From an article in Canadian Hedgewatch March 2013 titled “Observations On The Death Of Trend Following” by Brian Casselman:
Looking at S&P performance vs. that of equity managers, in 2008 the S&P returned -45% but from 2009 to date the S&P has rocketed about 130%. Yet many equity managers have not fared nearly as well in spite of the huge S&P tail wind; and during the four years following the stock market crash investors have fled stocks in droves while embracing managed futures mutual funds which are commonly trend following programs.
Investors are miss-stepping everywhere they tread, or so it seems. In light of such poor performance many commodity trading advisors (CTAs) and their panicked marketing agents continue to publicly argue that trend following is not dead. While the four to five-year drawdown that many otherwise successful commodity and currency managers have experienced continues to wreak havoc with long term track records, more and more of these battered traders are emphasizing that everything will be okay. Likewise, hedge fund managers representing various equity-based strategies believe they must stay the course and markets will sort out.
On the other hand, investors have increasingly come to believe that trend following is no longer profitable and their reasons are plentiful: market moving government rhetoric and intervention – with forced suppression of interest rates, central bank policy change, high-frequency trading, a proliferation of commodity and currency-based ETF’s, together with very large CTAs and hedge funds who throw around billions of dollars in the commodity space. So, what is contributing to such mass underperformance by so many managers?
Simply stated, trend following is at its core a long volatility strategy. In other words, it makes money when volatility expands (i.e., during trending moves). Conversely, it suffers frequent but small losses during nontrending periods in exchange for such infrequent but large gains. During non-trending periods the strategy attempts to tread water through the judicial use of stop loss orders until some market movement provides a large outlier move in which the strategy can profit.
While the complexion of markets frequently changes, since 2009 markets traded by CTAs have been in a prolonged period of non-trendiness combined with a great deal of volatility. Managers who trade US equities, while taking comfort in upward moving stock markets have also been subject to more volatile moves in both indexes and individual names.
Closer to home, Canadian equities have been full of noise. Let’s now review the underlying factors that are currently impacting investable markets with a goal of better comprehending manager strategy and limitations. Very few systems are profitable in all markets. Hence, it is generally agreed that a robust approach is far superior to an overly optimized one. Historically, together with prudent risk management the robust trading methodology has a far greater chance of surviving the rough patches. Thus, to make their programs truly robust many CTAs use 25 years of data. Further, they overlay risk management and manager-specific rules to the purely quantitative portion of the trading program.
During a conversation CTA Scott Stenger explained that “in my experience, I have seen situations where algorithms work until they don’t and then they blow up. From my perspective, algorithms can’t always adapt to new conditions as one can’t conduct a back test with similar conditions. Furthermore, Algo’s have to keep trading.” Such constraints on back testing and the inability to go to cash are common criticisms of quantitatively driven strategies. Some of the largest Quant firms readily admit that they don’t have an objective means by which to determine if their strategy is no longer profitable. Simply stated, it’s probably best to avoid most pure Quant approaches in difficult environments unless one can pinpoint the fail safes.
Next, the composition of price movements has altered during the past four or five years. One thing that is powerful to look at when comparing the large changes in performance from pre 2009 to date is the number of trending days2. Basically this is the number of days price is moving in the same direction on average across a set of markets. In short, 2009 had very few trending days. As a daily observer of the specific positions of several managers Greg Taunt further comments “I think when I see the big reversals that whipsaw managers forcing them to get out at the worst time before moving the other way the next day or, even the same day, you can see why there are struggles. We certainly haven’t had any clear direction since the end of 2008 as we lurch from crisis to crisis with the occasional bout of positive economic news.”3
Even more confusing, few very long term managers, who take a lot to stop out, as well as some shorter term managers, who very actively trade out of bad positions, are doing well. For example, one New York based CTA with a long track record found that their intermediate term trend following system was the best performer during 2012. But their short and long-term trend following programs lost money as did mean reversion and volatility breakout.
Such general comments emphasize that investors have to drill down. Quite unlike the early trend followers, not all current trend following programs are similar in composition. Various components or “overlays” of the core program make them unique. One really has to understand a manager’s strategy and its limitations. Perhaps most importantly, investors need to determine what performance results can occur by chance alone in the shorter run, without their being an underlying longer term problem. Then via simulation, the investor can allocate an appropriate amount of funds for a long enough period in order to let the statistical noise give way to profits.
Let’s look at the situation using a hypothetical trend follower as an example. In this case the manager’s buy and sell signals are tied to a 200 day moving average and stops are tied to a 20 day moving average. Now consider what would happen to this strategy’s performance in light of MA Capital’s observations of futures markets:4
1. Verifying Greg Taunt’s comments, when the past few years’ volatility of commodities common to most managed futures strategies is summed the size of daily and weekly moves has increased about 10%, compared to quarterly and annual moves.
2. Monthly and quarterly moves have also increased in relation to annual moves.
As the frequency of short term moves have increased more than long term moves, the added “noise” has created more stop outs for trend followers such as the hypothetical manager cited above, thus increasing short term losses considerably. MA Capital concludes that decreased performance of long term trend followers is “probably better explained by a relative pick up in short-term volatility as compared to the [changes in] long-term moves in markets.” This is certainly part of the picture. However, it is not quite as simple as looking at buy, sell and stop strategies – since shorter trend following and pattern recognition programs are also suffering, as are some reversion programs of the same periodicity. While one would think that increased shorter-term volatility would be good for active traders, certain markets continue to whipsaw and in some instances can also move rapidly for no apparent reason, like the rapid collapse of crude that occurred during Q3 2012 or the two week collapse of the S&P 500 in August of 2011.
Another good example is found in the currency markets. “The FX sector struggled in the shorter term timeframes in 2012. Due to the global central banks’ pivotal role, the FX markets remain the most susceptible to interventions and exhibit an extreme mean reverting character. This characteristic is self-reinforcing as it is supported by market participants trying to gain an edge, which means that it is likely to continue for the foreseeable future.”5
Thus, what may be a difficult time period to trade in one market may not be so in another market; we cannot generalize on the duration of time periods that may or may not be problematic. Investors must question which markets a particular manager is focused on, with which strategies, and then review the respective behaviour of such markets. Obviously, without such probing we can’t develop a more comprehensive picture of the range of risks inherent to specific strategies. Again, any strategy’s composition and limitations must be understood through the use of a comprehensive due diligence questionnaire and intense follow-up.
Let’s now dig a little deeper.
Richard Dennis is famous for teaching trend following strategies to a group of CTAs known as the “Turtles.” Throughout the 1980’s and to this day, several in this group became and still are the kings of trend following6. After returning many multiples of their investor’s original capital, for the most part since mid-2008 the performance of the remaining traders is essentially flat to negative. Additional factors that such trend followers are dealing with are:
Increased volatility that has caused more stop outs for trend followers near the end range of reverting price moves. And just as punishing, they also enter longs closer to the highs and shorts closer to the lows. Markets are not just noisier from volatility, but also more erratic, in so far as they unpredictably deviate from patterns we have previously seen. Take the stock market mini-crash of 2011 for example. Prior to the move tests by developers of hedge overlay strategies did not reveal the potential for a multi-day straight line dive in the indexes. They were crushed when weekly volatility far exceeded daily vol. In the “this time it’s different camp” it is likely that the majority of managers do not control for the role that new and unpredictable price patterns play in markets. Inter-market correlations have increased considerably and inconsistently. Correlations are now higher during large moves as a larger group (by numbers and AUM) of technical and fundamental traders get on board. Further, a move that starts in one grain market leads to the others, as wannabes jump on board. The risk-on risk-off concern we have been facing more recently has also increased the herd behaviour of what becomes an often large group. It is likely that similar influences are at play in equity markets.
The respective contributions of the above variables are not always simply an additive value. As there are several variables at play, different ones interact at different times in different markets to contribute to losses. We can’t simply sum the individual components and derive a risk number. It’s a large storm that’s hitting. For example, correlations between markets have been high but also are very unstable. As such, measuring correlations may not help assess risk – they could last weeks or months; it’s impossible to tell, so how can one change their trading based on this observation? Another interesting observation noted by Quest Partners and others is the asymmetry between profitable long and short trades. Over the last 25 years, particularly for the managers with larger AUM’s, long trades have been more profitable than short trades. One needs to inquire whether a given manager can perform both longs and shorts.
Quest concluded at the end of 2012:
“We enter 2013 expecting much of the same from the world’s central banks and fiscal authorities as they continue their efforts to stimulate with ever lower rates the moribund U.S. and Japanese economies and, to shore up the European monetary union. We expect programs will continue to perform in 2013 much as they did in 2012 until volatility suppression efforts are ineffective or abandoned and markets return again to a more free flowing natural state when systematic managed futures programs will again shine and demonstrate their value in a diversified portfolio.”7
Stanley Drunkemiller observes the same effects on equity markets from such sustained intervention. In a somewhat contrary stance, while the efforts of world-wide central banks and financial authorities push markets around, many years ago Robert Drach (and Robert Jr.) characterized the stages that stock markets go through during a recovery. This corrective process has not changed according to Drach Market Research. We are into a combined monetary infusion cycle, created by massive Federal Reserve and legislated stimulus actions. The central element of infusion cycles is additional money supply which eventually divides between increased productivity (public and private companies) and inflation.
Invariably, since creation of the current Fed, these cycles have resulted in significantly higher prices for productive assets, including common stock as the additional money works its way into the economy and stock market. Since the infusion cycle low (March 9, 2009), the S&P 500 has advanced about 130%.
Drach divides the ascent portion of infusion cycles into three parts.
1) A rapid, broad advance that ran from March to September 2009 with some of the worst quality companies rallying strongly.
2) A relatively lengthy period of volatile price gyrations in both directions with a net positive bias; this began September 17th, 2009. Since then, the market has exhibited gyrations in both directions with a net positive bias.
3) Gyrations are replaced by a more clearly defined trend-like advance ending with a speculative fluff.
While according to Drach we are currently phase two, value begins to assert itself as we progress through the phase. The transition from 2 to 3 is a function of internal pricing normalization, i.e. pricing relationships thrown into disarray by the recent huge recession and financial storm return to norms. While the normalization process can complete quickly or drag on, at this juncture (March 1st), Drach states that normalization is possibly close to completing provided that further massive monetary intervention does not delay it.9
Should Drach’s research again be proven correct, we could soon see more “normalized” moves in many financial and commodity markets, as well as the US stock market. On the other hand, remaining risks range from the macro effects of QE to the intentional and unintentional effects on the balance sheets of financial institutions, all with their potential to affect market trends.
We know that QE is a wild card; there is a significant probability that QE will continue to have diminishing returns, and there are significant unknowns regarding the effects of future Fed policy as we face issues never before seen (such as trying to unwind up to two trillion of dollars of assets)10. Currently, as PIMCO11 comments, interest rates at zero are forcing people into the equity markets, as there are no other choices. The Fed-created 50 to 75 basis point reduction in 10 year rates, translates into an 8% to 10% price subsidy for all financial assets, as equities price off of bonds. As such, PIMCO believes it will only take a small upward adjustment in rates to create dramatic reallocations to most portfolios.
So is it different this time? Are the observations I have noted an explanation for why the complexion of trends seems to have changed? While noting factors that may point to more normalized markets in the near future, Drach is also pointing us at the classic black swan condition through his above observations, i.e., events that have probabilities we can’t compute and consequences that are significant. Managers must avoid optimization now more than ever – which is always a temptation when markets become relatively stable. We can only make sound predictions in stable environments. With the VIX is hovering near all-time lows and the Dow at new highs as I write this, we are all looking over our shoulders for the next shoe to drop.
The more things change – the less weight we can place on the past – and most of the big blow ups are a consequence of naively applying statistical methods [or yesterday’s strategies] to a world housing more black swans than ever before.12 We now have a greater number of hedge fund managers than ever, with over ten times more assets than a decade ago; we all have broader and faster access to information, more computing power, etc. all of which have led to a reduced dispersion of fund returns. The talented innovators will always prosper but perhaps now, more than ever, the aggregate return profile of the mediocre will be more homogenous and more negatively skewed. In other words, it may be more difficult than ever to find truly skilled managers without embedded tail risk and to separate them from the pack.
Futures traders like to say “the trend is your friend.” Until trends return perhaps its best to keep your friends “closer” than your enemies – don’t simply read managers’ monthly reports, a rigorous program of ongoing due diligence is now more important than ever.
That’s it I guess. The evidence in my 4 books has been refuted by Scott Stenger (who?).