Systematic Trading research and development, with a flavour of Trend Following
Au.Tra.Sy blog – Automated trading System header image 2

Betafication of Alpha: towards a Commoditization of Trend Following?

March 29th, 2010 · 9 Comments · Fund Review, Trend Following

"from alpha to beta" slide from A. Lo's presentation: "What will happen to the quants in August 2017"

Above is a diagram by Andy Lo, illustrating how alpha morphs into beta over time, as the initial strategy becomes more common/popular. Could the phenomenon affect Trend Following? And could the 2/20 fee structure charged by CTAs now be inappropriate?

This Beta of Managed Futures paper (PDF) from the Conquest Group hints that Yes, the original alpha provided by Trend Followers has taken the path down the beta-fication route.
Further along these lines the paper claims that Trend Following Managers sell Beta disguised in Alpha clothes:

CTAs: Alpha or Expensive Beta?

Returns can be broken down into alpha (excess returns due to skill), beta and the risk free rate.
Beta is ubiquitous and easily replicable and therefore cheap, while alpha is rare, hard to replicate and therefore expensive. It makes sense to pay a premium for alpha, but there is no need to do so to access beta, as it is available from multiple, competing sources. CTAs purport to provide alpha and justify their fees on that basis. It is clear, however, that there is a good deal of beta in the returns of many alternative investments, which raises the possibility that alternative investments are providing beta but charging alpha prices for it.

The Benchmark

In order to measure the baseline for Trend Following (and derive each CTA’s beta to it), the first step is to establish the benchmark and measure its performance.

Instead of taking an index of Trend Following CTAs, the authors build an index representing investing like a CTA. They do this by implementing a simple mechanical Trend Following system (Donchian reversal system) that covers multiple timeframes (20: from 5-day to 200-day) and a wide range of diversified markets (55 markets with exposure to all asset classes and geographical locations). The system also implements volatility-based position sizing and sector allocations (reflecting average allocation from a sample group of long-term Trend Following managers – possibly a flaw in the benchmark construction as it does not make it completely independent from the CTAs).

Correlating the benchmark to CTA indices

After applying fees, slippage and interest, the returns of the benchmark were compared to the returns of well-known CTA indices (S&P Managed Future Index, CSFB Tremont Managed Futures Index, etc.). The correlation between the benchmarks and these indices range from 0.75 to 0.9 indicating that the benchmark seems to do its job of… well, providing a benchmark for Trend Following.

Measuring CTA alpha

The next step is to calculate each CTA’s beta and alpha to the benchmark. The correlations are still fairly high (between 60% and 90%) and most CTA exhibit negative alpha over the testing period. For example, Rabar and Millburn’s calculated alpha since 1990 is around -2,000%.

Fourteen of the twenty largest CTAs failed to demonstrate alpha to the Conquest Managed Futures Beta benchmark from their inception until December 31, 2004. Six of these CTAs did demonstrate alpha over this period. Alpha, of course, […] represents merely unexplained variation which can represent either skill, random fluctuation or some combination of the two.

CTAs: Trend Following Wizards or Not?

It is not explicitly stated from the paper but it is safe to assume that the results from CTAs are net of fees. It is arguable which performance to use for comparison (before-fees or net-of-fees). It would be interesting to see a comparison of CTA returns with the benchmark on a before-fees basis.

Since most CTAs charge (much) more than a 1% fee, a before-fees performance comparison should prove more to their advantage. One could speculate that most would actually display some alpha to this benchmark.

2/20 Fee Structure obsolete for CTAs?

If CTAs can display before-fees over-performance to the benchmark despite their net-of-fees under-performance, it simply points to CTAs over-charging for their fund management services and that they must evolve to adapt to the new market environment. In order to become more competitive (from a pricing point of view) and stay attractive to investors, CTAs must keep producing net-of-fees alpha to the benchmark. This can either be done by:

  1. lowering fees
  2. improve their returns

Option 1 is surely the easier way, but understandably less desirable for CTAs – nobody said free competition was good for everybody (that’s probably why we have all the current interventionism and protectionism in global markets!). But as more competition arises, they will have to – in a Darwinian fashion- adapt to this betafication of alpha or die out.

A probably fair approach would be to differentiate the beta performance from the alpha performance of the fund and charge fees accordingly. Possibly a combination of a flat fee to cover costs of the beta-tracking strategy (similar to a passive management fund) and a variable fee to charge the true “value-add” from the fund (i.e. alpha). The latter fee could be based on the Information Ratio using the Trend Following benchmark.

Obviously, this is based on the assumption that there can be a widely-accepted and valid benchmark for Trend Following (including variables such as slippage, roll yield, asset selection and allocation, etc.); or each fund could possibly designate their own benchmark for this calculation – to be agreed with investors.

Other Implications

There are a few good points to take away, as an investor:

First, it shows that Trend Following Wizards‘ returns are reproducible with simple strategies. As discussed earlier, saving fees for yourself can give you a big headstart.

Second, for those not inclined to build their own system, there should be more products and funds available charging a much more competitive fee – when the launch of a Trend Following ETF?

However, this could also be bad news for Trend Following in general: if there is only a finite amount of alpha/profit available in the markets for Trend Following, and if the Trend Following space becomes more crowded there will be less opportunities for each player.

This is also in line with another Darwinian market theory, from Andy Lo: the Adaptive Market Theory (AMH), whose papers are an interesting read. The AMH theory tries to reconcile the Efficient Market Hypothesis (EMH) with its numerous flaws, as described by behavioral investing theory. The main premise of the AMH theory is that markets’ target and ideal model is the EMH model. However inefficiences (profit opportunities) arise from behavioral biases and as a result markets oscillate towards/away from the EMH model. The level of inefficiency in a market is related to that market’s “ecology” – with more competition bringing on more efficiency.
Could too much Trend Following kill Trend Following?…

Related Posts with Thumbnails

Tags: ···

9 Comments so far ↓

  • Anarchus

    Could you provide a citation for the Andrew Lo schematic? I’ve generally not been a fan of simple diagrams like the one shown because I don’t think they illustrate points very well.

    There is a nuance missing from the schematic anyway – which is, this:

    1. Assume a statistical anomaly is discovered by one quant group that provides real trading edge. That allows quant group one to earn One Edge Unit.

    2. Over time in a competitive market for ideas, other quant groups discover the statistical anomaly and begin trading it. IMHO, the immediate impact of this development is counter-intuitive – the Edge earned increases to 2 or 3 units as the weight of money flowing into the idea accelerates future returns into the present by gradually or even rapidly moving prices to equilibrium and then beyond equilibrium.

    3. The edge goes to zero.

    4. Oftentimes in the last stage, there’s a massive whiplash as prices/values which have been pushed past equilibrium make a quantum leap back in the direction of fair pricing. This is what happened to the major equity quant funds back in August 2008, for sure.

  • Monday links: dollar doubters Abnormal Returns

    […] Have trend following CTA returns become commoditized?  (Au.Tra.Sy Blog) […]

  • Jez

    @Anarchus – Sorry, I did look left and right, up and down, used all my google skills, contacted some fellow bloggers but could not locate that original presentation that Lo gave (to the Society for Financial Econometrics in June 08)… I originally found the chart on All About Alpha (great blog) and the link to the presentation there is broken.

    With regards to your example, your step 2 is counter-intuitive indeed. If you consider arbitrage for example, the more players going after the anomaly, the least it should be allowed to exist (ie less opportunity) and the faster it should be arbitraged away. I don’t really see how more traders would increase the arbitrage opportunities? Might be different for Trend Following as there is more of a feedback loop effect to it…

  • Anarchus

    Even in arbitrage, it’d happen there too, but on a much quicker timetable. Remember that in arbitrage a key element in the investment return is “how long”. In my counter-intuitive scenario, arbs would put their positions on at some initial spread of 5% and expect to earn a 21% annualized return by having the two positions meet in 3 months’ time. In my money inflow scenario, those positions might go to maturity (or hedges intersect, or whatever you want to call it) in only one months’ time, so the effective annual return goes from 21% to 80%.

    I had a friend who did all sorts of futures and option index arbs back in the late 1980s when it was easier, and he always said he made more return taking his arbs off early than he did with his entry points.

    My background being in stocks, though, step 2 definitely happens over a longer time frame, and I think I know something about the grand quant equity debacle of 2008. Historically, valuation factors such as low P/E and low P/B had been associated with above average stock returns, and stock price momentum (especially the 12 mo-1 mo variety) had been associated with above average investment returns. In approximately the 2003-2006 era, big quant funds such as AQR and Goldman got large inflows of fresh assets and started investing in stocks with low valuation and good stock price momentum. In the first 3-18 months or so of the money flows, this substantially accelerated the positive returns to the strategy.

    Historically speaking, it would be uncommon for investment returns to the valuation factor and to the stock price momentum factor to be highly correlated, but in fact that’s what started to happen in the 2006-2007 timeframe. I strongly believe that the rising and unusual correlation between those normally uncorrelated return factors was because of the “weight of the money” flowing into quant funds which were all essentially using the same strategy. With leverage, in many cases.

  • Jez

    Thanks Anarchus – interesting points of view, especially regarding the equity market. I suppose this is a global phenomenon that affect most markets and make their “efficiency levels”/profit opportunities oscillate (maybe a reason to look into trading regimes – which have been popular lately in the blogosphere)

  • Erik

    For lower cost access to CTA-like systems, LSC and RYMFX are decent choices. LSC is commodities only, while RYMFX adds financial futures (currencies, bonds, etc.). Like most CTAs, both are down recently, but RYMFX did decent during the crash of 2008. LSC is much cheaper, but is an ETN and is smaller/newer.

    A dividend adjusted chart is here:

    S&P provides the backtested/pro-forma performance of the indexes, which is good, but may be vulnerable to frontrunning someday, since the strategy is mechanical/disclosed.

  • D. Hom

    Step 2 seems to have accounted for the extraordinary profitability of Long-Term Capital Management when it began doing spread trades. Other funds jumped onto the same trades and narrowed the spreads, increasing LTCM’s returns.

    The spreads went the other way in 1998 when everyone was forced to jump off those same trades – step 4.

  • Jake

    14 CTA´s are disciplined, and 6 are not.

  • BlackCat

    Hi, very informative blog, I’ll comment here since I found this post so useful.

    I was looking to buy into a CTA Managed Futures Trend Following fund. It seems that:
    – They all do the same thing: For the wizards I could get data for, correlation of their monthly returns (log difference) to those of the SG Trend Index (previously Newedge) was between 0.6 and 0.8.
    – They charge a lot, usually 2 and 20. For retail investors there are many different feeder funds, but they either have high management fees (Fund of Funds), or use a swap, which hides the “2 and 20” and then sticks an extra layer of fees on top.
    – The main risk of picking a fund is that their algorithms stop working (eg: Dunn WMA from 2003 to 2006).
    – Morningstar seems to like AMFAX, with annual fees < 2% and a reasonable minimum investment.
    – US mutual funds can only be bought by US residents. So too bad, I can't buy.

    I'll move on to trend following on equities next – a whole new ball game.

Leave a Comment