Performance Tests of TruValue Labs Volume, Insight, and ESG Activity Signals

Transcription

1 1 Performance Tests of TruValue Labs Volume, Insight, and ESG Activity Signals Results for All Country World Ex-US Index (ACWX) Stephen Malinak, Ph.D. Chief Data and Analytics Officer TruValue Labs Greg Bala Lead Data Scientist TruValue Labs Oct 2018

2 2 About the Authors Dr. Stephen Malinak Chief Data and Analytics Officer TruValue Labs Stephen leads TruValue Labs quantitative research team in applying artificial intelligence and machine learning techniques to create new financial signals from unstructured data. An industry leader in quantitative analytics, Stephen joined TruValue Labs from Thomson Reuters, where he spearheaded the company s quantitative analytics offering, StarMine, and developed over 15 quantitative models. He has an extensive track record of successful predictive models using widelyvarying techniques across numerous domains. He attended college at the Massachusetts Institute of Technology where he received his S.B. in Electrical Engineering and Computer Science. Malinak went on to receive his Masters and PhD in Engineering-Economic Systems from Stanford University. Greg Bala Lead Data Scientist TruValue Labs Greg is a technology leader and innovator with over 30 years of wide-ranging, cross-domain experience in data analytics, mathematics, program management, intellectual property creation, and software development. Greg attended the California Institute of Technology where he received his Bachelor s degree in Applied Mathematics. Performance Tests of TruValue Labs Volume, Insight, and ESG Activity Signals Visit to learn more. Copyright 2018 TruValue Labs

3 1 Table of Contents 2 Executive Summary 3 Summary of TVL Data Fields 4 ACWX Index Overview 6 Results Using Volume + Insight Strategy 9 Volume Relative to Size and Expectations 11 ESG Activity Signal 14 Strategy Variations and Sensitivity Analysis 19 Factor Exposure and Smart Beta Additivity 21 Conclusions and Plans for Further Research 22 References

4 Performance Tests of TVL Volume, Insight, and ESG Activity Signals Executive Summary This study tests the effectiveness of timely Environmental, Social, and Governance (ESG) signals as screening tools and quantitative alpha factors for the All Country World Ex-US index (ACWX) benchmark over the past ten years. Initial tests of this data on US large cap stocks (S&P 500 and Russell 1000) showed promising results, with outperformance of 3-5% annually vs. the benchmark depending on the strategy employed. This study investigates performance on the ACWX, which is a widely-=followed collection of large-cap stocks outside the US. TruValue Labs produces multiple ESG signals derived from unstructured text. The broad screening strategy based on TVL s Volume and Insight signals, previously demonstrated on the Russell 1000, also shows outperformance of 3.7% annually on the ACWX. Seeing global results very similar to the 3.5% outperformance demonstrated in the US increases confidence in the global reliability of the strategy. Looking outside the US across many countries where multiple languages are spoken raises some issues of comparability of signals across these different markets. Accounting for these differences provides an opportunity to improve performance in two ways: measuring volume relative to expectations for companies of different sizes and comparing companies within each country rather than across countries. These enhancements lead to a new ESG Activity Signal with the top quintile of companies beating the ACWX index by 7.9% annually. The ESG Activity Signal complements the broad screening strategy of Volume + Insight, enabling a quantitative rating of the full benchmark universe and more straightforward combination with Smart Beta factors. The ESG Activity Signal shows low correlation to typical smart beta factors, enabling top quintile performance of 9.0 percent and a quintile spread of 9.8 percent in combination with typical smart beta factors. 2

5 Summary of TVL Data Fields The previous US whitepaper [1] provides a basic introduction to the TruValue Labs historical data set, with more detailed descriptions of the ESG categories and data fields in product documentation [2] as well as detail on material categories by industry on the Sustainability Accounting Standards Board (SASB) website [3]. Figure 1 summarizes the signals available for backtesting as time series: Trailing 12 Months (TTM) Volume counts the articles for that company in the past 12 months; Pulse captures short-term day-to-day variations; Insight provides a longer-term rating equivalent ; and Momentum looks at changes in Insight over the past 12 months. Figure 1: Pulse, Insight, Momentum, and TTM Volume Signals These historical data fields are available for testing directly from TruValue Labs via SFTP as well as via select third-party data aggregators such as Open Factset. Scores are available for all 30 SASB categories, as well as two aggregates: all SASB categories together and just the material SASB categories according to the primary industry for a company. Based on findings of previous tests, this whitepaper focuses on results for all SASB categories as this approach provides the most information content. 3

6 ACWX Overview The All Country World Ex-US Index is tracked by the ishares ACWX ETF that discloses monthly constituents going back to early 2008 [4]. Figure 2 shows how the number of constituents in ACWX has doubled from around 600 in 2008 to over 1,200 in The studies in this whitepaper track these constituents as they were at each point in time, with monthly updates over the past ten years. Figure 2: Count of ACWX constituents over 10-year test period ACWX covers major developed markets outside the US, and it also includes several emerging markets as well. Figure 3 shows constituents by country as of May Japan accounts for the largest number of constituents. Notably the next two largest countries are emerging markets: China and South Korea. The United Kingdom, Canada, France, Germany, and Australia round out the top 8. Approximately 24% of the constituents come from countries where English is a primary language these countries are shown in blue in Figure 3. Some ex-us companies are represented as US-listed ADRs. Figure 3: ACWX holdings by Country as of May 2018 (English Speaking and Non-English Speaking) 4

7 The historical dataset for TVL signals currently consists entirely of scores from English-language documents. Not surprisingly, English-speaking countries feature higher levels of coverage by document count, as shown in Figure 4. 64% of companies in English-speaking countries have high or medium-level coverage, whereas 46% of non-english-speaking countries have high or medium coverage 1. Nevertheless, as will be shown later, it is still possible to develop an interesting and highly profitable signal in countries where English is not the primary language. Figure 4: TVL coverage for English-Speaking and Non-English-Speaking Countries as of May 2018 The backtests in this whitepaper were performed using Factset s Alpha Tester 4 platform, with the assistance of local Factset consultants to load the TVL scores as proprietary data fields for testing 2. 1 TVL defines high coverage as 26+ articles per trailing 12 months (TTM), medium as 5-25 articles, and low coverage as 1-4 articles per TTM. 2 Unlike the Russell 1000, it was not possible to test the ACWX on the Quantopian platform as Quantopian does not currently offer stock prices and total returns for securities outside the US. Seeing opportunities for synergies, Quantopian and Factset have announced plans for a joint offering to take advantage of different features available on each platform [5]. 5

8 Results Using Volume + Insight Strategy Studies of TVL data volume on the Russell 1000 showed interesting results that worked against the direction of the traditional size effect. Companies with high TVL data volume (the top half of companies), which tended to be larger companies, outperformed companies with low TVL data volume (the bottom half) by about 2% annually. Meanwhile, the traditional size effect worked as expected, with the smallest half of companies outperforming the largest half of companies by about 2% annually. Putting these two together showed additive benefits wherein companies in the top half of TVL data volume relative to size showed even more outperformance (4% annually) vs. companies in the bottom half of TVL data volume relative to size [1, Figure 20]. Over the same 10-year test period, the ACWX shows an even stronger size effect, with the smallest quintile of size outperforming the largest quintile of size by over 15.4% per year 3, as shown in Figure 5. Thus, the TVL data volume effect is fighting very strong headwinds to the extent that it tends to favor larger companies. Not surprisingly given those headwinds, the highest quintile of TVL data volume underperforms the lowest quintile, but only by 2.1% annually. Since companies with higher TVL data volume also tend to be larger companies, Figure 5 also compares the quintiles of TVL data volume to the quintiles of size as measured by market cap, revealing a positive quintile spread of 13% for volume relative to expectations based on size. Thus, higher data volume still leads to higher relative performance, after accounting for the underlying size effect. Later sections of this paper dig much more deeply into ways to measure TVL data volume relative to various proxies for company size. Figure 5: Return by Volume Quintile vs. Return by Market Cap Quintile 3 Note, the quintile returns in this chart and others that follow are active returns relative to the underlying cap-weighted ACWX benchmark. Active returns are measured as the annualized difference between the equal-weighted test portfolios, quintiles in this case, and the cap-weighted benchmark. 6

9 The Russell 1000 study [1, Figure 6] proposed a TVL Volume + Insight Strategy based on first filtering a benchmark for the top 50% by volume, then taking the top 5/6 by TVL Insight score, and equal-weighting the remaining portfolio. Tests of this TVL Volume + Insight strategy showed outperformance of 3.5% annually on the Russell Figure 6 shows performance of six equal-count fractiles 4 of the TVL Insight score for stocks in the top half of volume on the ACWX. Combining the top 5/6 fractiles to create the Volume + Insight strategy leads to 3.7% outperformance per year relative to the underlying ACWX cap-weighted benchmark. Figure 6 shows that the top 1/6th of Insight scores, though a much narrower portfolio, outperform the ACWX benchmark by 4.5%, while combining the top 2/6th (= 1/3rd) has outperformance of 4.3%. Figure 6: Return by Six Fractiles of Insight Score for Top Half of Volume Figure 7 produces the time series equity curve for the Volume + Insight strategy and compares it to the ACWX benchmark over the 10-year test period. Performance during the global market crash of 2008 is just slightly better than the benchmark, but full recovery from the initial drawdown occurs 9 months faster. Outperformance is reasonably steady afterwards, with more than 2.5x the total return over the 10-year period (102% total return vs 40% total return). 4 The motivation for looking at 6 fractiles was based on the Insight 360 user interface for inspecting TVL data in detail. One popular portfolio view chart looks at a 3x3 graph of Insight and Momentum scores. 6 fractiles provides a more detailed look at the Insight scores that fits within this charting framework. The quant studies that follow in this whitepaper fall back to the more common quant approach of using quintiles (5 groups of companies). 7

10 Figure 7: Equity curve for Insight + Volume Strategy compared to ACWX Benchmark The ESG Momentum strategy from the R1K whitepaper noted certain sweet spots of Insight combined with Momentum in the US. Outside the US there are certain sweet spots as well, but the size of the portfolios combined with variations in situations across different countries essentially renders those portfolios too small to test reliably. An overriding observation across the global data set is that ESG data volume is of more influence in finding alpha than ESG Momentum as defined in the TVL platform. ESG Momentum may play more of a role when data volume has ramped up to consistently high levels across the global platform. Given the current variability of data volume across different companies, the rest of this paper explores signals resulting from the combination of ESG data volume and the Insight score. 8

11 Volume Relative to Size and Expectations Figure 5 showed a substantial market cap effect on the ACWX which explains why volume works backwards compared to R1K, given these large headwinds. Adjusting ESG data volume relative to market cap would directly introduce one of the most common size factors to the portfolio, prompting an inquiry into what proxy for size might be more comparable to expectations of news volume and less correlated with market cap itself. The results for the Volume + Insight strategy were produced using global rankings of Volume and Insight scores. In the Russell 1000, all companies are US-based so the issue of country-relative scores does not arise. Digging into more detail on what drives volume, a first step is to look at country-relative scores. This enables more meaningful comparisons across country boundaries, as many of these countries have far different states of economic development and wide differences in the availability of English-language documents. Academic studies have shown a positive relationship between news coverage, trading volume, and market liquidity on a given stock [6]. Hence, trading volume can be used to capture the same basic size effect, with a closer expected relationship to news volume and less exposure to market cap and its side effects of price momentum and valuation. Trading volume can also be aggregated over the trailing twelve months (TTM) for direct comparison to TTM ESG data volume. Figure 8 shows that TTM dollar trading volume shows a bit less price sensitivity as a measure of size compared to market cap, on a quintile basis. Both measures of size have been ranked with Q1 being the smallest and Q5 being the largest, thereby creating a positive quintile spread for the size effect. Figure 8: Quintiles of TTM Trading Volume on ACWX vs. Quintiles of Market Cap Figure 9 shows that ESG data volume relative to dollar trading volume also shows a very similar price response as ESG data volume relative to market cap. But importantly, Figure 10 shows that ESG data volume relative to dollar trading volume has only a 0.50 correlation to market cap, whereas ESG data volume relative to market cap has a 0.89 correlation to market cap. 9

12 Figure 9: ESG Data Volume / TTM $ Trading Volume vs. ESG Data Volume / Market Cap on ACWX Figure 10: ESG Volume / TTM $ Trading Volume has less correlation to Market Cap on ACWX 10

13 Performance Tests of Insight, ESG Momentum, and Volume Signals ESG Activity Signal Looking at ESG data volume relative to expectations (using TTM dollar trading volume as a proxy for those expectations) enables ranking of the full stock benchmark universe rather than just those stocks with high ESG data volume. Even companies with low or no TTM ESG data volume can be compared to expectations of data volume based on TTM dollar trading volume. The ESG Volume + Insight strategy showed that both data volume and polarity of the Insight signal (positive or negative sentiment) have an impact on price returns. However, that strategy was based on a double screen (one screen for each of the two factors) and thereby did not rank the entire benchmark universe. That strategy also provided minimal guidance for likely relative price returns for companies that passed the double screen. Experimentation and observation have shown that the ESG Data Volume and Insight Score strongly complement each other on the positive side, with companies having both strong data volume relative to expectations and high Insight scores demonstrating outperformance. On the short side, however, many of the companies with strong data volume relative to expectations and highly negative Insight scores tend to crash quickly in an ESG crisis event, overreact from a pricing basis, and then recover slowly afterward. Meanwhile, companies with slowly leaking bad news are less likely to crash and overreact, and more likely to drift downward slowly. Studies of traditional short-term news analytics have also found evidence of overreaction to bad news on high levels of social media buzz, accompanied by underreaction to positive news [7]. Some behavioral finance investors capitalize on these types of media-driven behavioral market over-reactions and under-reactions [8]. Putting these observations together suggests that multiplying Insight rank by rank of ESG data volume relative to expectations could be an interesting way of combining these variables, rather than simply adding them or using them as double screens. Figure 11 shows quintiles for the multiplication of two rank variables. Using this approach, the top quintile mainly consists of cases with both high Insight rank and high (TTM ESG Data Volume / TTM Dollar Trading Volume) rank. However, it also includes low activity companies with the very highest Insight ranks, as well as extremely high activity companies with the lowest Insight ranks (these are the companies most likely to have a price crash overreaction to an ESG crisis). Note that all five quintiles in Figure 11 have the same number of colored dots, even if the numbers might appear to be different. Note further that the zone breakpoints were created using (100-rank) for each variable to create the curvature in this direction. This multiplicative approach works well for combining these two variables as a predictive stock ranking combination. (TTM ESG Data Volume / TTM $ Trading Volume) Rank has a strong linear price response but does not account for the polarity of the ESG news. Figure 11: Quintiles for Multiplication of Two Ranked Factors 11

14 Insight Rank shows outperformance for the highest Insight companies but also shows a bounce for the lowest ranking companies. Combining the two, as shown in Figure 12, produces an excellent combination ranking factor while incorporating the polarity of the ESG news on the positive side with the tendency of the most strongly reported negative ESG events to experience a price rebound in the longer term. For simplicity, going forward this paper will use the name ESG Activity Signal to refer to the product of the Insight Score Rank with the rank of the ratio of TTM ESG Data Volume relative to TTM dollar trading volume. All factor ranks are calculated on a country-neutral basis, and all quintile returns are active returns relative to the ACWX capweighted benchmark. Figure 12: Active Returns for Insight, Activity, and Insight x Activity The top quintile of the ESG Activity Signal shows strong returns over the backtest period vs. the cap-weighted ACWX benchmark. Figure 13 shows similar returns to the benchmark during the Great Financial Crisis of 2008, followed by steady and strong outperformance after that. The Total Period Return of 187% is 4.7x the Total Benchmark Return of 40% over the 10-year test period with monthly rebalancing. Figure 13: Returns over time vs. benchmark 12

15 Figure 14 shows the steady ramp-up in counts of stocks over time for this top quintile, in proportion to expansion in the count of benchmark constituents along with increased data availability. Figure 14: Counts of stocks over time 13

16 Performance Tests of Insight, ESG Momentum, and Volume Signals Strategy Variations and Sensitivity Analysis Traditional ESG factors have been used more for screening out classic ESG offenders such as tobacco, coal, or munitions. When tilting toward positive ESG scores, most traditional indices still use market cap weighting to minimize tracking error. Hence, long-running indices and ETFs based on traditional ESG factors do not show much outperformance. Figure 15 shows a sample index based on traditional ESG data using a closely related benchmark, ACWI (which is the global version of ACWX, including both ex-us and US stocks). Note that the lifetime (and 10-year) outperformance of the ESG version of the benchmark is about half a percent annualized, but with underperformance over the past 5 years [9]. Figure 15: Comparison of MSCI ACWI ESG Leaders Index to underlying ACWI Benchmark The TVL ESG Activity Signal, in contrast, provides strong outperformance across the 10-year test period. Interestingly, the signal performs well in both English-speaking and non-english-speaking countries that have stocks included in the ACWX benchmark, as shown in Figure 16. The top quintile for English-speaking countries outperforms the ACWX benchmark by 9.8% annually, with a quintile spread of 6.5% annually. In non-englishspeaking countries, the top quintile outperforms the ACWX benchmark by 6.4% annually, with a quintile spread of 5.3% annually. 14

17 Figure 16: Results on English-Speaking vs non-english-speaking Figure 17 extends this geographic analysis. In the ACWX, developed countries have a top quintile of 8.3% and a quintile spread of 6.7%. Emerging markets have a top quintile of 7.3% and a quintile spread of 2.8%. Figure 17 then shows detailed performance for the largest 20 countries in the ACWX (by stock count), sorted from the most positive quintile spread to the most negative quintile spread. Germany shows the strongest performance, followed by Hong Kong, Australia, and Switzerland. The TVL ESG Activity Signal struggles the most in Singapore, China, Taiwan, and Indonesia. Adding local languages in a future release of the signal will likely help with performance in these Asian countries. Figure 17: Results by Country Type and by Country for Top 20 countries 15

18 Many investors follow a sector-neutral strategy and want to understand how well a factor ranks companies within the same sector vs. picking among different sectors in different proportions to the underlying benchmark. Figure 18 shows a comparison of three approaches to neutralizing the exposure of the ESG Activity Signal. The middle chart, Country Ranks, is the approach followed above. The left chart, Global Ranks, allows the algorithm full discretion to favor any stock in any country in any sector, based on its global rank. The right chart, Sector Ranks, looks at the ability to sort stocks within the same sector, though those sectors can cross any number of countries in the ACWX benchmark. Global ranks enable slightly greater outperformance on the top quintile relative to the ACWX benchmark, 8.1% annually for global ranks vs. 7.9% annually for country ranks, but both approaches have the same quintile spread of 5.8%. However, country ranks do a better job of sorting across the spectrum of scores, with a more linear price response than the global ranks. This could be because of widely varying market structure and Englishlanguage document availability across the full ACWX benchmark. Sector ranks provide a top quintile of 6.3% and a quintile spread of 3.9%. Comparing to the performance of country ranks, restricting to a sector-neutral approach delivers 80% of the performance in the top quintile and 67% of the quintile spread. Hence the predictive power of the ESG Activity Signal is about 20-33% from its ability to pick sectors and 67-80% from its ability to pick companies within sectors. Figure 18: Global Ranks vs. Country Neutral vs. Sector Neutral for the TVL ESG Activity Signal As is the case with any quantitative factor, the ESG Activity Signal performs better within some sectors than others. Figure 19 shows performance results across the Factset Sectors for this test period for the ACWX. Note that these sectors do not perfectly align with SASB s sector system used to determine materiality, so that may raise certain issues that could be addressed on a company-by-company basis. Sector spreads look best in consumer durables and utilities, and worst in health services, technology services, and commercial services. Interestingly, all sectors show positive returns from the top quintile, other than industrial services, which is just barely below zero. For the three sectors with significantly negative spreads, in each case the negative spread is caused by outperformance of the lowest quintile rather than underperformance of the highest quintile. 16

19 Some of the variation is due to smaller samples and normal market variability. Some of the variation may be due to stronger signals within certain sectors. This question is best addressed by fundamental analysts with strong sector expertise, and it provides an opportunity to enhance results via a quantamental approach. Figure 19: Performance Results by Sector After comparing ESG data volume to trading volume, does the absolute level of ESG data volume have as important of an impact? Backtests of the ACWX universe suggest that ESG data volume relative to expectations matters more than the absolute level of data volume, as illustrated in Figure 20. Splitting the companies in each country into two groups by data volume shows that both the upper and lower halves of ESG data volume show similar quintile spreads within each. Interestingly, companies with high data volume are more likely to show a bounce or upward price reversion in the lowest quintile. The top quintile of high data volume outperforms the top quintile of low data volume, and the bottom quintile of low data volume underperforms the bottom quintile of high data volume; this observation supports the behavioral theory of greater overreaction to negative news on higher data volume. High ESG data volume companies show a top quintile of 8.3% and a quintile spread of 4.9%. Low ESG data volume companies show a top quintile of 7.1% and a quintile spread of 5.3%. Figure 20: TVL ESG Activity Signal for High Data Volume vs. Low Data Volume Companies 17

20 As was found on the analysis of the Russell 1000, TVL ESG factors have a negative correlation to classic quant value factors and do a better job of sorting growth (non-value) stocks than value stocks. Analysis of the ACWX leads to the same conclusion, as summarized in Figure 21. Growth stocks (companies in the bottom half of E/P) have a top quintile of 8.0% and a quintile spread of 6.7% when sorted by the TVL ESG Activity Signal. Value stocks (companies in the top half of E/P) have a top quintile of 6.9% and a quintile spread of 4.6%. Figure 21: TVL ESG Activity Signal for growth stocks vs. value stocks 18

21 Factor Exposure and Smart Beta Additivity Typical quant factors are often employed across a broad index such as ACWX as a smart beta approach. Using multiple uncorrelated factors generally leads to higher performance. Building such a combination smart beta approach provides a good basis to see how additive new factors are to traditional well-known and commonly used factors. Five commonly used quant factors are Size (market cap), Price Momentum (12M minus 1M), Minimum Volatility (trailing 3yr stock price volatility), Value (E/P), and Quality (ROE). Figure 22 shows how these common factors fare alone and in combination on the ACWX over this 10-year test period. All perform reasonably well on their own, other than the worst value and quality quintiles tending to spike upward rather than move downward over this test set. The five-factor combination yields an impressive top quintile of 8.4% and a quintile spread of 9.1%. Figure 22: Returns of 5 Smart Beta Factors and Combination Figure 23 shows that the TVL ESG Activity Signal has a negative correlation to four of these five quant factors, with only size (market cap) having a positive correlation of These low correlations to smart beta factors 19

22 suggest that the TVL ESG Activity Signal will be additive in combination to those factors, as is indeed shown in Figure 24. Figure 23: Correlation of ESG Activity with Smart Beta Factors Figure 24 starts with a 50/50 blend of the 5-factor smart beta combination and the TVL ESG Activity Signal, which is then re-ranked into combination quintiles within each country. The top quintile of the combination adds 0.5% to the smart beta signal annually, while subtracting 0.2% from the bottom quintile. Thus, the quintile spread increases by 0.7% annually. Note that this analysis makes no attempt to optimize the signal combination; for instance, performance results could clearly be increased by shifting relative weights for different classes of stocks, such as value vs. growth. Figure 24: Additivity of ESG Activity to Smart Beta Combination 20

23 Conclusions and Plans for Further Research This whitepaper has shown that TVL s ESG signals derived from English-language text documents show strong potential for alpha generation outside the US. Replicating the ESG Data Volume + Insight strategy developed on the Russell 1000 leads to 3.7% annual outperformance vs. the ACWX benchmark, similar to the 3.5% outperformance vs the Russell Combining the TVL Insight signal with ESG data volume relative to expectations based on dollar trading volume enables ranking of the full ACWX benchmark universe with the ESG Activity Signal, as well as delivering 7.9% annual outperformance for the top quintile vs. the ACWX. This ESG Activity Signal shows strong performance in both English-speaking and non-english-speaking countries, and the combination of these signals increases both the top quintile and quintile-spread performance of Smart Beta strategies. Similar tests of the TVL ESG Activity Signal also show similarly promising results on the Russell 1000, with an even lower correlation to size and smart beta. These results, as well as broader global results will be featured in an upcoming whitepaper. Other trading strategies to be investigated include shorter-term event-driven trading using the TVL Pulse Score and upcoming significant events database, which is currently under development. 21

24 References [1] Stephen Malinak, Ph.D., Junyi Du, and Greg Bala; Performance Tests of Insight, ESG Momentum, and Volume Signals: U.S. Large Cap Results for S&P 500 and Russell 1000 Benchmarks, , May 2018 [2] ESG Scoring Methodology, TruValue Labs, updated February [3] SASB Sustainability Framework: [4] [5] factsetandquantopianannounceplanstolaunchfinancialdataanalysisplatformtohelpinvestmentindustrycapitalize onrapiddatagrowth [6] [7] 15Jun2016.ElijahDePalma.pdf [8] [9] 22

25 About TruValue Labs is the first company to apply artificial intelligence to uncover timely Environmental, Social, and Governance (ESG) data on a variety of asset classes. The company s mission is to deliver increased transparency to investment professionals by providing data and analytics that go beyond traditional fundamentals. The flagship product, Insight360TM, delivers timely and investable insights by revealing intangible value and risk factors from unstructured data. Visit to learn more about the SaaS product and data service. Phone support@truvaluelabs.com (888)

26 Copyright 2018 TruValue Labs 24