Equity beta cap

Though equity indices did not make a positive contribution in 2022, the absence of net long beta adjusted exposure to equities was beneficial in a year when equities ended negative but did not follow a path that was easy for all trend followers to profit from. The nature of the price action in 2022 was such that some trend followers could have been “whipsawed” by several small and large countertrend rallies during the year. 

After a bull run in equities in 2023 some CTAs are likely to have long exposure at least to US equities, which could be vulnerable to a sharp equity market reversal. ISTEC has an overarching equity beta cap (zero forecast beta) applied to the program which limits its long exposures, even after a bull run. Traditional trend followers without an equity beta cap (such as the sister program CFM ISTrends) has a fluctuating equity beta that can take the program equity beta, at any moment in time, in particular in bull markets, to high levels. 

Going forward, CTAs that can swiftly build short equity exposure have potential to offer some crisis or crash protection, if the world economy does have a hard landing, and/or valuation multiples compress, possibly because interest rates stay higher for longer. 

Proprietary trend indicators 

ISTEC is an absolute return strategy based on more than 30 years of experience of trend following; ISTEC uses CFM’s proprietary trend models, which are relatively parsimonious in terms of the number of parameters. 

Agnostic risk parity

ISTEC has outperformed the SG CTA index in the first half of 2023, in part by losing less in March around the collapse of interest rates associated with the mini-banking crisis. This is partly a consequence of portfolio construction techniques across CFM, which are somewhat more diversified and less directional. Whereas many CTAs only had short bond positions in March 2023, ISTEC also had some long interest rate exposures, adding up to a relative value bet. “If two instruments are heavily correlated, CFM may have a preference for being short of one and long the other in order to reduce risk while maintaining returns,” says Philip Seager, Head of Portfolio Management and portfolio manager of the ISTEC program at CFM.

This has always been part of the program. The portfolio construction methodology aims to equally weight orthogonal and independent principal components to create a more diversified out of sample return profile with more independent bets. Any portfolio can be viewed through the prism of Principal Components Analysis (PCA), whereby portfolio construction is based on uncorrelated principal component portfolios rather than correlated instruments.

CFM has named its methodology including these techniques “Agnostic Risk Parity”, applied them firm-wide, and written papers on the topic, such as the 2016 paper Agnostic Risk Parity: Taming Known and Unknown-Unknowns, co-authored by Seager and six others at the firm. This shows how Sharpe ratios have been enhanced by around 10% by allocating somewhat more evenly across principal components. 

This does not imply equal asset class weights, which do vary to some degree according to the correlations of the assets in the universe. Seager clarifies, “Any variations in the weighting scheme across assets serves to achieve the most robust diversification on a forward-looking basis”.

Sub-strategies 

ISTEC has three strategies: long-term trend following, a short-term trend overlay, and long volatility, which make different contributions to returns during various sorts of market environments, corrections and crises over different timeframes. Around the Covid crisis, the short-term model and the volatility model both made sizeable contributions to the overall program, which made 8.3% in March 2020, and 13.2% over the whole of 2020.

Recently, the long-term trend following strategy, following trends between 7 and 11 months, has been performing best compared to shorter trend timeframes. 

Short-term models

ISTEC’s short-term models, trading timeframes around 2 months focused on the most liquid bonds and equity indices, did not in fact get long of bonds in March 2023, because the speed of the move was too fast to capture. 

VIX strategy 

The ISTEC VIX model, which holds a long futures position in the VIX index of implied volatility (hedged with long S&P 500 futures) detracted from returns over 2022. The slow drift down in markets combined with a low VIX led to negative performance – VIX accelerations need to outweigh S&P 500 declines for it to profit. “The expected Sharpe ratio from this strategy is near zero anyway, but this is better than the negative expected return from buying options, which have a much heavier negative bleed. The negative drift is well below that of options,” points out Seager.

Execution efficiency

Regardless of trading timeframes, CFM aim to optimize entry and exit price levels. ISTEC trades a similar universe of liquid markets in FX, long term bonds, short-term interest rates, credit indices, equity indices and commodities, to other CFM programs and benefits from the firm’s technology and execution teams and infrastructure, and investment into execution research since 2000, which has also led to academic papers on market impact published on the CFM website and elsewhere. CFM are experts at measuring, modelling and minimizing these costs. “As liquidity keeps going up, we have only ever seen our cost of trading go down. We have not seen any adverse impact from HFT style liquidity. We believe HFT reduces spreads even for retail investors, so that everyone benefits from improved liquidity,” says Seager. 

Cost modelling has not always been accurate in the past when costs of faster trading speeds were sometimes underestimated. “Much progress has been made in cost modelling over the years. Importantly the upward drift on transaction costs from rising assets under management has been controlled. Our costs have not gone up. They would have gone down without asset growth,” clarifies Seager.

Proprietary algorithms

CFM aggregate the signals, net them off, and then implement trades to try and minimize transaction costs, and maximize cost adjusted gains. Short-term trading predictors and algorithms are intended to minimize costs and measure slippage, which feeds back into the simulations. “We also benchmark our costs versus broker and off the shelf algorithms and find that our execution algorithms perform better than broker ones. We build our own algos and were early adopters of direct to broker and/or exchange trading back in 2002,” says Seager. 

Overall, CFM’s low latency network has direct access to over 130 exchanges and other trading venues or platforms. Everything except FX is traded via futures. “The FX market is more fragmented, but it has evolved a lot over the years, with major FX players including the interdealer market, ECNs and liquidity providers. ECNs have become the most widely used venue style for trading and we have adapted accordingly. There is no single dominant venue for spot FX,” explains Seager. 

“We have been honing and refining our execution for over 20 years. It is valuable at this level of frequency, even if costs as a percentage of assets are smaller for ISTEC and ISTrends than for other faster trading programs such as Discus,” says Seager.

Capacity, vehicle and fee choices 

CFM had assets of USD 11.9 billion² as of October 2023 and some strategies do have capacity constraints, but both CFM ISTrends, with assets of USD 739 million³, and ISTEC, with assets of USD 630 million³, have plenty of spare capacity as they are highly scalable. 

The UCITS fund structure delivers a 10% annualised volatility. There is a choice of fee structures: it is possible to pay a higher management fee and no incentive fee, or a slightly lower management fee with an incentive fee. Managed accounts can offer further scope to customise risk targets and fee structures.