BNP Paribas’s Theam Quant funds suite of systematic alternative strategies is one of the broadest in the industry. It offers investors a granular menu of signal types, strategies and asset classes. It includes strategies based purely on fundamental data, those using only technical or price data, and those blending both inputs. It covers strategies focused on one asset class – such as fixed income, equities, or commodities, as well as multi-asset class strategies spanning four or more asset classes. It includes alternative risk premia/style premia/factor premia isolating one type of risk premium – such as volatility risk premia – and broader strategies trading multiple risk premia. Some of the strategies available on a standalone basis are also building blocks for some of the multi-strategy products.
“Theam Quant funds range builds on the foundation of derivatives expertise in its investment banking platform, and extends this structuring acumen to investors. The firm was an early mover in offering a purely quantitative stock-picking approach, and the ethos of innovation continues with ESG,” explains Vincent Berard of BNP Paribas Quantitative Investment Strategies team.
“The Theam Quant Europe Climate Carbon Offset Plan fund was one of the first to focus on the dynamic goal of energy transition, reducing future carbon footprints, to align with the Paris Agreement target of limiting climate change to two degrees, rather than just investing in companies that already have a low footprint. This leveraged the quantitative resources for integrating data and scoring companies,” says Roberto Bartolomei, Head of Global Markets THEAM Quant Fund Sales at BNP Paribas. Several filters are applied. After excluding companies involved in controversial sectors, such as armaments, nuclear weapons and energy, and tobacco, two sets of quantitative rankings – BNP Paribas Asset Management’s internal rankings and those from Vigeo Eiris – are used to screen the Euro Stoxx 600 and invest in firms with a high E, S and G score. The next step focuses on high carbon emitters, and identifies those with the best energy transition strategy, while also constraining the overall portfolio to ensure a 50% carbon reduction. These filters do not completely eliminate the carbon footprint however, since very few companies have a zero or negative footprint.
“The next stage is to satisfy institutions’ aspiration to neutralize as far as possible the carbon footprint of funds, and to do so in an innovative fashion. For the first time, a carbon offset mechanism has been incorporated into a UCITS fund through the purchase of VER – verified emission reductions. These are a voluntary carbon credit as opposed to the regulatory carbon credit market used mainly by oil companies and utilities. We invest in projects that result in a concrete social and economic benefit in countries such as Kenya, where the Kasigau corridor REDD+ project reduces carbon emissions by one million tonnes per year over 30 years, and has other environmental and social benefits. It furthers supports the production of sustainable charcoal, wildlife protection and ecofriendly garments. These projects generate the credits and are also used by our parent bank, BNP Paribas, for its own offset activities. The concept could be extended to long/short or equity market neutral funds, and indeed to those investing across a range of asset classes,” explains Berard.
Zero and negative interest rates increase the need to try and extract alpha from fixed income markets since coupon income can no longer be relied on. The FIND strategy, launched in March 2019, is based on an academic research paper published in the Journal of Fixed Income: ‘Beyond Carry and Momentum in Government Bonds’, which explored how factors such as defensiveness, reversals, market cycles and central bank actions can complement carry and momentum. This was co-authored by Jérôme Gava, William Lefebvre and Julien Turc, who heads the BNP Paribas Quantitative Investment Strategies (QIS) Lab. It investigated return drivers in government bond futures markets, for directional and cross sectional moves.
“FIND trades only G5 government bonds and interest rates in five markets: US, Canada, UK, Germany and Japan, with no exposure to second tier or peripheral markets such as Italy or emerging markets such as Brazil. The objective is to isolate interest rate moves and avoid corporate or sovereign credit risks, in contrast to many absolute return bond strategies that make much use of credit risk. The portfolio can take some short-term interest rate duration exposure, and is limited to maintain a low to medium correlation to the bond markets over the medium-term, mainly trading relative value between government bonds,” says Bartolomei.
BNP Paribas are of the opinion that directional trend trading works best at the short end of the curve, while the back end is driven by a wider range of factors, including monetary policy, inflation and commodity prices.
Therefore, short term interest rate positions – Euribor and Eurodollar futures – are traded long or short, based on a proprietary trend indicator, including cyclical economic trends, which in turn drive expectations of interest rates.
Meanwhile, long term interest rates are traded on a long/short basis, using a multi-factor model with three inputs: carry, momentum and relative value. The largest factor, with a 50% weighting, is carry and roll down; this factor sometimes overlaps with momentum. The momentum and relative models, each weighted at 25%, are expected to diversify one another not least because the relative value model is partly based on mean reversion. It is expected to perform better when central bank actions are coordinated, and rates are converging, as in the second half of 2019, or 2009-2012, whereas the momentum model would typically do better when central bank policies are divergent, such as between 2013 and 2016. The decorrelation between the two helps to smooth out returns and contributes to the Sharpe ratio above one on the back test. Overall, the long term strategy has fifteen possible positions in terms of factors: long or short of carry, momentum and value, in the five 10-year government bond markets it trades.
Returns in the high single digits for the first quarter of 2020 were already ahead of the usual target for 6% annualized returns, and the strategy deleveraged to lock in profits, and maintain the volatility target of 4.5%. The collapse in short term US interest rates generated a substantial opportunity for the strategy in the first quarter of 2020, followed by carry and relative value. Exposures and positions sizes shift potentially daily with the signals. In March, the strategy had spread trades such as long US government bonds and short Canadian government bonds. It also had a long UK versus short Germany position, and earlier in 2020 had the opposite exposure. As of May, the strategy’s positioning included a long US and short EUR government bond wager on a convergence between US and European rates. A curve steepener could benefit the carry factor. The FIND strategy ended the first half of 2020 up 4.42%.
“For the time being FIND is only offered on a standalone basis, though it may in future become a component of other multi-strategy funds. FIND can also be accessed through an unfunded swap,” says Berard.
From July 1st, FIND now incorporates some ESG criteria. To be eligible for the bond portfolio, countries need a democracy score (based on human rights; controversial weapons; gender inequality; freedom of expression; corruption, political stability and rule of law) above a certain threshold. There is also a cap on long exposure of countries that do not adhere to the Paris Agreement on climate change.
For the first time, a carbon offset mechanism has been incorporated into a UCITS fund through the purchase of VER – verified emission reductions.
Vincent Berard, BNP Paribas Quantitative Investment Strategies team
Theam Quant equity market neutral strategy – Equity World DEFI Market Neutral – trades single stocks on the long book and uses stock indices on the short side (S&P 500, EURO STOXX 50 and Nikkei 225), to reduce beta to zero. It had a relatively strong 2018, which helped it to win The Hedge Fund Journal’s UCITS Hedge award for Equity Market Neutral Global – Quantitative Best Performing Fund in 2018. “Its 2019 performance was negative for the global strategy as US allocations lost money; the purely European version of the strategy actually profited. The value factor has been the culprit of losses,” says Berard. “Value had underperformed the benchmark by a larger margin than the outperformance of the other factors, low volatility, momentum and quality, which have performed relatively well,” he explains.
In 2020 value and momentum suffered during the selloff, as is typical, and their losses had outweighed the outperformance of low volatility and quality in the first quarter. The strategy recovered in April partly thanks to overweighting the best performing sector, healthcare, and underweighting the worst, financials. It ended the first half of 2020 up 1.34%, outperforming many market neutral funds.
Theam Quant’s Factor Defensive strategy trades the same four equity factors, but instead of shorting equity indices it adds a systematic options strategy to reduce volatility. A short term call option overlay is designed to enhance yield while a long term put option reduces volatility and drawdowns.
“Going forward, the strategy is maintaining exposure to value, in line with its design. The value style is the most undervalued versus growth since the TMT bubble in 1999. Inflation could also help value stocks,” says Berard.
Theam Quant’s trend following CTA, Multi-Asset Diversified (MAD), has received The Hedge Fund Journal’s UCITS Hedge awards for its longer term performance namely Best Performing Fund over 4 and 5 years in the Systematic/Quantitative Macro Category for periods ending in 2018. The strategy has underperformed in 2019 and 2020, mainly due to having lighter positioning in bonds than many other CTAs or quant macro funds.
Theam Quant’s new AI CTA strategy, Multi Asset Artificial Intelligence (MAD AI), seeks to produce more consistent returns than trend following. In common with MAD it is purely based on technical price data, but uses artificial intelligence neural networks algorithms, including non-linear statistics, to identify hidden patterns and opportunistically switch between trend-following and mean reversion modules. In years such as 2008, 2013 or 2014 there could be a higher weighting in trend, while years such as 2010 or 2012 would have had heavier exposure to mean reversion. The algorithm inputs include variables such as lookback periods used to define trends and mean reversion: trends are measured over periods between one month and one year, while mean reversion is defined on periods of between one week and one month. The low correlation between trend following and mean reversion means that the combined Sharpe ratio is higher than either would be alone. The back test has generated a Sharpe of over 1.
The network learns continuously from equity, commodity and government bond market data dating back to 1996, and is based partly on cross-asset signals. For instance, equity and commodity market behaviour can be used to derive signals for bond markets.
It trades the same asset classes as MAD but trades somewhat fewer markets: five equity indices, three bond markets and one commodity market. In these markets, MAD AI can have the same positioning as MAD when trend signals are being followed but will have opposite positions when mean reversion signals are activated. “For instance, in March 2020, the strategy was long of bond markets, and short of commodities, both based on trend signals, but had used mean reversion signals for equity markets. There can also be geographic differences: the two strategies may share common positioning in US equities but might have opposite positions in European equities,” says Bartolomei.
Berard acknowledges that, “running more models and more parameters is a challenge, which is why there are safeguards against overfitting. These include splitting the datasets into in-sample training sets out of sample validation periods, and training the data over multiple economic and market cycles”. The systems also cap the numbers of models and parameters, reducing the number of inputs to 42. An ensemble approach, averaging results from the top subset of models, is also useful. And the strategy is not based solely on past experience. Its machine learning approach is adaptive in using algorithms to test multiple models, and learning to pick the models that work best under different market conditions. The MAD AI strategy ended the first half of 2020 up 4.14%.
Theam Quant’s short volatility strategies are intended to generate a return profile somewhere between investment grade and high yield credit, with regular income and occasional drawdowns. The return pattern is also intended to be smoother than equities, with faster recovery of drawdowns. Where premium income outweighs the costs of exercised put options, returns should be positive, and BNP Paribas has grouped markets into four scenarios to predict returns. A sideways market is perhaps most intuitively a positive backdrop, as is a long bull market, albeit the strategy will likely lag long only equities in a bull run. A slow and long equity bear market can also be positive for returns, since equities will not fall enough in any month to trigger exercise of the put options. The worst case scenario involves equities plunging fast enough to trigger the option exercise, as happened in December 2019 or March 2020, but recovery times have been often swift, for example after the Lehman crisis in 2008.
The US and European strategies are in principle similar, but are implemented differently. The Equity US Premium Income Fund targets premium income of 5% per year and aims for premium income to cover losses from option exercise. This was not the case in March 2020 but was in April 2020. “This is on the face of it a simple and well-known approach of selling puts on stocks. This earns a risk premium, and could perform particularly well when it is able to earn more premium by selling more richly priced options, but can suffer downside like a reverse convertible or autocall strategy when the options are exercised,” says Bartolomei.
But the devil is in the detail. The income is drawn from selling 5% out of the money, one month puts on a basket of US equities: the BNP Paribas Stock Put Write US Index, which selects 25 stocks using a smart beta approach. The criteria are a good business model, price momentum, low correlation, attractive option premium in terms of implied versus realized volatility, and liquid option markets. The objective is to avoid the more risky stocks that are more likely to result in the short puts being exercised. In 2020, an emphasis on technology and healthcare sectors (shared with Theam Quant’s aforementioned market neutral strategies) has contributed to outperformance.
“The main advantage of the Theam Quant strategy is its structured and risk-controlled implementation,” says Berard. One month maturities are chosen to optimize the time decay income. Execution is important to diversify exercise risk, and the strategy sells puts daily on the one month maturities, to smooth its entry and exit timing decision. This daily rebalancing maintains preset levels of exposure and smooths out the “pin risk” that can arise if portfolios are only rebalanced once a month. Thus positions, strikes and maturities are all well diversified. “Going forward, high implied volatility creates potential for the strategy to generate good income,” says Berard.
The Europe – Target Premium Fund is conceptually similar but is structured differently. It sells 5% out of the money puts on a stock index rather than on individual stocks. It is also using 200% leverage and targeting somewhat higher income of 7%. To balance out this additional risk, it spends about 10% of the premium received on 10% out of the money index protection, covering 33% of gross notional exposure. So far, 2020 has proved to be a “perfect storm” for the strategy, since the market gap has triggered exercise of puts. It has also started to recover strongly; as the unexpired put options moved into the money, annualized income reached 39% at one stage this year. “The speed of the 2020 crash made the market dynamics very different from 2008, when a more drawn out multi-month bear market made it possible to recover losses through selling more expensive options,” explains Berard.
Going forward, high implied volatility creates potential for the strategy to generate good income.
Vincent Berard, BNP Paribas Quantitative Investment Strategies team
A third volatility strategy, the Dispersion US strategy, is trading the volatility of stock indices (shorting a volatility swap on the S&P 500) versus their constituents (owning options on the largest 100 stocks in the S&P 500) to exploit a structural anomaly. “Index volatility is often overpriced because it is the natural way for a macro fund or real money investor to buy protection. Meanwhile, single stock volatility is often underpriced because structured products, such as reverse convertibles or autocallables, are natural sellers,” says Bartolomei.
The objective is therefore to be short of index volatility and long of single stock volatility, and to size the ratios so as to earn positive carry – or at least avoid negative carry. In this respect it clearly differs from a long volatility strategy that can incur negative carry. This is a relative value strategy that sizes each side vega neutral to avoid making a directional call on vega, or implied volatility. When there is a shock, the strategy performance may see a setback if index volatility jumps more than single stock volatility, but this sets up a strong dynamic for rebalancing the trade, which can make high returns as markets normalize. “Clients like a product with fast reactivity but in practice this is difficult to time, so holding it over a long period can make sense if the carry is flat or positive,” says Berard.
BNP Paribas are constructive on the outlook for this strategy, especially in the context of the US Presidential election – an event that has historically led to sector rotations. Additionally, correlations could fall if the trade war heats up again. Moreover, probable downgrades to earnings estimates would likely impact sectors differently.
THEAM Quant – Raw Materials Income fund is exposed to a long/short commodity strategy that aims to extract a form of income by creating positive carry from the contango and backwardation of commodity market calendar spreads. For instance, if a commodity is in contango, it will short the front month and own longer dated contracts. In March 2020 this profited from the spot oil price falling much further than the distant price, and also benefits from rolling down the curve.
In a backwardated market, it would go long of the front month and short the back end. A steepening of commodity curves in April 2020 has allowed the strategy to go short the front end in contracts between May and July 2020, and long the back end around December 2022.
“In theory if a commodity term structure is completely flat the strategy might not have any exposure. In practice it is quite rare for commodity curves to have no gradient,” says Berard.
It uses two sets of indices. The BNP Paribas DR Alpha Ex-Agriculture and Livestock strategy is a long index that either captures backwardation or reduces negative roll from contango, using the Commodity Curve Alpha signal, and invests via S&P GSCI Dynamic Roll mono-indices to access five energy and four base metal markets. The short book is the regular index. Raw Materials Income strategy was up 13.39% in the first half of 2020.
Investors should watch this space for multi-strategy launches.
Past performance is not a guide to current or future performance. For indicative purposes only. The value of investments can go down as well as up and investors might lose part or their entire investment. This does not constitute an offer or a solicitation to engage in any trading strategy or the purchase or sale of any financial instrument.