Nipun Capital, a minority and women owned business enterprise based in Foster City near San Francisco, was conceived in 2011 to implement its founders’ vision to be an institutional provider of systematic, fundamental Asian equity exposure, initially running Asian long/short equity and later adding Asian long only equity and China long/short equity, before broadening the investment universe in 2016 to include other emerging markets. The strategies have in 2020 and in recent years performed well in contrast to some other quantitative approaches that have fallen prey to generic factor exposures and violent factor rotation.
The name Nipun, which means “skilled” in Hindi and “standing on a solid foundation” in Chinese, combines the ancestry of the founding partners.
The name Nipun, which means “skilled” in Hindi and “standing on a solid foundation” in Chinese, combines the ancestry of the founding partners: one is Indian and two are ethnically Chinese. Co-Founder and Portfolio Management Partner, Pooja Malik, who grew up in India, says, “quant investing and entrepreneurship is in my DNA: my father retired early from the Indian army to start his own firm in the 1980s, investing in Indian equities using a quant approach. He imported Osborne computers and installed them in our backyard! I took to coding at an early age”. However, she got into quant investing somewhat by a happy accident. After earning an MBA in Finance from the Indian Institute of Management in Bangalore, and spending three years working for Indian bank ICICI and the Indian School of Business, she moved to the US and initially joined Barclays Global Investors (BGI) in the corporate strategy team. “I soon discovered I had a natural talent, intuition and capability for quant investing. I love what I do and wouldn’t trade it for anything,” she says.
Her co-founders and colleagues share this passion and Nipun’s core team of senior portfolio managers, researchers and the Head of Operations also worked together at BGI. Most of them have been at Nipun from the start, which means they have worked together for more than 20 years. The team have forged their own fiercely independent corporate culture. “It can be harder to hire from other organisations that have a different mindset or who are accustomed to working in a very siloed organization,” reflects Malik, who featured in The Hedge Fund Journal’s 2016 ‘Tomorrows Titans’ report published in association with EY.
The three partners all have extensive experience in systematic equity research and now have complementary roles in portfolio management, risk and research at Nipun. Malik worked at BGI, from 2001 to 2010, rising to Managing Director in 2008 and became Head of Portfolio Management for North American equities including the US and Canada. Nipun’s Research and Risk Management Partner, Dr Ken Hui, who rose to Head of Equity Risk Models at BARRA before joining BGI, had covered European equities, ran Canadian equities, and latterly Japanese equities during his time at BGI between 1999 and 2010. Meanwhile, Research Advisory Partner, Dr Charles M.C. Lee, had led research at BGI between 2004 and 2008, before leading US and global equity, defining the agenda for all active systematic strategies. Lee is a tenured professor at Stanford and works for Nipun part-time in an advisory capacity. “He edits and reviews articles for many journals and also has a role at Beijing University that places him close to the academic network in that region. Lee is the bridge between Nipun and the academic community, alerting us to many ideas quite early on. This gives us a competitive edge as many insights in quantitative investing have come from academic research and having access to early stage working papers before they are published enhances our advantage,” points out Malik.
The trio, who ran the gamut of equity strategies from long only to 130/30, long/short and market neutral, at BGI, set up Nipun in 2011 with the blessing of former colleagues after BlackRock took over BGI in December 2010. Launching Nipun’s first fund in May 2012 with $18 million of mainly partner and friends and family money, the partners have always eschewed seed or acceleration capital deals for fear that sharing revenues would take resources away from research. “For several years we reinvested all profits into the firm. The management company has also borne some costs of new vehicles in order to keep their expense ratio at a reasonable level,” says Malik. This plucky spirit has also been inspired by her family history: “My father died when I was 12. Being brought up by a single mother with 3 kids – and my mum had never worked outside the house until then – taught me a lot about grit and resilience”.
The core investment philosophy, based on Shiller’s behavioural finance model, is that short term asset pricing deviates from long term fundamental value or intrinsic value due to sentiment and noise trading. This means that trading dynamics are very important and Nipun specifically targets more neglected and inefficient markets as being richer sources of alpha. “We think we could generate returns in developed markets, but we believe there is more alpha in less efficient Asian and emerging markets,” says Malik. Since Nipun started, the investment universe has included twelve Asia Pacific markets: Australia, China, Hong Kong, Indonesia, Japan, Malaysia, New Zealand, Philippines, Singapore, South Korea, Taiwan and Thailand. China overseas and onshore listings and Taiwan make up over half of the exposure in the Asian long/short fund. The universe includes small caps and mid-caps, which attract less institutional interest and buyside coverage than large caps in Asia. Nipun finds higher alpha in these under-owned, non-benchmark stocks which are typically neglected by foreign investors.
For instance, Malik discusses the stark comparison between Amazon and momo.com (a Taiwanese e-commerce company): “There are over 56 analysts that cover Amazon. Only 2 global sell side firms cover momo.com and 7 local brokers, who are not widely followed by the global investor community. Institutional holding in Amazon is over 62% with over 4,500 institutions holding the stock. In contrast, only 16% of the stock of Momo.com is held by institutions with only 91 institutions holding the stock”.
She continues: “The inefficiency in Asia can also be seen on an aggregate statistical basis. In fact, even within Asia, there is a contrast between developed and emerging markets. For example, in developed markets like Australia and Hong Kong, almost half of the listed stocks have at least three analysts covering them whereas in markets like China A, Korea and Taiwan, only 20% of the listed stocks have greater than three analysts covering them”.
Asia and emerging markets provide a rich canvas for alpha generation. These markets are dominated by retail investors who are more susceptible to behavioural biases and do not pay enough attention to company fundamentals. As a result, there is larger mispricing in these markets. Nipun goes further to identify opportunities based on the idiosyncrasies of each country. Their research is bottom up, country specific and is able to identify localized insights that drive alpha generation. In 2020, the investment universe for the long/short fund was broadened to include global emerging markets. These markets share many of the same characteristics as the Asian markets including retail participation, state ownership and country-specific data and disclosure.
“Asian equity markets are retail dominated and have less sophisticated arbitrage capital than the US and developed markets to iron out inefficiencies and dislocations,” says Malik. For instance, academic research has identified greater overconfidence in China and Taiwan than in the United States and has linked this behavioural bias to sensational, lottery-seeking trading that has often resulted in substantial losses. Nipun can use social media and blogs to track rumour-driven retail stocks and will tend to size positions accordingly. Another important bias is investors’ anchoring to, and extrapolation of, historical sales and profits growth, not taking sufficient account of new developments. Nipun systematically identifies such biases and builds models to use them as alpha drivers.
Unlike developed markets, data is disparate in Asia and hard to collect, standardize and cleanse. Most Asian countries follow IFRS accounting standards, however the exact accounting rules are interpreted and defined differently in each country, making cross country comparisons harder.
It is widely believed that there is not enough data for stocks in emerging markets. Nipun debunks this myth. “There is a high level of data and disclosure in these countries, if you know where to look,” claims Hui. As Head of Research, the task of finding and cleansing the data is mostly for his team. For example, stocks in China are regulated by multiple entities and make different disclosures to a whole range of regulatory bodies. The disclosures are often in local languages and it takes an extensive effort to find and cleanse the data.
China, in particular, is very different from other emerging markets. Recognizing the alpha potential in China, Nipun was an early investor in Chinese equities and first started investing in Chinese ‘A’ shares in 2015 through the Hong Kong Connect program. Based on extensive research, they implemented their investment principles in China in a highly localized manner. This has allowed them to generate alpha from stocks that other investors typically avoid. For example, Nipun finds that State Owned Enterprises (SOEs) are a source of alpha as Nipun’s earnings surprise models work just as well for SOEs as for non-SOEs.
Every researcher must articulate the hypothesis by setting out what is the inefficiency, why does it work, when does it work or not in different markets.
Dr. Ken Hui, Co-Founder, Research and Risk Management Partner, Nipun Capital
Availability of stock borrow on the short side is another source of inefficiency that needs to be carefully navigated in Asia. Nipun uses single stock swaps for shorting across all markets, primarily for alpha, but also because the available equity indices are not a good hedge for their long book in markets such as China. Additionally, and in contrast to most western equity markets, single stock short swaps costing between 0 and 10% annualised are generally cheaper than Chinese equity index shorts, which have sometimes cost as much as 10 or 15% annualized to borrow. The reforms announced in November 2020 should make it easier to short in China, as QFII holders can lend out their inventory and restrictions on charges and term of stock lending are lifted. However, Nipun works with multiple brokers and has not had any trouble accessing borrow for shorts, even after the 2015 crash when the authorities in China did restrict some types of shorting.
Indeed, short sale bans in 2020 – in Korea, Malaysia and Indonesia – are a periodic phenomenon in Asia. However, Nipun has learned to adapt to short sale restrictions and occasional bans over the years. “For the most part in 2020, we can still build our desired portfolio on the short side,” says Hui. For instance, the strategy has adapted to quotas on short interest in Taiwan, partly by trading daily. Short sale bans do not generally require closure of existing positions, though they do apply to the swaps that Nipun use. Nipun has also developed methods to optimize short books amongst its brokers based on relative costs.
Short selling is only one challenge, as the fragmentation of Asian markets and multiple currencies also creates more choices over how to access them on competitive terms and with reasonable transaction costs. Nipun execute long and short trades through swaps with major bulge bracket investment bank brokers, partly to avoid the counterparty risk of local brokers (some of the big brokers also have tie ups with local brokers anyway). Geffen Advisors, run by BGI’s former head of counterparty relationships, David Geffen, was helpful in selecting the brokers back in 2012. “He put in place a systematic framework of quantitative and qualitative processes to interview the prime brokers, provided quantitative data on borrow availability, execution and pricing, and helped to negotiate ISDA terms,” recalls Malik. Nipun find overall trading costs are lower with prime brokers than with executing brokers, because the prime brokers look at Nipun from a holistic wallet angle, including total revenue including security lending and leverage.
For execution, Nipun mainly use broker algorithms modified for time of day and individual market microstructure. Microstructure differences include when and how markets open and close. Occasionally, high touch voice execution is done for opportunistic or local market specific reasons (Taiwan did not have a continuous trading market until this year). Nipun are not using DMA (direct market access) nor colocation, which are often associated with higher frequency strategies, but Nipun have anyway noticed their trading costs declining over the years. “Overall trading costs in Asia have come down over the past decade as liquidity has gone up. Market impact and commissions have trended down. Bid/offer spreads do vary based on liquidity and did widen out in March and April 2020, but the long run trend is down,” says Malik.
Nipun launched in Asia because they determined that Asian markets offered superior alpha generation potential, based on the founders’ extensive experience across nearly all global equity markets. Investors may target positive alpha every year but in practice few deliver it each and every year. Long term performance for the market neutral strategy, Nipun Asia Total Return Fund, has been virtually pure alpha, and uncorrelated to conventional asset classes and other market neutral hedge funds, as measured by the HFRI. Short-term performance has varied, partly with the opportunity set for fundamental investing in general and shorting in particular. The first three years of the strategy, between 2012 and 2014, saw low double-digit annual returns. “The market was pricing in fundamentals, and there was enough volatility to generate alpha on the short side too. There was also some dispersion between market performances within Asia,” recalls Malik. In contrast 2016-2017 saw high single digit annual losses. “In 2017, Asian markets shot up 40% in a straight line. This was a classic dash for trash junk rally with highly levered, high beta, and high volatility stocks doing well. It was an extreme market environment marked by dislocations we had not seen in 20 years. Our proprietary definition of value stocks with strong cashflows and improving sentiment underperformed more expensive stocks with poor cashflows,” says Malik. Most of the losses in 2015-2017 came from the short book; Nipun’s long only strategy did reasonably well. The drawdown in 2016-2017 has led to some changes.
Nipun’s first fund launched in May 2012 with $18 million of mainly partner and friends and family money. The partners have always eschewed seed or acceleration capital deals.
The drawdown prompted changes to risk management, country weightings and accelerated the pace of innovation. “Risk management changed after the 2017 drawdown to allow for more variation in risk targets. Pre-2017 the risk target had been 8-10%, but now overall risk is more aggressively adjusted when volatility goes up and liquidity tightens. The risk process has also cut losses faster,” says Malik. Nipun has shifted exposure away from Japan and towards China, which was a more fruitful source of alpha. But the biggest lesson was to be more nimble and more quickly adapt models to current market conditions, which notably came to fruition in 2020.
Overall Nipun evolved but did not change their DNA because they believed in the merits of their approach. As fundamentals reasserted themselves in 2018-2019, the portfolio rebounded strongly, while traditional or generic value – the Achilles heel of some quant managers who have shut down funds this year – has been muted. “Our own definition of value has outperformed generic value metrics in recent years,” says Malik. Some quantitative managers have argued that the value factor is at its cheapest ever levels and could see a huge recovery but again Nipun is not fixated on traditional value. “We do not expect to generate strong performance from a resurgence of the traditional value factor. We did not see much downside from the underperformance of value and do not expect big upside from any recovery,” explains Malik. Traditional value is not the only generic factor that Nipun has minimal exposure to; the manager also avoids generic momentum.
Nipun is substantially different from other quant managers. The market neutral strategy has no traditional style/risk/factor premia exposures or biases, judging by a regression of its returns versus the excess returns of the MSCI Asia Pacific Ex-Japan Diversified Multi Factor Index, which is a composite of four factors: value, momentum, quality and size. Therefore, the huge divergence between growth and value styles seen in the US, Europe and Asia, is of limited relevance to Nipun’s largely factor neutral approach.
One example of Nipun’s insights is to trade factors that exploit mispricings thrown up by retail investors’ behavioural biases. “For instance, we have created a suite of proprietary factors that we call non-traditional value, in the sense that it is not defined by multiples of price or market capitalization. Our proprietary value factors complement and diversify traditional value factors and have materially outperformed traditional value factors,” says Malik.
As well as avoiding common or generic quantitative factors, Nipun more broadly seeks to sidestep overcrowded positions. “We have developed proprietary tools and measures, and an automated process, to track crowding for every factor or model. This includes tracking volume and momentum for real time measures of crowding. Metrics based on regulatory filings are slower because they are released with a time lag,” Malik explains.
The lessons learned during the drawdown were manifested several years later in a swift response to Covid-19, which led to some revisions that demonstrate how Nipun is more nimble than some other quants. Nipun has some degree of discretion over model data inputs, including in 2020 when Covid-19 rendered historical data irrelevant. “We have been building models for 20 years and understand their strengths and limits. We sometimes need to add data and signals. After the Covid-19 shock and halt to economic activity for many firms and industries, we quickly adapted models by changing inputs to make them better and more relevant. We removed some stale historical data, factored in global linkages between stocks, and changed our interpretation of balance sheet strength and access to cash. Traditionally issuance of debt or equity has predicted underperformance but in this environment the ability to raise capital and bolster balance sheets is a sign of strength and investor confidence. We also observed this phenomenon during the global financial crisis of 2008,” says Malik.
These thoughtful changes have paid off as Nipun saw a double digit return in the third quarter of 2020 as investors refocused on fundamentals and earnings, where Nipun’s models had accurately predicted earnings surprises, and some proprietary valuation measures including wagers against glamour stocks generated profits, particularly in China. The attribution was broad based, across nearly all regions and countries. High stock dispersion bodes well for the strategy, which has no meaningful country, currency or sector exposures. Attribution shows that most of the returns can be explained by idiosyncratic stock risks, and this conclusion is the same whether using Nipun’s proprietary in-house risk system or standard vendor risk packages.
Though Nipun is not trading conventional factors, sensitivity to how factors perform differently in various Asian markets remains an important element of customizing models and signals to local markets. “For instance, the momentum and liquidity factors have historically generated positive information ratios in developed markets but negative returns in Chinese equities. One possible explanation for momentum is retail investors’ short average holding periods of 25-50 days, which might lead to faster price reversals, and may be influenced by leverage,” says Hui. Inefficiencies persist for longer in Asian markets, so a multi-month holding period makes sense, in contrast to a statistical arbitrage approach that might have multi-day or even intraday holding periods.
Governance has always been an important part of models, particularly in China since 2015. We have found that companies with poor governance tend to have weaker cashflows and earnings in future.
Pooja Malik, Co-Founder, Portfolio Manager
Each market in Asia has its own nuances, in terms of factor performances and other features. “China and Taiwan are geographically very close, and both are retail dominated, but they are very different in terms of behaviour, data, disclosure and governance issues,” points out Malik. For instance, momentum can work well in Taiwan.
Moving from countries to sectors, Nipun report standard GIC sector exposures for investors but use finer and more granular definitions for comparative analysis and risk control internally. The standard sector categories are anyway somewhat fluid, for instance China’s Alibaba moved from technology to consumer discretionary last year.
The balance between data types is roughly two thirds fundamental and one third technical, and this has been steady over time. “The methods of extracting fundamental data have changed as we use more smaller vendors who cover more mid cap stocks, and get more granular information from footnotes,” says Malik.
Even where data types are uniform, they need to be parsed in different ways. “Financial data on cashflows and balance sheets is common to all markets, but the way in which we interpret it is differentiated partly due to local accounting standards. For instance, companies in Australia can write up balance sheet values. Other data such as sentiment data is completely localized. Investor flows data is key for gauging sentiment,” explains Malik.
Nipun’s research agenda is county specific but covers a broad variety of data sets. “We have reviewed a lot of alternative data, but its application in the portfolio is limited. Some data sources might only apply to 50 or 100 stocks, which means its risk allocation and impact on portfolio returns is likely to be small. Other data types such as satellite data are becoming less valuable as they become commoditized and more widely used. We are however using NLP to read unstructured data such as media blogs, company reports, financial statement and footnotes,” says Malik. Nipun believes their advantage comes from the insights they generate, not from data alone.
Nipun’s research into areas such as how factors perform differently in Asia has also spawned a long only strategy. For some firms that run both long/short and long only, the long book of the long/short program closely or sometimes even wholly overlaps with the long only. Yet Nipun’s market neutral strategy is very different from its Nipun Emerging Markets Alpha Fund, launched in February 2016. “The research engine is similar across the strategies, but strategy design and portfolio construction are both very different. Average holding periods are 3-4 months on the market neutral and 9-12 months on the long only. The long only strategy has a low beta approach, using defensive stocks to protect downside, whereas the long/short strategies aim to protect downside using their short books. The long only strategy explicitly has a big bias to the “low volatility” factor, because we expect this will structurally outperform. One source of the anomaly is risk-loving investors seeking out lottery-like stocks. The low volatility anomaly may be greater in emerging markets, given that retail investors have larger behavioural biases,” says Malik. This can be further amplified by institutions in emerging markets that are also biased towards concentrated, high volatility portfolios. The leverage aversion issue – leading investors who cannot leverage lower volatility assets to seek riskier stocks – could be more significant in emerging markets where costs of leverage are high. And across all markets, benchmark-conscious investors cannot countenance the tracking error of a low volatility approach. Nipun’s research suggests that a common definition of defensive stocks (MSCI Minimum Volatility Indices) outperformed by nearly three percentage points in emerging markets between 1994 and 2018, but by less than one percentage point in developed markets.
Nipun does not use a common off-the-shelf minimum volatility portfolio primarily due to concerns around crowding. They have developed a proprietary approach to building a low volatility portfolio in emerging markets, one that has performed significantly better than traditional low volatility approaches.
Nipun was an early mover into China and started trading China ‘A’ shares a month after the Hong Kong Stock Exchange Stock Connect launched in 2015. “We like to work collaboratively with counterparties and service providers such as data vendors. We partnered with the prime brokers not only to access stock borrow but also to help them shape and develop their China offerings,” says Malik.
The initial interest in China was to replace Japanese exposure with Chinese exposure in 2015 within the market neutral strategy. This later led to Nipun launching a dedicated China long/short strategy, Nipun China Total Return Fund, in January 2019. Nipun added China just as China ‘A’ shares were becoming investible from 2014, and stock borrow was becoming obtainable in 2015. The Hong Kong Stock Connect was chosen as the most efficient way to access China A shares, though Nipun can also access QFII via its brokers, the supply of which is now uncapped due to the November 2020 reforms. The ChiNext exchange was recently added after it joined HK Connect, taking investable Chinese stocks up to about 1,800. (The long only fund can also trade Chinese ADRs.)
The case for China is partly based on the size and liquidity of the market: China ‘A’ shares are the world’s second largest equity market while Shanghai and Shenzhen together are the most liquid as measured by daily turnover. The asset class has been under-owned by foreign institutional investors, and benchmark-conscious investors need to increase allocations as MSCI Emerging Markets Index has added Chinese large-caps and mid-caps. Over time, weightings should rise further as inclusion factors increase due to higher free floats as restricted shares are unlocked and foreign ownership limits are raised or lifted. Nipun envisages Chinese shares as a whole (including China A shares, Hong Kong and US listed Chinese stocks) could eventually make up over half of the MSCI Emerging Markets Index, and over 10% of the MSCI All Country World Index (ACWI). Nipun has been broadly positive on the valuation of Chinese equities, but more importantly they are a compelling arena for alpha generation due to the aforementioned inefficiencies.
Despite the strong performance resurgence, assets of $440 million in September 2020 remain well below the peak of $825 million reached in 2016. Nipun has stayed focused on constantly honing and refining its research engine throughout the nine years of asset inflows and outflows. Nipun’s approach for testing ideas starts off by insisting that ideas are sensible based on some strong economic rationale or intuition. “Every researcher must articulate the hypothesis by setting out what is the inefficiency, why does it work, when does it work or not in different markets,” says Hui. The predictive power needs to be precisely defined. “We need to ascertain why the idea works. For instance, does it directly predict an earnings surprise, or does it work indirectly by predicting institutional buying,” adds Hui. Ideas also need to be consistent and additive relative to the existing library of insights.
“The testing of ideas is now standardized and almost automated, which enables rapid platform testing of research. The fastest onboarding of a new signal was about two weeks. We have between 70 and 80 insights at any time. This fluctuates as 16 were added in 2019, and some can also be deleted. An insight is a signal,” says Malik. The Covid-19 revisions were a good example of how swiftly Nipun can adapt to changing market climates.
Nipun views statistical techniques as a tool to alpha generation. They do not have a goal of using AI or machine learning processes. Rather, their goal is to find alpha generating insights and they view the latest modeling techniques as an additional tool in their toolkit. “Our experience of researching AI and machine learning processes has been that they result in a momentum bet, based on factors that worked well recently. We do however have some non-linear models such as a probabilistic fraud detection model,” says Malik.
The research agenda is evolving with key areas being: “Building dedicated country specific models, adding local data providers to enhance datasets, web-scraping for data collection, adding corporate governance metrics, and refining sentiment models to track short term shifts in sentiment,” adds Malik.
Corporate governance is one aspect of ESG. “Governance has always been an important part of models, particularly in China since 2015. We have found that companies with poor governance tend to have weaker cashflows and earnings in future. This year the E and S have also been incorporated on the risk side,” says Malik. Nipun are not currently UN PRI signatories, but Malik, “considers our approach to be broadly aligned with UN PRI principles”. Nipun has for some years run exclusionary mandates avoiding industries such as tobacco, alcohol or armaments, for some clients such as endowments and foundations, and ESG is now an increasing source of positive alpha signals.“Nipun is a small organization, but we believe that our actions can have a much larger impact than our size may suggest. The leadership team strives to model behaviour in their individual lives that demonstrates that they take sustainability seriously, and we encourage the broader team to do the same,” says Malik.
Lower transaction costs could lead some types of inefficiencies to erode in future, as could inflows: “We expect that most markets, including the Chinese market will become more efficient over time as more institutional capital flows in, since institutions are less constrained and less susceptible to behavioural biases,” says Malik. For instance, institutional investors, who typically have much longer holding periods than retail investors, are steadily growing their share of the Chinese market. This could be one long term reason for broadening the investment universe, while short sale bans in some Asian markets provides near term motivation. In 2020, Nipun decided to add Eastern Europe, Middle East and Latin America, increasing the investment universe from about 6,000 stocks to over 7,000 stocks. “We see the same inefficiencies as in Asia. For instance, Saudi Arabia has many state-owned enterprises and retail investors,” says Hui. All in all, Nipun is upbeat about the climate for alpha generation in a post-Covid-19 world.
The average span of a hedge fund is five years. Nipun has far outlasted that, despite the volatility since they launched. Further, they have executed a strong turnaround, continued to invest in research, explored new markets and built new strategies. “Our self-belief in the process, the confidence in the team and the goal of building a strong track record are what keep the partners motivated. We have come a long way with our recent wins and believe we are off to great places,” says Malik.