Investment Strategies Based on Flow Signals

Dr. Sebastian Wenz and Martin Brückner

First Private Investment Management KAG mbH

March 2023

(This article first appeared in German in issue 1-2023 of Absolut|alternative)

Introduction

Capital flows are a diffuse and often only superficially understood concept in capital markets. There are two contradictory viewpoints on flows that one encounters in the financial marketplace. On the one hand, for many practitioners it is an obvious fact that flows affect prices. For the pre-trade analysis of stock orders in intraday trading, so-called market impact models, which predict the influence of an order on the market price with relative precision (usually as a function of the percentage participation in the trading volume), have been used for many years.[1]

On the other hand, this does not fit with the widespread academic view that share prices are determined by rational expectations about future cash flows (theory of efficient markets). Traditionally, financial market research claims that flows do not have a strong impact on market prices. This is because the efficient market theory assumes that supply and demand are highly responsive to price. Prices should therefore not deviate from the fundamental value, just because of large capital flows into or out of a security. So, it is not surprising that in a recent survey among 300 economists the consensus answer to the question “How much does the aggregate value of equities change if an investment fund allocates an additional $1B to stocks?“ was "zero".[2] 

Inelastic Market Hypothesis: Flows Move Markets

Nonetheless, financial market research has also recognized that there is a gap between academic models and reality that has not yet been bridged. In 2013, Robert Shiller won the Nobel Prize in economics for - among other things - research that showed that stock prices move much more than fundamentals justify: the Dividend Discount Model was not able to explain the empirically observed volatility in the markets[3]. This observation has entered the academic debate as the "excess volatility puzzle”. Since then, there has been a debate about a scientific explanation for the inexplicably high price fluctuations. Some researchers explain the volatility by rational changes in the discount rate or risk aversion among investors, whereas others look for the answer in behavioral finance and limits to arbitrage. From the perspective of the practitioner, there is another intuitive explanation: capital flows are not irrelevant for price formation, as in the theory of efficient markets, but rather flows move markets - especially in the short-term - much more than theory would suggest. Finally, there is academic support for this point of view in a controversial new research article by Xavier Gabaix of Harvard and Ralph Koijen of Chicago. Gabaix/Koijen propose an unorthodox, if not heretical, idea: The stock market is so inelastic that every dollar that is invested in the stock market increases the total valuation by five dollars.[4] Their "inelastic markets hypothesis" posits structural inelasticity as the answer to the long-discussed problem of excessive volatility. The core finding of Gabaix and Koijen in the context of capital flows is that most buyers of securities are highly price insensitive or inelastic. How can this be? In principle, one can assume that the global market portfolio is more or less in equilibrium. Every asset is owned by an investor at every point in time. Transactions take place between the marginal buyer for that asset and someone who owns that asset (and is willing to sell it). To know how this transaction affects the market price, it is important to know who needs to change their position and why. If the buyer has a higher incentive to trade, he must make a concession (in the form of a higher price) if he wants to motivate a seller to transact. How inflexible the buyer is determines the willingness to accept price concessions to get into the position. In other words, the degree of price elasticity is crucial for short-term price formation. For example, large buyers in the capital markets tend to be institutional investors who must implement strict constraints for their portfolio mix, for example to hold one hundred percent in equities or 70/30 in equities/bonds or similar. When an investor invests a dollar in such a fund ("inflow") to invest into equities, the fund must deploy that money according to its mandate. Fundamentals or market assessment have little to do with it.

 Current Trends in Market Structure

The financial press had also recently increasingly taken up the topic of flows; not least due to some of the peculiar features of the markets in the crisis years 2020 and 2021 (including the U.S. Treasuries sell-off in March 2020, stock market rally in 2020, meme stocks, etc.). So the flows triggered by fiscal and monetary policy - first Quantitative Easing in 2020, then Quantitative Tightening in 2022 - have certainly dominated the headlines over the past two years. But in the background, there are two megatrends whose dynamics have decisively changed the market structure over the last ten to twenty years.

First, there is the inexorable rise of passive and inflexible strategies such as index funds and ETFs, fixed volatility target strategies, target date funds and others. The result is that an ever-increasing share of all assets under management is invested in funds that operate with fixed rules regarding when and what to trade. Any inflow into a passive index fund must be invested immediately by its fund manager. This shows that completely passive investing is an illusion - these funds also act in the market (besides inflows/outflows, index changes are the triggers here) according to a fixed rule: every dollar of inflows is immediately invested in the index, regardless of fundamental valuations. There are many different statistics about the share of passive strategies in the overall market. A recently published paper by Alex Chinco of Baruch College and Marco Sammon of Harvard Business School shows that passive investors holdat least 37.8 percent of the U.S. equity market in 2020 - more than double the Investment Company Institute's semi-official estimate (which is derived from the size of the index fund universe). And that share will continue to grow if one looks at trends in inflows and outflows into passive and active funds.

The second megatrend with an impact on market structure can be observed with regard to the role of market intermediaries. Prior to the financial crisis, intermediaries of flows or liquidity providers were primarily banks (including proprietary trading desks). Banks, for a variety of reasons (including regulation such as the Dodd-Frank Act), have retreated significantly from this and have been replaced by high-frequency trading firms (HFTs) and hedge funds, which, however, operate much more flexibly and vary widely the liquidity they provide depending on the market environment. The result is a much more fragmented and volatile liquidity environment. This can probably also be seen as the main reason why the market structure has become more unstable, and we are seeing increasingly large (intraday) market movements. This situation has worsened again in the aftermath of the COVID-19 crisis. If we look at the market structure before and after the pandemic, one of the most striking changes is that market liquidity is much lower (over 50% according to an internal measurement by JPMorgan). Due to the lower liquidity, every Dollar someone buys or sells moves the market more than before.

 Classification of Flow Effects as a Source of Return

How can the flow effects described here be classified as potential sources of return? Simply put, there are only two ways to generate returns in the markets which, to avoid confusion, should not be mixed. One is alpha, a second is risk premiums flowing from capital users to capital sources – beta. 

Beta is the more accessible source of return; it involves buying and holding of assets that have positive risk premiums and includes investment styles such as long only equities, bonds, and multi-asset. Here, because of the relatively low risk/return ratios, very long investment horizons are required to be reasonably confident that the expected risk premium can be realized. However, these risk premiums are at least accessible to everyone.

Alpha, on the other hand, can also be seen as the difficult task of buying an asset before everyone else buys it and selling it when everyone else has bought it. Alpha is thus more difficult to generate because, unlike beta, it is a zero-sum game. To generate alpha, you have to beat other market participants - their loss is your gain.[5] As a rough classification, alpha can be obtained at three different time horizons, each with different signals matched to it:

1. High-frequency trading (HFT): In the ultra-short-term, market participants with the best technology, economies of scale and personnel in the field of high-frequency trading/market making can capture tiny spreads with very high turnover, mainly for providing liquidity. The investment horizon is sometimes in the millisecond range.
2. Short-term flow effects: between one day and one month, flow effects dominate the influence on the price development - only after that, fundamentals and risk premiums start to play a bigger role.
3. Classic, long-term fundamental analysis - the identification of undervalued stocks or informational leads in terms of fundamental data (e.g. corporate earnings). The optimal investment horizon to realize alpha here is usually in the range of one to twelve months, or even longer.

The shorter-term alpha sources are interesting mainly because they usually allow higher Sharpe ratios. The major drawback is that the capacity for capital allocations is inversely related to the investment horizon because trading costs for large orders quickly exceed the alpha. Longer-term fundamental-oriented strategies often have much lower Sharpe ratios because it takes time to establish the fundamental signal versus the "noise." And true to market wisdom, "markets can stay irrational longer than you can stay liquid", these strategies can be correspondingly more difficult to sustain. On the other hand, they have the great advantage that significantly more arbitrage or investment capital can be accommodated in these strategies.

Two Examples of the Implementation of a Flow Signal in Quantitative Systematic Strategies:

It stands to reason that the most sustainable advantages that can enable an investor to beat the market are found in its structure - in cases where market participants are predictably forced to act. Many different strategies have been found over time for the implementation of flow signals in purely quantitative systematic investment strategies. To illustrate this, two examples are briefly presented here, which are also currently in use:

• Market Neutral Equity Strategy Based on Flow Signals Derived from Passive Index Rebalancing Flow

As a stock selection criterion, a flow signal can be easily implemented in a market neutral equity strategy. In this example, the signal is the inclusion or deletion of individual stocks in widely followed stock market indices[6]. This happens in most stock indices on a regular basis, as well as ad-hoc in case of certain corporate actions. Around the so-called index rebalancing date, predictable flows into and out of certain stocks then occur, as index-oriented investors are forced to add the new candidates to their portfolios and sell the departing stocks. Due to the ever-increasing proportion of index-tracking managed assets, this usually leads to significant (temporary) price distortions. Stocks that are added to a popular index gain strongly before they are added, while deletions lose before they are dropped (Fig.).

Since the rules for the composition of most indices are publicly available, with access to the right data it is possible to implement a market-neutral long-short strategy that positions itself correctly in various indices in advance over the course of the year. In recent years, the strategies used by First Private have achieved a performance of approx. 7.5% p.a. with a volatility of approx. 5%.

• Long-only Equity Portfolio Insurance Incorporating Alpha-Signals from End of Month Flows

Turn-of-the-month effects have been known for some time in the equity space and have been academically documented in various out-of-sample studies. For example, in "Dash for Cash: Monthly Market Impact of Institutional Liquidity Needs" (2019), Etula/Rinne/Suominen/Vaittinen present evidence that the monthly corporate payment cycle causes systematic patterns in equity markets that coincide with end-of-month cash needs. In a risk budgeted equity strategy we implement, the integration of a flow signal based on this evidence ensures that when re-entering the equity market, we do not buy after a regularly predictable institutional flow but are already positioned in the market before the flow.[7]

FIRST PRIVATE QSL VS. MARKET AND VS. RISK BUDGED OF 20% p.a. 15 YEARS (2008-2022)

First Private Portfolio insurance QSL is essentially based on three pillars:

• Allocation of the risk budget to quarters
• Re-entry after stopping out, taking flow signals into account
• Path-dependent variable risk buffer per quarter with fixed annual risk budget

As seen in the attribution analysis, the flow signal historically contributed to more than one-third of the strategy's outperformance relative to a simple version with an annual risk budget approach.

Process for the Identification of Flow Signals

As the examples show, it is possible to generate alpha by understanding flows, predicting them, and then being correctly pre-positioned. Trading these flow signals means legally anticipating the flow by examining public data. It is not actual knowledge of the order. How then can signals be generated based on flow events? In principle, it is necessary to identify which assets certain investor groups hold and how they have or will behave in response to certain trigger events. A simple process for identifying flow signals can therefore be outlined as follows: 

1. Definition of different investor groups acting as homogeneously as possible (e.g. index funds, gamma hedgers or similar).
2. Identification of and focus on flows that are inelastic.
3. Forecasting the size of these flows: Analysis based on historical data - how did the identified investor group behave and what were the triggers?
4. In-depth understanding of flexibility in trading and the trading strategies that the identified investor group may employ to be executed.
5. Analysis of the extent to which this flow (has) affected market prices.

The analysis of flows and derived alpha signals becomes easier the more rule-bound and subject to constraints the analyzed flow or investor group is, and the more transparent the data

RETURN ALANYSIS-OUTPERFORMANCE VS. MARKT FIRST PRIVATE CAPITAL PROTECT AND ITS COMPONENTS

of the underlying decision-making is. An important prerequisite for inclusion in quantitative systematic investment strategies is a trigger that can be quantified in advance - usually a date or a trigger that can be modeled (volatility, price, etc.).

Categorization of Flows

Every day, there is a multitude of market-moving flows. To provide a systematic overview, flow effects can be categorized based on their underlying trading motivation as follows:

 1.      Hard Constraints:

• Hard benchmark constraints: for passive (index) funds due to investor mandate (index rebalancing); month-end rebalancing to fixed investment ratios for blend funds; vol control flow for funds with fixed volatility target/risk budget.
• Institutional or policy mandates: QE/QT, bond auctions by fiscal authorities.
• Margin calls: forced selling due to losses that have depleted risk capital
• ...

2.      Incentive Structures (monetary, regulatory, etc.):

• (Delta) hedging flows of banks incentivized to hedge market risk; hence, gamma, vanna, and charm effects are observed especially around option expiration dates.
• Stop-loss flows (which can lead to short squeezes in extreme cases).
• Systematic trend following/ CTA momentum models with fluctuating investment degree
• Futures rolls usually occur in the most liquid time window near the contract expiration date
• …

3.      Behavioral (Behavioral Finance):

• Performance chasing
• Fund unit redemptions/redemption flows due to negative performance
• Quarter/year-end window dressing.
• ...

These are all types of more or less predictable flows. Some flows are already optimized by the trading parties to minimize the market impact when trading, others are very inelastic. Therefore, when examining each of these flows, it is important to assess the degree of price elasticity on the opposite side individually.

Conclusion:

Flow signals are currently perhaps the greatest alpha opportunity in the markets. Not least for Wall Street banks, they are one of the last bastions of exceptionally high profitability in liquid market segments.[8] At the same time, the search for alpha with flow signals, like that for all potential sources of alpha, is no simple undertaking. The capital market environment can also play a role in the availability of the alpha opportunity. If capital and credit are easy to come by, there is a corresponding amount of arbitrage capital, so that the most obvious flow effects are largely arbitraged away. Or too many other market participants have predicted the coming flow and are now dependent on it coming, because otherwise they themselves become inelastic buyers - in which case a short squeeze can occur, for example. These are just some of the circumstances that have to be taken into account and require continuous monitoring of the relevant market structures and appropriate adaptation. In the field of quantitative-systematic strategies, this is therefore also a fertile field for the use of modern machine learning methods or AI - especially in the short-term trading, where large amounts of data are available.

Ultimately, flows not only fill a gap in understanding market movements in practical applications, but also in the gap between the theory of efficient markets and price movements that are too often seen as irrational. The conundrum of seemingly random movements in markets is replaced by the more tractable problem of understanding the determinants of flows in inelastic markets.

[1] Bouchaud, Inelastic Market Hypothesis: A Microstructural Interpretation [2022], Tóth et al.

Anomalous price impact and the critical nature of liquidity in financial markets. [2011]

 [2]  Gabaix/Koijen, In Search of the Origins of Financial Fluctuations: The Inelastic Markets Hypothesis“[2021]

[3] Shiller, R. J., Do stock prices move too much to be justified by subsequent changes in dividends? Am. Econ. Rev. 71, 421–436, DOI: 10.3386/w0456 [1981]

[4] Gabaix/Koijen [2021]

[5] With a few exceptions: there are only a few special market participants whose goals are not defined by the optimization of investment returns. They act in the markets for social or political reasons: Central banks, governments.

[6] Analyzed in detail, for example in “Buy High and Sell Low with Index Funds!” Arnott/Kalesnik/Lillian Wu (June 2018)

[7] First Private Portfolio insurance QSL is an innovative solution that adds some new ideas to the classic "risk budget per calendar year" approach, delivering clear performance benefits, much lower drawdowns and reduced volatility. The concept does not require the use of derivatives and is only invested in equities and cash respectively.

[8] ?…a trading desk that generates more revenue per employee than almost any other at Goldman. The unit, with about 20 people, crafts algorithms to help wager Goldman’s cash, raking in profits by Anticipating and reacting to changes in the world’s biggest stock indexes. (Natarajan/Abelson, “They Quit Goldman’s Star Trading Team, Then It Raised Alarms“, Bloomberg [02.08.2022])