?
To understand the ‘Why’, we can create a small market model which simulates a small portfolio. Let’s assume there is a human manager who prefers to select winners and wants to select a stock and put it in his/her portfolio. And to select, the human manager has to score, subjectively or objectively this selection from a set of short-listed candidates. The score should assist the manager to explain that stock belongs to one of the following four categories. 1 - It’s a winner, which should continue to win i.e. rich get richer (RGR). 2 - It’s a loser, which should continue to lose i.e. Poor get poorer (PGP). 3 - It’s a winner, which may lose i.e. the Rich get Poor (RGP). 4 - It’s a loser, which may start winning i.e. the Poor get Richer (PGR).
Image 1: Four Probabilistic States: Rich Get Richer (RGR), Rich Get Poor (RGP), Poor Get Richer (PGR) and Poor Get Richer (PGR)
The probability of a stock to be in any of the following four states [4] is ¼. This means that lack of a scoring method can already put a human manager at risk, because he/she may just have ¼ chance to identify the rich get richer selection. Assuming this human manager luckily and or with his/her skill can make the selection. He/she still has another challenge. The rich selection does not only have to get richer; it has to get richer faster than the poor gets richer selection he/she has not made.
RGR > > PGR (i)
Because the rich do not always grow faster than the poor, the S&P 500 portfolio which has the PGR selection, has an advantage over the sub-selection of other managers i.e. the human manager’s portfolio. We can assume the poor get poorer (PGP) and the rich get poor (RGP) selections don’t change during the period.
More complexity can be easily added to this simple model of market and multiple situations can be added to showcase that sub-selection have to compete with the complexity inside the master group (S&P 500 components) and the probabilities can keep stacking up against the human manager, who has an impossible hand to beat the broad market consistently, anyway he/she selects his sub-selection, the RGR way, or the PGR way. If we take a simple scenario of RGR having a ½ chance to be better than the PGR sub-selection, the overall probability of success is still 1/8 (¼ * ½). This is why human managers don’t beat the market, because the complexity of Intra S&P 500 components, is so high, that the human managers have no chance to beat it.
RGR >> PGR (ii)
RGR << PGR (iii)
Image 2: Sequence of selections at time zero (to) are hard to predict at a point in time (ti).
The RGR Mathematics
So, it’s not the RGR mathematics which fails the human manager, it is the challenger, called the PGR, which spoils the party. Just because market is growth biased, does not mean, owing FANG stocks will make you beat the S&P 500 consistently. You have to understand that the story of anticipation, which is intelligence, is algorithmically simple and it does not need all the information and computation of the world. It starts with rehashing the problem of whether RGR will beat the PGR or vice versa. Which in another way means that anticipation and intelligence to understand complexity is about understanding whether momentum, a certain trend persistence (continuing trend) will beat reversion (a change of trend) and vice versa. The dynamics of mean reversion and its failure is at the essence of anticipatory intelligence. Will the market (selected stock) keep going up or change trend. The interaction of the two forces of push and pull is what creates, noise, trend, failure of trend and everything related to nature of a living entity. And market is a living entity. This is why Keynes famously said, the markets can remain irrational longer than you can remain solvent, which is another way of saying, market can refuse to reverse and continue to stay in momentum waiting for you to lose your position and pride.
Apple did not just happen. It came from near bankruptcy and decades of lacklustre price. What was good for Walmart/IBM/ GM/GE was good for America, now what’s good for Apple is good for America. The second mover invariably replaces the first mover. A few scientists call it the luck factor (Speculative Science). There is a fundamental problem with the mathematics of RGR. RGR is a Siamese twin with the RGP. And the PGP as miserable it sounds, contains the magic beans of tomorrow, which take you to the clouds. So basically, it’s a probabilistic challenge. You simply can’t separate the RGP risk from the RGR and you can’t find the magic without suffering the PGP. And hence if you select from RGR only you have a significant chance to underperform over the long run, and if you do PGP, the RGR winners can road roll you. This is poetic logic of the S&P 500, which is why the compelling SPIVA statistics. Don’t try to beat us, just invest in us. There is a logical, mathematical, scientific solution to this quagmire, but that’s for another time. This feature is only about acceptance that Active management needs to transform or become irrelevant.[5]
Image 3: The probabilistic sequences are interconnected.
AI without the model is redundant
The idea of building AI without a probabilistic model is like constructing a building without a foundation. AI without a mechanism is a new method of creating new information, it has nothing to do with anticipatory intelligence. Machine learning without mechanism thinking is illusory intelligence, which cannot build market beating alpha processes and will stay befuddled by the market’s probability stack. Machine learning without a model is machine learning without purpose, noise enhancing, bias not behaviour seeking, risk increasing not risk reducing.
The Real AI
The real AI [6] has a chance to outperform the market, consistently and can break the game against the market. Though at some level, it is about understanding complexity and chaos, but the game is not against the market initially, it is against the 9 asset managers who don’t outperform. Beating human managers is where it begins, before the Real AI starts building Science, which can understand complexity, and anticipate chaotic systems like weather, sentiment, climate, flock of birds and jumping electrons. Such an AI does not just beat the benchmark, it first goes and finds a benchmark that is worth beating for the coming year. “Would it be commodities this year? Would it be Japanese market? Or would it be Vietnam?” The machine may first go about curating a mandate for itself before it goes about doing SWH on it. This is why learning machines would be disruptive because they don’t just become better at the game, they break it. Commoditizing alpha is not about beating the market, it is about showing that the best hedge fund of the world’s record at around 20% annualized over 30 years is beatable and it all starts with a mechanism.
How will active asset management survive these machines?
The good news is that nothing dies, it reduces in size, in clout, in relevance, but it does not die. So, the asset management power will reduce, assuming they are resistant to change and won’t transform with the changing times, and may continue to adopt an ostrich approach. The other good news, the asset managers, who are open, and keen to adapt would benefit at the expense of the managers not keen to adapt. The larger ones, will consolidate, the weaker ones will sell their book, industry consolidation will ensue. But there is a segment of asset managers who will thrive, because like everything else, crisis is an opportunity for the few, who dare to adapt and take risk. Change starts with acceptance. Finance is not facing irrelevancy; human asset managers are. The whole idea that physics is dangerous for your wealth, is not helpful when it comes to embracing evolving technologies.
Acceptance Number 1. Discretion is an art form. Bull markets bring Midas touch. Bear market brings financial scams. Fundamental investing is an art form for the less than 1% of the asset managers and not even 1% of those 1% are giving double digit annualized returns. There is a search cost for finding them. Most likely there is a performance fee and an unaffordable minimum ticket. Quite possibly they operate and cover Pan-Asian markets and not the overstretched North America. And even if you are lucky to find them, you may not be lucky and emotionally strong enough to hold on to them through their natural reversion cycle of underperformance [7]. Say you are persistent to hold on to them, you still need to be wise enough, to not put all your allocation with one manager. Extracting intelligence from information is an impossibility because information has been juiced for eternity. Everybody has access to 99% of it. The 1% that is privy enters the area of material non-public information. So either you use the Mosaic theory or generate meta information from information, it simply is too much computation and brain power for alpha generation, not an efficient process for investment managers and not cost effective for an investor. Discretion needs to rely on machines, if it has to stay competitive.
Acceptance Number 2. Finance theories are challenged, and Economics is not Science. The Nobel prize in economics have been given to two kind of groups. One group believes reversion is a reality, whatever you may do, it is impossible to make superior gains over the long term. This school is the ‘Efficient Market’ [8] School, which relies on the PGR and RGP school of thinking i.e. things always reverse and come back to the mean. The other school is the ‘Inefficient School’ [9], which goes about explaining why the ‘Efficient School’ is wrong and markets are inefficient, and rich do get richer and poor do get poorer and markets are not efficient. Both these Nobel prize winners are arguing from two ends of the same mechanism and hence financial and economic theory has been challenged as both these schools have not been able to demonstrate market beating performance. Factors, the mainstream of finance today, suffers the RGR, PGR fate. Factors work, fail in a chaotic way, this is why naive diversification [10] beats all factor investing. Financial theory is not the answer, if you plan to beat the market consistently.
Acceptance Number 3. Weak AI is as the name suggests, weak and narrow. Understanding complexity is not for the weak. You can’t master data science and think you have deciphered complexity. To beat markets, you need to make your model of complexity. What is complexity? How is complexity created? How does nature operate? How does it create noise? How does it sieve through it? If you are not interested in the broad strokes of nature, you are better off borrowing someone’s model thinking. There are a lot of models in history that you can use. But without the common sense of a mechanism, you are building something weak, which is going to frustrate you and it’s going to transform you into a statistic like the 300 years statistic behind you. Beating the market needs something systematic, scientific and replicable. If one of the elements are missing, you are playing the losers game.
Image 4: 30-year annualized performance. Best case, S&P 500 and a performance estimate for the rest of the active managers net of fees.
Innovation is not incremental. Innovation is dramatic. Dramatic is how change happens. Alpha will eventually get democratized when a passive investor would break the market better than the best hedge funds of the world for zero management fees.
Bibliography
[1] Pal, M. “AlphaBlock”, SSRN, 2017
[2] Investment Management Fees – Compounding in Reverse, Long View Asset Management Ltd., 2014
[3] "World’s biggest pension scheme overhauls fee structure", Financial Times, 2018
[4] Pal, M, “Mean Reversion Framework” SSRN, 2015
[5] "Fabozzi: Finance Must Modernize or Face Irrelevancy", CFA Institute, 2019
[6] Pal, M., “Human AI”, SSRN, 2017
[7] Goyal, A. Wahal, S. “The Selection and Termination of Investment", The Journal of Finance, 2008
[8] Pal, M., “Reversion Diversion Hypothesis”, SSRN, 2017
[9] Pal, M, “Is Smart Beta Dumb?”, SSRN, 2017
[10] Pal, M, “How Physics Solved Your Wealth Problem!”, SSRN, 2016