Strategic Risk Management
Strategic Risk Management

Strategic Risk Management

Active funds devote considerable effort to the search for excess returns, but risk considerations often fail to get anywhere near the same level of attention. (Location 437)

In the standard risk management approach, the focus is on setting volatility constraints associated with various targets and benchmarks. Such constraints often are based on the probability of a significant downdraft that could adversely impact the current investment strategy. (Location 442)

This book’s authors make the case that the position of a fund relative to its risk boundaries should be integrated into any consideration of investment shifts. The challenge is garnering sufficient incremental return from new investments to justify all incremental risks. (Location 451)

Significant risk events are likely to spawn a need for portfolio rebalancing. Most funds do rebalance at both prescribed time intervals and following sufficiently sizeable market moves. (Location 460)

This auto-rebalancing protocol is based on the presumption that significant market events do not seriously impact going-forward prospects. (Location 463)

This presumption is based on the belief that, over time, the market presents the same face to investors—both before and after major market moves. This equilibrium-mandated framework is comforting because it relieves fund managers of the need to peer into the clouded world of uncertainty and tease out revised policy portfolios. (Location 464)

They make the case that, rather than being based on a fixed periodic timetable, rebalancing should be more closely attuned to market conditions. For example, discernible changes in a market’s prospective risk should play an important role in shaping the rebalancing process. (Location 469)

For example, they show how volatility scaling provides a risk management function by reducing allocations when risks are increasing. They also study a range of investment strategies and assess how each strategy performed historically in times of crisis. (Location 473)

It appears that both volatility scaling and strategic rebalancing did serve to improve portfolio performance. (Location 477)

One of our core beliefs at Man Group is that risk management of portfolios is just as important as alpha generation. (Location 489)

In Seeking Crisis Alpha (Chapter 1), we write about a theme that is close to our hearts: the ability of time series momentum to produce strong returns in weak market environments. (Location 493)

We also show that time series momentum has some similar features to a long (put and call) options strategy. Aside from our momentum funds, we have directly used the protective feature of momentum in our long-only multi-asset programs. (Location 496)

We argued that put options are the most reliable, but most expensive, strategy, and that U.S. Treasury bonds have historically been unreliable. (Location 503)

Time series momentum and quality combine the attractive features of positive returns in both good and bad periods (at some reliability cost). We have, over the years, built solutions for clients utilizing equity options, credit protection, times series momentum, and equity quality to fit specific investor preferences. (Location 505)

Risk Management via Volatility Targeting (Chapter 3) takes a different approach to risk management by focusing on methods to keep asset and portfolio volatility stable over time. (Location 508)

Many systematic hedge fund strategies use some form of volatility targeting, while risk parity is one of the few long-only approaches to use this technique. We show that scaling positions by an expected volatility (using recent historical returns) produces more stable risk outcomes in all the asset classes that we study (i.e., reduced tail losses and more stable experienced volatility). (Location 511)

In Strategic Rebalancing (Chapter 4), we summarize several years’ worth of research on the impact of rebalancing on portfolio returns. (Location 516)

Rebalancing has many benign features, including the obvious risk balancing and less obvious return improvement. However, we show that rebalanced portfolios generally underperform buy-and-hold portfolios in extreme market environments where assets show strong momentum (because the rebalancing keeps buying the underperforming asset and selling the outperforming asset). (Location 519)

The chapter was prompted by a client remarking that rebalancing is a “short volatility” strategy, which caused us to start exploring the topic in much more depth and realizing the importance of active choices in rebalancing strategy. (Location 523)

We show that drawdown rules can be effectively used to weed out strategies (or managers) who lose the ability to generate alpha, and that this improves portfolio risk-and-return characteristics. (Location 527)

All of the tools we advocate are quantitative. In Man vs. Machine (Chapter 6), we look at both the risk and performance of systematic versus discretionary hedge fund strategies. (Location 530)

We found that discretionary and systematic macro managers are united in their long exposure to volatility, which can help in crises. For equity funds, discretionary managers have shown higher performance than systematic ones, but this difference is entirely explained by discretionary managers having larger factor exposures, especially to the market and size factors. (Location 532)

While we are clearly not out of this period of turbulence, we believe that our approach to strategic risk management provides some guidance on how to better manage risk through difficult periods. That’s half the challenge of being a portfolio manager, and often the more-overlooked half. (Location 542)

The idea of risk management is to provide some protection during adverse events. However, the cost of that protection must be balanced against the benefit. (Location 560)

By contrast, we focus on the idea of crisis alpha, which uses dynamic methods that lower risk and also preserve excess returns. (Location 562)

Trend following is one technique that works especially well with a crisis-alpha strategy. Theoretically, trend-following strategies sell in market drawdowns (mimicking a dynamic replication of a long put option) and buy in rising markets (mimicking a dynamic replication of a long call option). (Location 565)

Our starting point is a deep dive into time-series momentum (trend-following) strategies in bonds, commodities, currencies, and equity indices between 1960 and 2015. Over the last few years, institutional investors have turned to futures trend-following strategies to provide “crisis alpha.”2 (Location 571)

Performance was strong in not only the worst but also the best equity and bond market environments, revealing a well-known equity market smile and a lesser-known, but even more pronounced bond market smile. (Location 625)

While we start the evaluation of momentum strategies in 1960, our data start as early as 1950 to allow for a so-called warm-up period for obtaining the volatility and correlation risk estimates needed in the strategy construction. (Location 641)

Choosing equal weights is quite common in academic studies (albeit usually for dollar allocations), as it’s in a way a model-free choice. Any other weighting scheme would require justification for exactly how and why you deviate from equal weighting. (Location 874)

We also decompose the strategy returns into the performance from the four different asset classes: bonds, commodities, currencies, and equities. (Location 1108)

The performance of commodities displays more of a left skew, with performance particularly strong during the worst periods for equities and bonds. (Location 1111)

(i) using a 12-month rolling performance evaluation window (rather than 3 months) and (ii) starting the analysis in 1974, when we have data for currencies. In both cases, we find that the momCTA strategy does well in both the worst equity and worst bond market environments. (Location 1113)

Specifically, we run versions of the momCTA strategy where positions in equities are capped at zero. This will ensure that the strategy is well-positioned during periods of equity market decline (as it can never be long). Obviously, this will also ensure that during an equity bull market the strategy can only be flat or short (i.e., erroneously positioned in equities). (Location 1120)

For an investor who cares about both the equity and bond crisis alpha return profile, the situation is more nuanced, however, due to an unfavorable cross-effect. (Location 1144)

In this chapter, we have introduced a key component of strategic risk management: identification of active strategies that serve to cushion portfolios in times of stress. (Location 1163)

Despite 56 years of supportive empirical evidence, it is natural to ask whether momentum strategies will continue to be profitable. In this respect, it is worth noting that academic papers on the topic date back to at least Jegadeesh and Titman (1993) and performance has persisted since then. (Location 1174)

For example, Lou, Polk, and Skouras (2016) present evidence that stock momentum is different from other trading strategies in that professional, institutional investors tend to “trade against the momentum characteristic.” (Location 1178)

Many additional considerations play an important role when running a live momentum strategy on futures, including fine-tuning the trading signal definition, portfolio construction, risk management, and execution. (Location 1182)

While time-series momentum strategies tend to do well, on average, in periods of poor equity and bond performance, there are key questions that remain unanswered, in particular: (Location 1184)

In the previous chapter, we introduced a key concept in strategic risk management: integrating strategies that have favorable risk management characteristics into the asset selection process. (Location 1358)

Further, we showed that this protective property appeared robust to different economic environments since 1960. However, trend following is only one possible strategy that has protective characteristics. Further, it is important to diagnose the performance of various strategies in specific economic episodes. This is what we endeavor to do in this chapter.1 (Location 1360)

analyze the performance of different tools that investors could deploy. For example, continuously holding short-dated S&P 500 put options is the most reliable defensive method but also the costliest strategy. (Location 1366)

The typical investment portfolio is highly concentrated in equities, leaving investors vulnerable to large drawdowns. (Location 1374)

Time-series momentum strategies add to winning positions and reduce losing positions, much like a dynamic replication of an option straddle strategy (see discussion in Chapter 1).3 We show that such strategies performed well over the eight equity drawdowns and three recessions. We also explore limiting the equity exposure (i.e., no long positions allowed), which we find enhances the crisis performance. (Location 1400)

Finally, we show that futures time-series momentum strategies and quality long-short equity strategies are not only conceptually different, but also have historically uncorrelated returns, (Location 1411)

Once a drawdown has begun, the subsequent rolls of the options become more expensive as implied volatility rises, increasing the cost of the hedge. This effect then requires accelerated price decreases to produce the same hedge return. (Location 1500)

An equal-weighted combination of a long S&P 500 investment and the long put strategy has a negative excess return in each of the eight crises, as well as a negative overall excess return. Including the transaction costs of trading options (which are relatively high) would make the return of this strategy even more negative, underlining our observation that it is an expensive strategy. (Location 1506)

Long credit protection strategies have generally benefited during drawdowns as the spreads between corporate and Treasury bond yields widen. (Location 1515)

It does not provide a dividend, but, as a real asset, it can help offer protection against certain sources of long-term inflation. (Location 1585)

We introduce a simple futures time-series momentum signal, like we did in Chapter 1, as the compound return over the past days, scaled by volatility: (Location 1614)

For the purpose of analysis, we consider 1-, 3-, and 12-month momentum strategies to capture short-, medium-, and long-term momentum trading. That is, in Equation 1.1 is set to 22, 65, and 261 days, respectively. (Location 1625)

We divide the momentum score by the rolling standard deviation of security returns to calculate a risk-adjusted market target allocation. The strategy performance is then given by multiplying the market target (Location 1628)

allocations by a gearing factor and the next period’s return, and then summing across securities: (Location 1629)

We report the total return of the time-series strategies for equity drawdowns in Table 2.1 and for recessions in Table 2.2. The one- and three-month unconstrained strategies have tended to perform well during equity crises, consistent with the results in Chapter 1, which show that faster trend strategies are particularly good at providing potential crisis alpha, and during recessions. (Location 1653)

On the other hand, the 12-month unconstrained strategy has negative returns during the three most recent equity drawdowns (where the 2018 fourth-quarter selloff can be considered out-of-sample, per our discussion before) and performs notably less well during recessions. (Location 1657)

We now turn to a second active strategy, long-short U.S. equity strategies that use quality metrics. (Location 1769)

Also noteworthy for its return during equity drawdowns is the stock momentum factor, which in this case is traded at the stock level and in a cross-sectional (dollar-neutral) fashion, and so differs from the futures time-series momentum discussed in a previous section. (Location 1804)

In contrast, the value factor has been much less effective as an equity market drawdown hedge than the quality and profitability factors over our sample. (Location 1810)

The futures time-series momentum strategies (1-, 3-, and 12-month momentum with equity positions capped at zero) demonstrate negligible correlation with any of the quality stock strategies (profitability, payout, growth, safety, and the grand quality composite). Hence time-series momentum and quality stocks are complementary defensive strategies. (Location 1953)

This chapter had multiple goals. With strategic risk management, certain strategies are blended into a portfolio that lessen the pain in negative market environments. (Location 1989)

Two conceptually different classes of strategies emerge as credible candidates in our view. First, the futures time-series momentum strategies studied in Chapter 1, which resemble a dynamic replication of long straddle positions, performed well during both severe equity market drawdowns and recessions. Restricting these strategies from taking long equity positions further enhances their protective properties, but the cost is lower overall performance. (Location 2000)

Second, strategies that take long and short positions in single stocks, using quality metrics to rank companies cross-sectionally, have also historically performed well in equity-market selloffs and in recessions, likely a result of a flight-to-quality effect. We analyzed a host of different quality metrics, and point out the importance of a beta-neutral portfolio construction, rather than the dollar-neutral formulation that is more common in most published papers. (Location 2004)

Every crisis is different. For each one, some defensive strategies will turn out to be more helpful than others. Therefore, diversification across a number of promising defensive strategies may be most prudent. (Location 2011)

That is, the portfolio construction method itself is of vital importance. Also, the sizing of positions matters, and in the next chapter we show that downsizing at times of increased volatility provides another tool for managing risk. (Location 2015)

Understanding the historical track record of these investments is an essential component of strategic risk management. However, there are other tools in the investor’s arsenal that positively contribute to risk management, including rebalancing strategies (Chapter 4), drawdown strategies (Chapter 5), as well as the subject of this chapter, volatility targeting.1 (Location 2223)

A portfolio strategy that targets certain levels of volatility may act similarly to the positive convexity strategies that we discussed in the first chapter. (Location 2227)

For example, research has documented two features of volatility. First, volatility is persistent (sometimes described as clustering). High volatility over the recent past tends to be followed by high volatility in the near future. (Location 2228)

Second, for equity markets there is a negative relation between volatility and return realizations. As a result, a portfolio strategy that targets a certain level of volatility will be reducing weights in assets where volatility is spiking, which naturally reduces the severity of drawdown. (Location 2232)

Conditioning portfolio choice on volatility has attracted considerable recent attention. (Location 2238)

Recent studies show that a constant-volatility approach results in higher Sharpe ratios than a constant notional exposure. (Location 2240)

While most of the existing studies have concentrated on equity markets, we investigate the impact of volatility targeting across more than 60 assets, with daily data beginning as early as 1926. (Location 2243)

We find that Sharpe ratios are higher with volatility scaling for risk assets (equities and credit), as well as for portfolios that have a substantial allocation to these risk assets, such as a balanced (60–40 equity-bond) portfolio and a risk parity (equity-bond-credit-commodity) portfolio. (Location 2244)

Risk assets exhibit a so-called leverage effect (i.e., a negative relation between returns and volatility), and so volatility scaling effectively introduces some momentum into strategies. (Location 2247)

As previously mentioned, periods of high volatility are associated with negative returns and volatility scaling reduces losing positions—the same type of effect that one would expect from a time-series momentum strategy. (Location 2248)

We will show for other assets, such as bonds, currencies, and commodities, volatility scaling has a negligible effect on realized Sharpe ratios. (Location 2251)

The impact of volatility targeting goes beyond the Sharpe ratio; we find that it reduces the likelihood of extreme returns (and the volatility of volatility) across our 60+ assets. (Location 2253)

Volatility targeting also reduces the maximum drawdowns for both the balanced and risk parity portfolios. (Location 2257)

Following this are the analyses for the multi-asset balanced and risk parity portfolios. Finally, we discuss the leverage effect to provide further insights as to why the Sharpe ratio of risk assets is improved by volatility scaling. (Location 2261)

It is evident that volatility tends to cluster. In our data from 1926–2017, volatility is persistently high during the 1930s (Great Depression), the early 2000s (following the bursting of the tech bubble), and 2007–2009 (Global Financial Crisis). (Location 2406)

Both the vol of vol and left tail (mean shortfall) materially improve with volatility scaling, and the improvement is greatest for the most responsive volatility estimates. (Location 2451)

Hence, volatility scaling cuts both the left and right tails. Consistent with this, in Appendix 3.A, we show that kurtosis is much reduced when volatility scaling is used. (Location 2454)

The realized volatility of volatility-scaled returns is much more stable over time. This is also evident from the vol of vol metric (i.e., the standard deviation of the rolling one-year realized volatility) reported in the legend: 4.6 percent for unscaled returns versus (Location 2487)

Returns were less volatile pre-1980, and much less so pre-1967. Hence it seems that bond markets have gone through different volatility regimes, lasting multiple decades. (Location 2533)

Moving on to credit, we see in Table 3.7 a substantial increase in the Sharpe ratio when using a relatively fast volatility (10-day half-life) estimate. For slower estimates (longer half-lives) the situation reverses. (Location 2626)

In Figure 3.9, we show the autocorrelation of the monthly variance of daily returns for the 50 futures and forwards markets in light gray. The average for different sectors is superimposed in darker shades. (Location 2647)

Because of the aforementioned asset-level scaling to 10 percent volatility in all cases, the 60–40 split here is in risk terms. That is, 60 percent of the risk will be allocated to equities rather than 60 percent of the capital. This implies a dollar allocation lower than 60 percent given that equities are riskier than bonds. (Location 2686)

The Sharpe ratio, vol of vol, and expected shortfall (left tail) all improve from asset-level volatility scaling. (Location 2689)

In this section, we examine possible explanations for why volatility scaling improves the Sharpe ratio for risk assets, such as equities and credit, but has no effect on the Sharpe ratio of other assets. (Location 2762)

Our analysis suggests an answer that can be split into three parts: (1) only risk assets empirically display a so-called leverage effect, (2) the leverage effect effectively introduces some momentum, and (3) such a momentum overlay is beneficial for the Sharpe ratio. (Location 2764)

That is, negative returns tend to be followed by a reduction in the position size (as volatility is higher in that case) and positive returns tend to be followed by an increase in the position size (as volatility is lower in that case). (Location 2789)

The evidence suggests that time-series momentum strategies have historically performed well (Location 2809)

We find the shorter-term, one-month, momentum of returns to be most relevant here. Using a 20-day half-life for volatility scaling, the R-squared is 45 percent. (Location 2814)

In contrast, this chapter has focused on portfolio management strategies—that is, the dynamic management of the assets in the portfolio. The initial strategy we examined is volatility targeting investment weights rather than focusing on dollar weights. We contrasted the performance of both individual assets and portfolios with a constant notional exposure (unscaled) to strategies that target a constant level of volatility (scaled or volatility targeted). (Location 2826)

Our results show that this boost is specific to so-called risk assets (e.g., equity and credit) or portfolios that have a sizable allocation to these risk assets. (Location 2832)

While the Sharpe ratio is important, most investors have broader investment objectives. We show that volatility scaling has one unambiguous effect across assets and asset classes: It reduces the likelihood of extreme returns (and the volatility of volatility). (Location 2834)

First, the detailed analysis for equity and bonds was done for U.S. assets, for which we have the longest daily return history. (Location 2839)

Second, while the focus in this chapter was on volatility scaling, there are other methods with the potential to improve the risk management of a long portfolio. Chapter 1 shows that trend-following strategies tend to work particularly well at times of equity and bond market sell-offs. Hence a trend-following overlay may further improve the risk and return of a long portfolio. (Location 2844)

Finally, while we explored intraday data for S&P 500 and Treasury futures and found some benefits vis-à-vis daily data, we believe this chapter only scratches the surface of this topic. (Location 2849)

Volatility scaling for risk assets provides a type of protection that is similar to investing directly in positive-convexity dynamic strategies like trend following. (Location 2854)

We will argue there is a big gap between the beliefs of the benefits of rebalancing and what actually happens. Unbeknownst to many, rebalancing is an active strategy that buys losers and sells winners. Mechanical rebalancing increases—not decreases—the risk of a portfolio. (Location 2856)

In this appendix, we explore the following additional risk metrics to contrast unscaled and volatility-scaled returns: skewness, kurtosis, tail skewness, tail kurtosis, and (maximum) drawdown. (Location 2861)

we argued before, we believe investors likely care about the left tail much more than the right tail (for a given Sharpe ratio), rendering skewness or tail skewness (which are impacted by both) less useful risk metrics. (Location 2895)

To further illustrate that equity volatility clusters, we show in Figure 3.B1 (left panel) the autocorrelation of the monthly squared volatility (i.e., variance) of daily returns.26 The variance of adjacent months is around 0.6 correlated over the full (1926–2017) sample period. (Location 2927)

As was already visible in Figure 3.1, the middle subsample corresponds to a period with fewer extreme bursts in volatility in the first place. (Location 2932)

Bond variance is much more persistent, and in fact the autocorrelation only slightly falls from lag 3 to 12. This is just another manifestation of the various prolonged volatility regimes the bond market has experienced, which we discussed earlier. In contrast, equity markets experience volatility clusters of, say, half a year, but do not exhibit the prolonged regimes the bond market experiences. (Location 2938)

The previous chapter established that an active portfolio construction technique, volatility targeting, was particularly effective for risk assets in reducing downside exposure. There are two additional techniques that we will examine: rebalancing and drawdown control.1 (Location 3102)

However, we will argue that rebalancing is poorly understood. If implemented in a naïve, mechanical fashion, it can increase the risk of your portfolio. For example, if the equity market is in a prolonged selloff, rebalancing will be purchasing additional equity all the way down—increasing the size of the drawdown. (Location 3107)

So, obviously, an unrebalanced portfolio will eventually lead to the portfolio being undiversified by being concentrated in the high risk–high expected return asset. (Location 3113)

Under a mechanical rebalancing strategy, such as a monthly or quarterly reallocation toward fixed portfolio weights, winning asset classes are sold and losers are bought. If winners continue to outperform, that detracts from the portfolio’s overall performance. If losers continue to underperform, that also detracts from the portfolio’s overall performance. (Location 3117)

During crises, when markets are often trending, this can lead to substantially larger drawdowns than a buy-and-hold strategy. (Location 3121)

We show that time-series momentum (or trend) strategies, applied to futures on the same stock and bond markets, are natural complements to a rebalanced portfolio. (Location 3146)

As was mentioned in Chapter 3, the bond market was structurally different in the pre-1960 period and was characterized by unusually low volatility, reflecting the intervention of both the Treasury and the Federal Reserve. (Location 3153)

An alternative to a trend allocation is actively timing and sizing rebalancing trades, which we label strategic rebalancing. (Location 3161)

We explore different heuristics as well as trend-based strategic rebalancing rules and show that the strategic rebalancing rules are particularly helpful for reducing drawdowns for a 60–40 stock-bond portfolio. (Location 3175)

Large pension plans and sovereign wealth funds often explicitly target a fixed 60–40 asset mix. For example, the Norwegian Government Pension Fund Global in 2007 adopted a 60 percent target allocation to equities, with the remainder mostly invested in fixed income (see e.g., Chambers, Dimson, and Ilmanen 2012). In this section, we start by considering a two-period model to illustrate the difference between monthly rebalancing to a constant asset mix (Rebal) and buy-and-hold (Hold). (Location 3185)

In a multi-period setting, the return difference between a monthly rebalanced and buy-and-hold portfolio is similar to that of a short straddle written on the relative performance of stocks and bonds. (Location 3230)

The Rebal-plus-trend combination has a fluctuating stock and bond allocation, but no long-term drift. It has a slight long bias with, on average, a 5.3 percent long stock and a 4.7 percent long bond futures position, coincidentally almost exactly replacing the 10 percent reduction in the 60–40 rebalanced portfolio (see Table 4.1, Panel C). (Location 3391)

Now, we will study whether an investor can get similar benefits by smartly timing and sizing rebalancing trades, which we call strategic rebalancing. We consider both commonly used heuristics and trend-based rules. (Location 3403)

Typically, one varies the rebalancing frequency, takes a threshold-based method, or combines these two approaches. Also, rather than rebalancing fully toward the target asset mix, one can rebalance partially and so reduce turnover (and save on trading costs), providing yet another knob to turn. (Location 3406)

Threshold-based rules seem slightly less potent, where we consider rebalancing if the fraction of stocks is outside of the 60 ± 2 percent and 60 ± 4 percent range (but we also considered other ranges, which did not materially improve performance). (Location 3421)

In Table 4.3, we show results when rebalancing is delayed if the stock-bond spread trend is negative, positive, or continues to be in the same direction (i.e., to rebalance only when the trend direction now is in the opposite direction of a month ago, which likely corresponds to a not-so-strong or inconsistent trending environment). (Location 3425)

This is intuitive, as drawdowns typically occur when stock returns are negative and so a delay of rebalancing means not buying back stocks to bring the portfolio back in line with the 60–40 mix. (Location 3454)

The second portfolio management tool that can impact the severity of portfolio drawdowns is strategic rebalancing. While most investors rebalance, many do not realize that mechanical rebalancing can exacerbate drawdowns in market selloffs. (Location 3524)

On the other hand, a pure buy-and-hold portfolio is untenable for most investors as it leads to highly concentrated, undiversified portfolios. (Location 3526)

We show that the negative convexity induced by rebalancing is effectively countered with a trend exposure, which exhibits convexity and can be either implemented as a direct allocation to a trend strategy or with a strategic trend-based rebalancing rule. (Location 3529)

For example, investors can also use monthly in- and out-flows to move back toward the target asset mix. For example, Chambers, Dimson, and Ilmanen (2012) mention that the Norwegian Government Pension Fund Global directs monthly inflows into the asset class that is most underweight relative to the benchmark. (Location 3532)

We note that a stock-bond trend exposure is just one way to mitigate drawdowns at times of continued stock market losses. An investor has more arrows in her quiver. A good starting point is a more diversified portfolio that includes more asset classes and has an international exposure. An allocation to a broader trend strategy that benefits from trends in other macro assets at times of equity market distress may further dampen equity market losses (see Chapter 1). (Location 3536)

As shown in Chapter 3, volatility targeting can help manage the risk of a 60–40 stock-bond portfolio. We now turn to Chapter 5 where we show how the drawdown statistic can be used to make allocation and redemption decisions. (Location 3540)

One could argue that part of the improvement in the drawdown characteristics is because of the divestment of 10 percent of the rebalanced portfolio and is due in part to the defensive nature of the 10 percent trend allocation. (Location 3578)

A rule to hold off rebalancing when the stock-bond trend is negative tends to improve drawdowns, as it did for the 60–40 stock-bond portfolio considered before in Table 4.3. (Location 3630)

Chapters 3 and 4 focused on two portfolio mechanisms that serve to reduce the severity of portfolio drawdowns: strategic rebalancing and volatility targeting. (Location 3777)

The size of the drawdown is impacted by key parameters (or assumptions). We use the term “drawdown Greeks” to refer to the sensitivities of the probability of hitting a given drawdown. (Location 3793)

These are the key drivers of the maximum drawdown, and we identify the following: the evaluation horizon (time to dig a hole), Sharpe ratio (ability to climb out of a hole), and the persistence in risk (chance of having a losing streak). (Location 3795)

Drawdown-based rules can be particularly useful for improving investment performance over time by detecting managers who lose their ability to outperform. (Location 3801)

We also recognize that the timing of these replacements matters as a Bad manager can do more harm the longer they are managing assets. (Location 3805)

Investors face considerable uncertainty around the quality of the managers (or strategies) when selecting them. (Location 3955)

In order to have a more consistent rate of replacement through time, we now also consider a drawdown cutoff value that increases with time. Specifically, the cutoff is: (Location 4110)

In Figure 5.10, we illustrate the effect of a drawdown-based rule, where a risk reduction of 50 percent is triggered if the drawdown dips below a cutoff value. (Location 4145)

Full risk taking is restored if the manager recoups half of the losses. That is, they would have recovered the peak-to-trough loss if their risk had not been reduced by half. (Location 4146)

In reality, the two are complementary, where the relative weights on total returns versus drawdown depend crucially on the assessment of how likely it is that a Good manager can transition into a Bad one. (Location 4187)

Obviously, alternative criteria that may hint at a possible deterioration of a manager’s quality (such as turnover, fast asset growth, publication of their secret sauce) may provide a warning that an investor should start to place more weight on the drawdown statistic. (Location 4189)

We have set out here the foundations of strategic risk management. We began by detailing how allocation to particular strategies improved the risk characteristics of portfolios. Next, we detailed three portfolio management mechanisms, volatility targeting, strategic rebalancing, and drawdown control, which also serve to improve the risk characteristics of portfolios. (Location 4195)

There is one commonality to all of these ideas: they are all quantitative. Next, we will explore the interplay between discretionary and systematic asset management. (Location 4197)

In our experience, some allocators to hedge funds, including some of the largest in the world, either partially or entirely avoid allocating to systematic funds. (Location 4359)

These reasons seem to be consistent with a distrust of systems, or “algorithm aversion,” as illustrated by a series of experiments in Dietvorst, Simmons, and Massey (2015). In line with our experience and algorithm aversion, only 31 percent of hedge funds are systematic and they manage just 26 percent of the total of assets under management (AUM), as at the end of 2014. (Location 4362)

We find that systematic and discretionary manager performance is similar, after adjusting for volatility and factor exposures (i.e., in terms of their appraisal ratio). It is sometimes claimed that systematic funds have a greater exposure to well-known risk factors. (Location 4371)

In the second row, we report the amount of the return that can be attributed to well-known and easy-to-implement risk factors, based on a regression analysis. For discretionary funds, more of the return can be attributed to factors than for their systematic counterparts. (Location 4378)

Systematic macro stands out with an average risk-adjusted return of 4.9 percent. Discretionary macro has an average risk-adjusted return of 1.6 percent, while systematic and discretionary equity funds have similar values at 1.1 percent and 1.2 percent, respectively. (Location 4422)

Buffett’s performance can be largely explained by “exposures to value, low-risk, and quality factors” together with “a leverage of about 1.6-to-1.” (Location 4501)

While cross-sectional momentum strategies applied to U.S. stocks were well known before 1996, time-series momentum applied to futures has been documented only much more recently, and is therefore not included. (Location 4503)

In this third column, we also add dynamic factors. For systematic macro managers, there is a large exposure to U.S. stock momentum, which again can be understood from the prevalence of trend following in this category. Discretionary macro managers have a modest positive exposure to U.S. stock momentum, and also to FX carry. (Location 4627)

For discretionary macro funds, the average unadjusted return is 2.86 percent. Based on the baseline case specification, 0.74 percent of that is attributed to the equity factor, and also 0.74 percent to the bond exposure. The attribution to the vol S&P 500 factor is –1.28 percent, leaving an alpha of 1.57 percent after taking into account the smaller effects of other factors as well. (Location 4799)

For discretionary equity funds, the average unadjusted return is 4.09 percent. Based on the baseline case specification, 2.51 percent of that is attributed to the equity factor, leaving an alpha of 1.22 percent after taking into account the smaller effects of other factors as well. (Location 4988)

However, discretionary and systematic funds within macro or within equity are historically more highly correlated (in the 0.6–0.9 range). This suggests to us that discretionary and systematic managers’ investment strategies are more similar than one might think. (Location 5006)

Our results show that an aversion to systematic managers, as displayed by some investors, and in line with a more general “algorithm aversion” phenomenon, may be unjustified. (Location 5168)

We believe it is likely that some market inefficiencies are more suitable for a systematic approach while others are better exploited by a discretionary approach. Also, most of our analysis was for hedge fund–style index returns. (Location 5170)

A related paper by Chincarini (2014) compares performance and fees of quantitative and qualitative (as he calls it) funds. (Location 5215)

In the volatility targeting analysis in Chapter 3, we argued that sizing positions in proportion to volatility, rather than holding a constant notional exposure, creates a more balanced return stream. Empirically, in case of risk assets like equities, volatility targeting resulted in a higher Sharpe ratio of returns, correlating to a lower impact of volatility. In this chapter, we show that volatility targeting led to a reduced drawdown and higher cumulative returns for equities over the first quarter of 2020 as well. (Location 5484)

We note that the COVID-19 selloff in the S&P 500 was much faster than most other selloffs. Buying puts provided a good offset, but short credit risk was more potent with a +102 percent return (in excess of T-bills) over this period. (Location 5500)

we note that all time-series momentum (mom) strategies did well over the COVID-19 equity selloff period. As could be expected for a fast selloff, one-month mom did best. (Location 5523)

Of the various quality strategies, profitability held up well during the recent selloff, as did growth. Safety did not do well, particularly when implemented as a beta-neutral strategy. This is similar to results seen during the Global Financial Crisis in 2007–2009. We argue this is due to tightening credit conditions. (Location 5526)

It is perhaps surprising that the mom strategies did so well over the recent selloff, while the actual performance of trend followers over this period was slightly negative on average, albeit with considerable dispersion across managers, with some doing very well. (Location 5531)

Second, simpler trend strategies (like mom) seem to have worked better during this particular selloff. This is consistent with the performance of the SG CTA Mutual Fund index, which had the best results of the three. Third, trend followers typically employ slower models, and it is faster trend models that performed best during the COVID-19 equity drawdown. (Location 5546)

In practice, trend followers often employ moving-average crossovers (macs) rather than the simpler mom strategies. (Location 5562)

However, as can be seen in Table 7.3, during the recent, very fast equity selloff, mac models’ more gradual trading led to substantially lower crisis performance compared to the simple mom strategy. (Location 5579)

Across all nine equity market selloffs, trends in fixed income are most profitable. In the most recent COVID-19 selloff, a main driver of the performance difference between mac and mom strategies is that mom strategies do better in equity indices. (Location 5595)

In Chapter 3 we showed that sizing holdings in an asset to target a constant ex-ante volatility, rather than targeting a constant notional exposure, leads to improved risk characteristics. (Location 5615)

The first quarter of 2020 provides an interesting out-of-sample period to evaluate how volatility targeting performs. In Figure 7.4, we contrast the cases of constant notional (left panels) and volatility targeting (right panels) for U.S. equities. We use exponentially decaying weights for the volatility estimate with a half-life of 20 days. (Location 5621)

In the four lower panels, a constant notional exposure leads to a much worse drawdown, with the cumulative return dipping below –30 percent, while for target-volatility investing, the trough is only around –17 percent. (Location 5626)

When we extended the sample through the first quarter of 2020, volatility targeting significantly outperformed, reducing drawdowns by one half. The spike in volatility led to sharp reductions in allocation to risk assets—at the right time. (Location 5707)