According to our data, 70 per cent of the UK&#8217;s blue-chip firms were not meeting expectations on climate related disclosures. You can read our ESG Research Director, Diana Rose&#8217;s thoughts on this in the latest This is Money article below.
&nbsp;
<a href="https://www.thisismoney.co.uk/money/markets/article-11089707/Grocers-Tesco-Ocado-falling-short-green-promises.html">https://www.thisismoney.co.uk/money/markets/article-11089707/Grocers-Tesco-Ocado-falling-short-green-promises.html</a>

household-names-from-tesco-to-ocado-failing-to-give-enough-information-about-how-they-will-survive-in-a-greener-economy-

Household names from Tesco to Ocado failing to give enough information about how they will survive in a 'greener' economy

<a href="https://ifamagazine.com/article/lcaw2022-balancing-the-net-zero-ambition-with-the-greenwashing-tightrope-diana-rose/">https://ifamagazine.com/article/lcaw2022-balancing-the-net-zero-ambition-with-the-greenwashing-tightrope-diana-rose/</a>
Our ESG Research Director Diana Rose was quoted in IFA Magazine regarding her thoughts on the 4th annual London Climate Action Week and the increasing role of finance in achieving our net zero ambitions in the UK.
You can read the article in full using the link above.

balancing-the-net-zero-ambition-with-the-greenwashing-tightrope

Balancing the net-zero ambition with the greenwashing tightrope

ESG Research Director Diana Rose recently spoke with Elric Langton of Small Company Champion.
In his editorial for ShareScope, Elric looked at how ESG will impact lives, investments, and Corporates’ role in the implementation and consequences of attempts to bypass ESG. He also looked at how Insig AI is helping fund managers navigate and build wealth through the use of AI.
You can read Elric’s insights and Diana’s comments here:
<a href="https://knowledge.sharescope.co.uk/2022/06/29/how-important-is-esg-to-the-global-financial-institutions-and-global-business-and-are-we-ready/">https://knowledge.sharescope.co.uk/2022/06/29/how-important-is-esg-to-the-global-financial-institutions-and-global-business-and-are-we-ready/</a>

How-important-is-ESG-to-the-global-financial-institutions-and-global-business-and-are-we-ready

How important is ESG to the global financial institutions, and global business and are we ready?

tesla-kicked-out-of-s&p-ESG-index

Tesla kicked out of S&P’s ESG Index: What just happened

This case may be the first of its kind, but we all know it won&#8217;t be the last.  The SEC&#8217;s greenwashing taskforce is on a mission and this will be a wake up call to fund managers.  This particular case is around the claim that all stocks went through an ESG quality review as part of its investment process, which they apparently didn&#8217;t.
Overstating the ESG qualities of a fund to attract capital (ie greenwashing) is a very real problem which risks damaging not only the credibility of ESG, but its potential to actually get capital flowing in ways that will benefit society and help make the planet liveable for our children.
However, I must admit I don&#8217;t envy the regulators trying to police ESG space.  It often feels like it&#8217;s going through the tricky teenage years and right now they&#8217;re having a pretty wild party.  What is an &#8216;ESG quality review&#8217; in the first place?  Every fund will take a different approach to this process, but it&#8217;s guaranteed there will be subjectivity and maybe even smoke and mirrors.  What can bring integrity back to ESG funds, (aside from telling the truth about whether or not these quality reviews are being done at all) is oversight and transparency on the process; consistency and due diligence in the method.
Regulation is coming down the pipeline soon to crack down on the freeloaders at the party.
At Insig AI, we speak to a lot of asset managers anticipating regulation and striving for best practice.  They talk about the time pressures to turn around ESG analysis, and costs and expertise overheads of doing the research they want to on stocks as limiting factors.  This is why we&#8217;ve designed AI tools to take the pain out of doing due diligence on your portfolio&#8217;s public disclosures on ESG with transparency and consistency &#8211; to bring integrity to the ESG review process so that our clients aren&#8217;t looking over their shoulder for the greenwashing police.

sec-issues-first-greenwashing-fine-to-an-investment-adviser

SEC issues first 'greenwashing' fine to an investment adviser

Insig AI&#8217;s ESG Research Director <a href="https://www.linkedin.com/in/ACoAAAEN8ukB24Qgb2FV7o0onAJyWf9meMTRnpw" data-attribute-index="0" data-entity-hovercard-id="urn:li:fs_miniProfile:ACoAAAEN8ukB24Qgb2FV7o0onAJyWf9meMTRnpw" data-entity-type="MINI_PROFILE">Diana Rose</a> was quoted in the recent Financial Times Sustainable Views article regarding the need for increased transparency on ESG ratings.
You can read the full article here:
<a href="https://www.sustainableviews.com/sustainable-companies-are-more-expensive-says-study/">https://www.sustainableviews.com/sustainable-companies-are-more-expensive-says-study/</a>

sustainable-companies-are-more-expensive-says-study

Sustainable companies are more expensive, says study

How can you tell if a company is not only talking the talk but also walking the walk when it comes to ESG? in this article, our ESG Research Director Diana Rose sets out how Artificial Intelligence can be the key in figuring who isn&#8217;t keeping their promises.
You can read the full Computer Weekly article here:
<a href="https://www.computerweekly.com/blog/Green-Tech/Using-AI-to-discern-if-enterprises-are-walking-the-talk-on-ESG">https://www.computerweekly.com/blog/Green-Tech/Using-AI-to-discern-if-enterprises-are-walking-the-talk-on-ESG</a>
&nbsp;

using-ai-to-discern-if-enterprises-are-walking-the-talk-on-esg

Using AI to discern if enterprises are walking the talk on ESG

Between 2015 and 2020, the amount of ESG disclosures quadrupled, but we are by no means at a peak. 
You can read our ESG Research Director Diana Rose&#8217;s thoughts on what the future of ESG disclosure might look like in the latest Pensions Expert blog using the link below.
You can read the article in full here:
<a href="https://www.pensions-expert.com/Law-Regulation/What-does-the-future-of-ESG-disclosure-look-like?ct=true">https://www.pensions-expert.com/Law-Regulation/What-does-the-future-of-ESG-disclosure-look-like?ct=true</a>
&nbsp;

what-does-esg-disclosure-look-like

What does the future of ESG disclosure look like?

The &#8216;G&#8217; in ESG is widely known as governance, but what exactly is governance? And how can it be used to ensure better long-term performance in a company?
You can listen to our ESG Research Director <a title="https://www.linkedin.com/in/ACoAAAEN8ukB24Qgb2FV7o0onAJyWf9meMTRnpw" href="https://www.linkedin.com/in/ACoAAAEN8ukB24Qgb2FV7o0onAJyWf9meMTRnpw">Diana Rose</a> discuss governance on the latest <a title="https://www.linkedin.com/company/financial-times/" href="https://www.linkedin.com/company/financial-times/">Financial Times</a> Adviser Podcast using the link below:
<a href="https://podcasts.apple.com/gb/podcast/exploring-governance-the-g-in-esg/id1437151302?i=1000556289364">https://podcasts.apple.com/gb/podcast/exploring-governance-the-g-in-esg/id1437151302?i=1000556289364</a>

what-exactly-is-governance

What exactly is governance?

<h2>TCFD: Task Force on Climate-related Financial Disclosure</h2>
One could argue that this four-letter acronym is the start of the green revolution for the corporate reporting world. With TCFD reporting now mandatory from 6 April 2022 for over 1,300 large corporations with &gt;500 employees and &gt;£500m in annual turnover, the UK is the first of the G20 to make it law, with others set to follow.
One could also argue that the regulatory burden of disclosure and reporting is going to cause a huge amount of pain. Not just for the companies required to report, but every investor required to analyse the mountain of disclosure documents coming their way, in order to substantiate their investment decisions.
The TCFD was established in 2015 and finalised in 2017 to drive the reporting of climate-related financial information in publicly available Annual Reports. Widely applicable across all sectors and jurisdictions, disclosures are around 4 core pillars that demand oversight of climate risk at every level of business: governance, strategy, risk management, and metrics and targets.
There is no question that the merits of TCFD are going to change the way companies operate and report, and will certainly change the way investors evaluate the companies that they buy into. The near-term pain and overhead of the reporting requirement will also hopefully accelerate corporate awareness on their impact on climate and, over time, we all start to see and feel the benefits of a recovering environment.
With the 6th of April looming, the big question is: how many of the major FTSE companies are prepared for the disclosure task ahead of them or risk falling foul of regulation? What’s telling is to see how long a company has been talking about TCFD, and where they publish their disclosures. The point here is that TCFD requires significant lead-in time to implement and requires that climate risk is addressed in the fundamentals of the Annual Report.
<h2>What Does the Data Say?</h2>
Using the Insig.AI database of corporate ESG and financial documents, the goal was to identify trends around TCFD disclosure over time and across report types.
The approach was to use a keyword and phrase search across the entire database on: “TCFD”, “climate related”* and “climate risk”.
COMPANY, TIMEFRAME AND DOCUMENT UNIVERSE 
Companies: FTSE 100 (plus a few other UK names) 
Timeframe: 2015-2021 
No. of documents queried: 3,345 
Document Types: Annual Report, Integrated Report, Sustainability/ESG Report, Climate Change Report, Climate Change Policy, Financial Document, ESG Data
*Searching “climate related” also accurately surfaces “climate-related” which is the TCFD naming convention.
<h2>Summary Observations</h2>
<h3>Who was first to start referencing TCFD?</h3>
Aviva, Barclays, BHP and LSEG were the first FTSE companies to make reference to TCFD back in 2016, before the TCFD recommendations were even finalised.  This indicates foresight and these companies have continued to be leaders in TCFD disclosure over the past 5 years.
<img class="wp-image-2540 size-full aligncenter" src="https://insigai.io/wp-content/uploads/2022/03/esg_tcfd_1.png" alt="" width="904" height="346" srcset="https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_1.png?resize=200%2C77&amp;ssl=1 200w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_1.png?resize=300%2C115&amp;ssl=1 300w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_1.png?resize=400%2C153&amp;ssl=1 400w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_1.png?resize=600%2C230&amp;ssl=1 600w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_1.png?resize=768%2C294&amp;ssl=1 768w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_1.png?resize=800%2C306&amp;ssl=1 800w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_1.png?fit=904%2C346&amp;ssl=1 904w" sizes="(max-width: 904px) 100vw, 904px" />
<h3>References Over Time</h3>
FTSE companies have been actively referencing “climate related” since 2017 and the trend has continued to grow year on year. Reference to “climate risk” only started to gain traction in 2020. TCFD references have doubled year on year since 2017.
&nbsp;
<img class="wp-image-2541 size-full aligncenter" src="https://insigai.io/wp-content/uploads/2022/03/esg_tcfd_2.png" alt="" width="932" height="592" srcset="https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_2.png?resize=200%2C127&amp;ssl=1 200w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_2.png?resize=300%2C191&amp;ssl=1 300w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_2.png?resize=320%2C202&amp;ssl=1 320w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_2.png?resize=400%2C254&amp;ssl=1 400w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_2.png?resize=600%2C381&amp;ssl=1 600w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_2.png?resize=768%2C488&amp;ssl=1 768w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_2.png?resize=800%2C508&amp;ssl=1 800w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_2.png?fit=932%2C592&amp;ssl=1 932w" sizes="(max-width: 932px) 100vw, 932px" />
<h3>Leaders and Laggards</h3>
Analysing combined TCFD relevant disclosures over this same period, there are clear and consistent leaders. At the other end of the spectrum, there is a stark difference in level of disclosure.  It raises some questions about the level of commitment and readiness for TCFD reporting come April. Whilst the ‘materiality’ of climate change risk and impact does vary between sectors, which means some industries would start with more established reporting, a principle of the TCFD is that climate change is a risk to all businesses.
 <img class="size-full wp-image-2542 aligncenter" src="https://insigai.io/wp-content/uploads/2022/03/esg_tcfd_3.png" alt="" width="928" height="574" srcset="https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_3.png?resize=200%2C124&amp;ssl=1 200w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_3.png?resize=300%2C186&amp;ssl=1 300w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_3.png?resize=400%2C247&amp;ssl=1 400w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_3.png?resize=600%2C371&amp;ssl=1 600w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_3.png?resize=768%2C475&amp;ssl=1 768w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_3.png?resize=800%2C495&amp;ssl=1 800w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_3.png?fit=928%2C574&amp;ssl=1 928w" sizes="(max-width: 928px) 100vw, 928px" />
<h3>Disclosure by Document Type</h3>
Taking 2020 as a snapshot of where TCFD disclosures are found across reports, Annual and Climate Change Reports make up just over a third each.  22% are found in Sustainability and ESG Reports.  Companies not yet reporting to the TCFD in their Annual Reports will be required to start in the 2022-23 reporting cycle.
<img class="aligncenter wp-image-2543" src="https://insigai.io/wp-content/uploads/2022/03/esg_tcfd_4-800x720.png" alt="" width="600" height="540" srcset="https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_4.png?resize=200%2C180&amp;ssl=1 200w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_4.png?resize=300%2C270&amp;ssl=1 300w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_4.png?resize=400%2C360&amp;ssl=1 400w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_4.png?resize=600%2C540&amp;ssl=1 600w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_4.png?resize=768%2C691&amp;ssl=1 768w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_4.png?resize=800%2C720&amp;ssl=1 800w, https://i0.wp.com/insigai.io/wp-content/uploads/2022/03/esg_tcfd_4.png?fit=1011%2C910&amp;ssl=1 1011w" sizes="(max-width: 600px) 100vw, 600px" />
&nbsp;
&nbsp;

tcfd-which-ftse-companies-are-ready

TCFD: Which FTSE companies are ready for the 6th of April…and which are not?

what-is-so-hard-about-pdf-text-extraction

What is so hard about PDF text extraction

If you ask any data scientist what they spend most of their time doing, they won’t tell you that it is solving the world&#8217;s hardest problems, or fine tuning an incredible new life changing algorithm, or even predicting the next bitcoin bounce. No- what they would say is, collecting and cleaning data. Period. Once they have done this, they still have to transform it into a format that is machine readable. The reality is that these seemingly mundane tasks are 80% of their day job. In the asset management industry, there is a constant drive to find a trading edge or to find a new signal in a dataset that informs an investment decision. You only need to look at firms like Renaissance, TwoSigma and Citadel to find evidence of advanced data science driven strategies. They have invested millions of dollars into curating their own datasets and building teams to find value in their data. Anyone new to the game, however, has to start from scratch. This is why getting access to transformed, structured, labelled/tagged and machine readable data is so valuable.
<h3 class="MuiTypography-root MuiTypography-h3 css-1inx43p-MuiTypography-root">Initial success</h3>
In late 2020, we were developing a new machine learning model for an asset management client who was looking to identify the strategic direction of a company based on earnings call transcripts and annual reports. We started with the S&amp;P 500 and very quickly were able to get machine readable transcripts from Capital IQ along with companies&#8217; annual reports from the SEC Edgar database. The data was well formatted, tagged, easy to download and there was a lot of history &#8211; a data scientist’s dream! We used a combination of NLP to analyse what was being said in the text, and predictive machine learning algorithms to analyse the company financials. This approach resulted in an ecstatic client and an expanded mandate.
<h3 class="MuiTypography-root MuiTypography-h3 css-1inx43p-MuiTypography-root">The challenge</h3>
Filled with confidence (perhaps even a hint of arrogance) following our recent success, we set out to repeat the exercise on European companies. The earnings call transcripts were again, readily available and these were ingested and processed quickly. Next, we set out to add the European annual reports and filings, expecting to find the same clean dataset, beautifully structured as was held in the Edgar database. Instead, we found ourselves in a data-science nightmare. European companies do not have to report in a uniform, consumable standard, nor is there a central database to access files. They are free to do whatever they want and can publish the content anywhere, whether this be the investor page on their website, to LinkedIn. Put simply, there is no consistency of format, or filing. Given we had promised our client a quick turnaround (based on our first experience), we had a problem.
<h3 class="MuiTypography-root MuiTypography-h3 css-1inx43p-MuiTypography-root">The solution</h3>
In a scramble to find a reliable source of data that would help us avoid having to find and download thousands of PDF documents that we would have to manually clean and tag, we stumbled across an article by a small company called FDB Filings. They had apparently come across the same problem and built a business around it!
&#8211; Comprehensive coverage of European annual, interim &amp; press release reports
&#8211; PDF documents converted into a machine readable JSON format
&#8211; Link back to every original source document
&#8211; Point-in-time entity mapping
&#8211; Clear and well-defined tags on revenue guidance, executive speaker, business risks &amp; more
&#8211; 10 years of filing history
&#8211; Data accessible via an AWS S3 bucket
We set up a trial to test the data, and to our collective delight, the quality of their technology, data structuring and team was outstanding! We were able to ingest and score 10+ years of annual reports for 1,000 European companies, almost immediately and deliver back the results our client was expecting. The team at FDB had set out to solve a very hard data structuring problem and succeeded so much so that In November 2021 Insig AI acquired FDB.
<h3 class="MuiTypography-root MuiTypography-h3 css-1inx43p-MuiTypography-root">What we learned</h3>
Starting from scratch doesn’t necessarily require billions. This process demonstrates that a combination of thoughtful data structuring, good timing, and robust infrastructure, is fertile ground to create an AI-powered analytical platform, that delivers unbiased, uniform, value-add insights to an entire sector, across different markets, in an accessible, and easily digestible format. Financial Institutions are entering an entirely new paradigm, one where stakeholders and shareholders alike are squarely focused on sustainability and accountability. This is bringing a significant layer of complexity and cost to a financial institution. Technology and data companies are emerging with cost effective and transparent data scientific propositions that enhance the due diligence and risk management capabilities of an institution, while informing and improving investment decisions. For the heads of risk and sustainability at financial institutions, the choice of data provider is yours, but the time for acting, is now!

european-machine-readable-company-filings-a-data-scientist's-dream

European machine readable company filings: a data scientist's dream

Read the latest Pensions Expert article to hear the thoughts of our ESG Research Director Diana Rose on whether its time for pension schemes to consider natural capital.
You can read the article in full using the link below:
<a href="https://www.pensions-expert.com/ESG/Is-now-the-time-for-schemes-to-consider-natural-capital?ct=true">https://www.pensions-expert.com/ESG/Is-now-the-time-for-schemes-to-consider-natural-capital?ct=true</a>

is-now-the-time-for-schemes-to-consider-natural-capital

Is now the time for schemes to consider natural capital?

As part of International Women&#8217;s Day 2022, our ESG Research Director Diana Rose talks about the development of her role and her experiences in IFA Magazine #IWD2022
You can read the full article, including Diana&#8217;s top tips for Women starting out in the industry using the link below:
<a href="https://ifamagazine.com/article/iwd2022-women-in-finance-experts-reflect-on-their-experiences/">https://ifamagazine.com/article/iwd2022-women-in-finance-experts-reflect-on-their-experiences/</a>

women-in-finance-experts-reflect-on-their-experiences

Women in finance - experts reflect on their experiences

Our CPO Steven Cracknell spoke to Funds Europe about how there has been a shift away from the AI black box system, but not completely. Stating the difference today is that #AI and machine learning are a lot more accessible.
The full article can be found using the link below:
<a href="https://www.funds-europe.com/december-2021/technology-for-explainable-ai-look-in-the-white-box">https://www.funds-europe.com/december-2021/technology-for-explainable-ai-look-in-the-white-box</a>

technology-for-explainable-ai-look-in-the-white-box

Technology: For explainable AI, look in the white box

We spend a large part of our time working with text data and as a result we often find ourselves implementing some kind of NLP tool either as part of our product or on behalf of customers. One of those tools is Named Entity Recognition (NER) – a process for identifying entities, e.g. people and places, in text. There are a wide range of possible use cases for NER but ours are usually focused on investing where we have used it to augment search tools, automatically connect individuals and companies, and screen huge volumes of text for known bad actors. While we work with NER regularly, and with text data every day, we don’t consider ourselves to be experts in NER specifically – our interest is practical in nature and as such our experience is in implementing other people’s models rather than in developing new models from scratch.
This short piece contains some of the lessons we’ve learned: it’s primarily aimed at the potential NER user without much experience of the technology but we hope there will be something here too for the algorithm developers and vendors who may find it useful to see the experiences of a more narrowly commercially focused organisation.
<h2 class="MuiTypography-root MuiTypography-h2 css-ur62r9-MuiTypography-root">What is NER?</h2>
Before we go any further, for those who aren’t familiar with NER it is roughly speaking the process of going from this:
Elon Musk invites Vladimir Putin to connect on the Clubhouse app.
to this:
(Elon Musk)[PERSON] invites (Vladimir Putin)[PERSON] to connect on the (Clubhouse)[ORGANISATION] app.
As you can see there are basically two components to NER: (1) identifying the strings (i.e. words) which are entities, and (2) identifying the type of entity. The main benefit of this technology is that because it is automated you can process volumes of text that would be completely unfeasible for a human with the result that several tools and applications which were previously impossible become possible.
<h2 class="MuiTypography-root MuiTypography-h2 css-ur62r9-MuiTypography-root">One NER algorithm to rule them all?</h2>
One of the most immediately striking things when you attempt to implement NER in real use cases is that the available solutions are very generic and may not fit that well for your specific use case. One way in which this appears is in the labels used to categorise entities.
The overwhelming majority of solution developers and vendors use 4 categories of entity: people, places, organisations, and miscellaneous. This is mostly for the historical reason that these were the categories used in the CoNLL-2003 shared task which has acted as a touchstone for the industry ever since. Nor are the categories bad or unreasonable – given a task to group entities into four intuitive groups it is likely many people would arrive at something similar.
The problem is that reasonable though those categories may be they don’t necessarily make sense in every use case. In our use cases, for example, we typically don’t care about organisations as such but about companies. The difference being that ‘organisation’ is a broader category which includes things like the U.N. and the FBI.
The practical implication of this is that many off-the-shelf NER models may simply not be that useful for your application and may require some additional work or post-processing steps to deliver what you want. This situation seems to be improving as some more targeted vendors/developers are beginning to offer a wider range of commercially useful categories but given the number of possible entity categories we might want to target we are probably some way off having readily available models for all of them and it may not even be possible. We are also seeing more platforms which offer some ‘train your own’ capabilities and this may be a promising path.
What this touches on is that, practically speaking, it is a little misleading to think of NER as a generic problem with a generic solution. Perhaps one day there will be one algorithm to rule them all but today the best solution for identifying drugs in research papers is unlikely to be the same one as for identifying companies in regulatory filings. It may be more helpful for both consumers and producers to think of NER as a family of solutions to a range of different problems which share some formal characteristics. From a commercial perspective this is no bad thing because it is usually far easier to solve problems in a domain or application specific way than in a general manner as we will see below.
<h2 class="MuiTypography-root MuiTypography-h2 css-ur62r9-MuiTypography-root">What does good look like?</h2>
One of the surprising facts about implementing NER solutions in the wild is not just that it is tricky to work out which solution is best, but that it is also much harder than you might think to establish some objective and general measure of quality against which you can measure them.
If you wanted to know which NER solution was best for you, a reasonable starting place might be the publicly available benchmarking data. Unfortunately, our experience is that these benchmarks have not been especially useful. Any benchmark is only useful if it matches your intended use case in the relevant ways and we have found that the absolute and relative performance of NER algorithms on benchmarks was quite different to the performance on our specific data (mainly company filings and regulatory releases). Further, just as with a CPU or database benchmark, implementation details matter and can contribute significantly to the outcome. Even when the implementation details in a benchmark have been honestly chosen they may not be applicable to your circumstances and again result in the benchmark being unrepresentative.
These issues are really true of all benchmarking processes to a greater or lesser extent, but there are some features of NER that make it difficult even in theory to establish a general and objective measure of quality. Consider an algorithm which returns us the following:
Philip (Seymour Hoffman)[PERSON] was a prolific actor.
It correctly identifies “Seymour Hoffman” as a person, but misses the full name. Should this receive a full point, half point or no point? And what about cases in which the correct string is identified but the wrong label is applied? There just isn’t a right answer here which doesn’t depend on the context and use case – even among the relatively homogenous set of use cases we have encountered it has sometimes made sense to use very strict scoring criteria and at other times to weight it differently.
What this means is that if commercial circumstances allow it is far preferable to have (1) an independent benchmark which reflects the data you will be running the models on and (2) a scoring model which reflects the value of different kinds of partial scores in your use case. This might seem intimidating or impractical but we have found it more achievable than you might expect – we have built test datasets of &gt;90,000 tokens which have been hand labelled multiple times (with additional human reviews) in under a week with a small team. In general for situations where you are considering investing substantial resources into a solution or where you are sensitive to performance differences we have found the ROI to be compelling.
<h2 class="MuiTypography-root MuiTypography-h2 css-ur62r9-MuiTypography-root">Algos in action: YMMV</h2>
You could be forgiven for assuming that all NER solutions would perform about the same – after all they are all using roughly similar methodologies, designed by people with similar skills and experience, and tend to be in the same ballpark on claimed performance. This has not been our experience at all and we have seen very wide differences in performance on our independent benchmarks. For example below is a table showing the F1 scores (this is a value between 0 and 1) of different providers (OS = Open Source, COM = Commercial) in one of our benchmarks and you can see that in a sense the best performing model was almost twice as good as the worst.
<table style="border-collapse: collapse; width: 100%;">
<tbody>
<tr>
<td style="width: 10%;">Provider</td>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 10%;" scope="col">COM1</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 10%;" scope="col">OS1</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 10%;" scope="col">OS2</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 10%;" scope="col">OS3</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 10%;" scope="col">OS4</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 10%;" scope="col">COM2</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 10%;" scope="col">COM3</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 10%;" scope="col">OS6</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 10%;" scope="col">COM4</th>
</tr>
<tr>
<td style="width: 10%;">F1-Score</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 10%;">0.36</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 10%;">0.42</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 10%;">0.44</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 10%;">0.46</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 10%;">0.47</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 10%;">0.54</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 10%;">0.56</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 10%;">0.63</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 10%;">0.68</td>
</tr>
</tbody>
</table>
There are of course other dimensions on which one algorithm can be superior to another including speed (latency and/or throughput), ease-of-use, cost etc. NER algorithms are highly parallelizable so for self hosted solutions the limiting factor is usually going to be the size of the compute grid you have access to and the cost of running it. Here again the differences can be enormous and a range of 100x is not uncommon which, depending on the size of your dataset, may make some options unfeasible.
Another important and often overlooked consideration is that with a vendor API you do not typically have visibility into which model they are using or what changes are being deployed. There are ways to monitor the stability of a vendor algorithm in production but in practice these are cumbersome and not widely used – functionally these are black boxes. Again the exact use case matters, for use cases that are highly sensitive to changes or refinements in the underlying model you may need to self-host, seek guarantees from your vendor, and/or put a monitoring system in place.
A side note on performance is that we have seen some success using post processing steps on the outputs of the models. Because we tend to care about specific data types (in our case usually company filings) we can often achieve meaningful accuracy improvements with simple heuristics which exploit domain specific formal or content based features of our datasets. Below is the same table as above showing F1 scores after some very simple post processing steps were applied. We can’t promise that these results will be replicable in every case but for commercial applications this is certainly worth considering because the datasets being used will often have features that lend themselves to this kind of exploitation – for example, formal/structural characteristics or a tight domain focus.
<table style="border-collapse: collapse; width: 100%;">
<tbody>
<tr class="MuiTableRow-root MuiTableRow-head css-1hsm0p4-MuiTableRow-root">
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 16.7257%;" scope="col">Provider</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 10.708%;" scope="col">COM1</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 7.61062%;" scope="col">OS1</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 7.78761%;" scope="col">OS2</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 7.87611%;" scope="col">OS3</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 7.87611%;" scope="col">OS4</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 11.1504%;" scope="col">COM2</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 11.2389%;" scope="col">COM3</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 7.9646%;" scope="col">OS6</th>
<th class="MuiTableCell-root MuiTableCell-head MuiTableCell-sizeMedium css-g07xji-MuiTableCell-root" style="width: 11.2389%;" scope="col">COM4</th>
</tr>
<tr class="MuiTableRow-root css-1hsm0p4-MuiTableRow-root">
<th class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 16.7257%;">F1-Score</th>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 10.708%;">0.36</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 7.61062%;">0.47</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 7.78761%;">0.49</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 7.87611%;">0.51</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 7.87611%;">0.52</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 11.1504%;">0.61</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 11.2389%;">0.61</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 7.9646%;">0.67</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 11.2389%;">0.75</td>
</tr>
<tr class="MuiTableRow-root css-1hsm0p4-MuiTableRow-root">
<th class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 16.7257%;">Uplift (%)</th>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 10.708%;">0.2</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 7.61062%;">11.2</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 7.78761%;">11.4</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 7.87611%;">11.2</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 7.87611%;">11.3</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 11.1504%;">12.2</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 11.2389%;">9.7</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 7.9646%;">5.2</td>
<td class="MuiTableCell-root MuiTableCell-body MuiTableCell-sizeMedium css-6inj7x-MuiTableCell-root" style="width: 11.2389%;">10.0</td>
</tr>
</tbody>
</table>
<h2 class="MuiTypography-root MuiTypography-h2 css-ur62r9-MuiTypography-root">Takeaways</h2>
As people that work with text data everyday we are excited about the potential role NER technologies can play in a wide range of applications as they improve and evolve. To recap our observations for potential users and developers are:
Many solutions currently available are very generic meaning a little more effort may be required to deliver what you want. A definition of ‘good’ is less obvious than it might seem and benchmarking against your definition is both advisable and surprisingly achievable. In our experience performance varies greatly across different providers and it may be possible to improve results with some post processing steps.

some-observations-from-implementing-named-entity-recognition-algorithms-in-the-real-world

Some observations from implementing Named Entity Recognition algorithms in the real world

why-is-impactful-data-hard-to-find-and-what-it-means-for-the-future

Why is impactful data hard to find and what it means for the future?

In the past few years we have worked with many fundamental investors to help them get value from new data and over that period we’ve seen first-hand the problems funds routinely run into and the frustration that can arise as a result of them. Over time we’ve learnt that many of the problems stem from the same small group of mistakes and that these mistakes can be fixed. In this short article our goal is to describe and to codify two of the most common problems, to explain why they tend to occur, and to suggest some actionable remedies.
<h2 class="MuiTypography-root MuiTypography-h2 css-ur62r9-MuiTypography-root">The incompatibility problem</h2>
<h3 class="MuiTypography-root MuiTypography-h3 css-1inx43p-MuiTypography-root">Problem</h3>
One of the most common problems funds hit when bringing on new data or new data methods is difficulty getting the analysts/PMs to engage. A common pattern is that a data vendor or internal data researcher is pitching something they think should be interesting but find the analyst/PM is not engaging in the way they hoped. To the data person this can feel like outright irrationality and evidence that fundamental investors are not long for this world; to the analyst/PM it can seem like a waste of resources on hype driven but irrelevant work. In practice something like this situation often happens with sentiment scores: the data team works hard to produce scores which capture the positive/negative sentiment in a quantitative way and those scores may perform very well in a predictive capacity but when introduced to the fundamental investors after a brief spell of curiosity they fall flat and end up being ignored.
<h3 class="MuiTypography-root MuiTypography-h3 css-1inx43p-MuiTypography-root">Explanation</h3>
What has gone wrong in these situations is that the investor has been presented with data which is incompatible with their process. You can think of fundamental investors as machines for converting data into investment decisions, and like any kind of machine they need the right kind of input to generate the desired results – if you put diesel into your petrol powered car you won’t get very far. While no two fundamental investors will have identical processes, as a group they are similar enough for us to make some generalisations: the kind of inputs investors thrive on are things like company filings, earnings calls, broker research, industry analysis, and news. Those are datasets – and they are datasets, even if that is not how we usually think of them – which are compatible with the investment process.
What then makes one dataset compatible and another incompatible? Ultimately it comes down to understanding. For data to be compatible with an investor’s process they need to understand where to place it in their analysis, how to weigh it against other inputs, when to trust it, and when to ignore it. Sentiment scores typically fail on this test: it’s hard to know how to weigh a negative sentiment score against your higher than consensus earnings forecast, or positive M&amp;A expectations, or long term bullishness on demographic driven sales growth; and it’s hard to know when a sentiment score pointing in one direction should trump a thesis pointing in the opposite direction. Compatibility is not a fixed concept but relative to an investor: different investors may be able to work with different datasets and the same investor may, over time, find that what is compatible for them changes as their skills and experience change.
The roots of this problem lie in a common but mistaken – in the context of fundamental investors – approach to data and extracting insights from it. Many investors don’t encounter this problem until they start working with new data, so there can be a tendency to think that the problem is the data, but really the problem is what is being done to the data. You can see this by thinking about sentiment scores again: they turn something investors are very comfortable working with (earnings calls, say) into something very alien. The reason new datasets are especially subject to this is that many of them are not consumable by investors in their original or raw form, meaning there is an extra step required. This extra step introduces data science and data scientists into the mix (not necessarily in your fund, they may be on the vendor side) and this creates fertile ground for problems to appear.
A key part of the problem is that most data scientists are trained to think about data science as an independent activity instead of one which fits into a bigger pipeline. As a result they define ‘good’ in terms of certain objective measures like model performance and tend to optimize for improving these measures. This makes perfect sense in an environment where data science is the entire process – like in a quant fund – but works very badly when data science is occupying an intermediary role – like in a fundamental fund. In our experience funds do well when instead of focusing on objective measures of correctness instead they focus on compatibility, i.e. ensuring investors understand and are comfortable working with the data they receive. Data scientists in this environment need to understand that their role is not to do the best data science in a global sense, but the most useful data science.
<h3 class="MuiTypography-root MuiTypography-h3 css-1inx43p-MuiTypography-root">Solution</h3>
We have two general rules you can apply to help mitigate this problem. The first is to make the output of any data analysis, model, or signal directly, logically, and obviously impact or predict some company KPI or accounting line item. It follows from this that data scientists should avoid focusing on share prices or entirely abstract things like sentiment. This rule doesn’t prevent the data from being bad but it does guarantee that it is recognisably about something investors understand and will be compatible with their process. It also importantly gives the investor and the data scientist a common language to communicate in which allows for iteration and improvement as both parties become more familiar with the working practices and preferences of the other: &#8220;this just isn’t relevant to me&#8221; becomes &#8220;I don’t care about KPI X for company Y for these reasons…&#8221;.
The second rule is to do as little data science as is necessary, or in other words to give investors the rawest data they can handle. For obvious human reasons this can be difficult for data scientists to follow and to some it may even sound wilfully perverse. But there is a good reason, namely that every extra step in processing or each additional element of complexity adds more distance in understanding between the investor and the data.
Importantly we find reducing complexity gives the best overall results even when more complexity and processing is objectively better in terms of accuracy and performance. Organisationally this means bringing investors into the data process early and making sure they understand all of the important steps in the process even when those are technical. Once again this is a dynamic state of affairs: different investors have different degrees of understanding and any one investor can learn more over time and thereby increase the range of data that is compatible with them. It also means that investors who want to get more value from data must engage with their data teams and vendors and should not shy away from digging into the nitty-gritty of collection and processing.
<h2 class="MuiTypography-root MuiTypography-h2 css-ur62r9-MuiTypography-root">The actionability problem</h2>
<h3 class="MuiTypography-root MuiTypography-h3 css-1inx43p-MuiTypography-root">Problem</h3>
Another common way in which data can fail to deliver for the investor is that the insights from the data may not be actionable or may be uneconomical if the cost to produce them outweighs their value. As an example you might be able to track social media engagement for a consumer brand which demonstrates that it has meaningfully better engagement than its main competitors – but that is unlikely to support an investment decision in the absence of other views on the company. The problem here is that while you have an insight it is a weak insight, meaning that on a standalone basis it is not enough to motivate an investment decision. Weak insights are not necessarily a problem, in fact most insights from data are weak in this narrow sense, but are only useful if you can ensure that you have pre-existing investment theses to plug them into or if not to ensure the cost to produce them is low enough to economically use them in screening and background research.
<h3 class="MuiTypography-root MuiTypography-h3 css-1inx43p-MuiTypography-root">Explanation</h3>
This problem occurs because (1) it is hard for most fundamental investors to trade small insights in isolation and (2) because the process of mining insights from data is very costly. The first issue is really less about being fundamental as such, and more a byproduct of most fundamental investors being quite concentrated: investors who run concentrated portfolios have a higher bar for inclusion that an idea must meet than a more diversified manager with an otherwise identical strategy. In our experience investors running portfolios with &lt;100 names will often find that the insights they get from the data alone are insufficient to meet that bar (we should note that this is only generally true, in practice there are exceptions). This is one of the reasons why quantitative investors can often structurally pay more for certain data because it is easier for them to trade on all of the insights in a given dataset. The second issue comes down to the contingent fact that in 2021 data and the people who work with it are very expensive, which in turn means that the cost of insights from data are also very high. These are the two factors weighing on the ROI of your data strategy and keeping them in check requires a structured process and good discipline.
<h3 class="MuiTypography-root MuiTypography-h3 css-1inx43p-MuiTypography-root">Solution</h3>
We find one of the best solutions to this problem is to work ‘from the idea to the data’. When you decide to bring in new data to your investment process there are broadly two ways you can do it: you can begin with the data and mine it for insights (going from the data to the idea), or you can begin with fully or partially formed investment ideas and look for data which would support or weaken them (going from the idea to the data). Most fundamental investors, most of the time, will find that they achieve significantly better ROI following the second approach. In practice this means reviewing your positions/ideas and asking what data in an ideal world would help you to confirm or disprove an investment thesis, and then going to look for that data. When stated like this most fundamental investors find this intuitive but the difficulty comes in maintaining discipline when confronted with seemingly infinite datasets and institutional pressure to do more with data.
This amounts to running a kind of ‘actionability’ filter ahead of time and ensures resources are only spent looking at datasets where successful insight mining will have an investment impact. It has two additional benefits: first that it gives you from the outset clear metrics for success which helps cut through vendor marketing and technical blather, and second that it allows you to estimate (necessarily very roughly) ahead of time the value of a given insight and to budget accordingly. Initially you may want to run a review process over all of your holdings and ideas in a strict fashion, but over time this is likely to become less burdensome as you only really need to go through this exercise on new names and ideas. You will also likely find some data sources demonstrate their evergreen usefulness which you keep on even if holdings change whereas others are tactical, picked up and discarded as needed. As part of the review process you will need to think about the stability and longevity of your interests in companies and sectors, particularly as you consider situations where data cannot be collected after the fact, and therefore need to decide whether you are willing to invest today to answer investment questions in the future.

how-to-get-more-value-from-data-actionable-solutions-to-common-problems

How to get more value from data: actionable solutions to common problems

Like most people when possible we prefer to get our data from the official source, we also like it when companies are compelled to provide the data under pain of legal or regulatory penalties. One example of that is the National Storage Mechanism, the official document storage for companies listed in the UK – now maintained by the FCA, but for a long time outsourced to Morningstar. According to Listing Rule 9.6.1. companies are supposed to forward their annual reports to the FCA’s document viewing facility (i.e. the NSM) and issue a notification via an RIS (i.e. the RNS). Many people assume that when the source is official and the data collection is governed by regulations that the underlying data will be reliable – but here we show you a couple of examples demonstrating this is not always the case.
So if we look at Tesco we see the system working like it should, on 13th May 2020 they issue an RNS notifying the market that they have published the Annual Report and have submitted it to the NSM.
<img class="alignnone wp-image-2314 size-full" src="https://insigai.io/wp-content/uploads/insights/bad-data-image-0.png" alt="bad-data-image-0" width="822" height="400" srcset="https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-0.png?resize=200%2C97&amp;ssl=1 200w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-0.png?resize=300%2C146&amp;ssl=1 300w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-0.png?resize=400%2C195&amp;ssl=1 400w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-0.png?resize=600%2C292&amp;ssl=1 600w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-0.png?resize=768%2C374&amp;ssl=1 768w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-0.png?resize=800%2C389&amp;ssl=1 800w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-0.png?fit=822%2C400&amp;ssl=1 822w" sizes="(max-width: 822px) 100vw, 822px" />
If we check the NSM we find that annual report with the correct date:
<img class="alignnone wp-image-2310 size-full" src="https://insigai.io/wp-content/uploads/insights/bad-data-image-1.png" alt="bad-data-image-1" width="829" height="400" srcset="https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-1.png?resize=200%2C97&amp;ssl=1 200w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-1.png?resize=300%2C145&amp;ssl=1 300w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-1.png?resize=400%2C193&amp;ssl=1 400w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-1.png?resize=600%2C290&amp;ssl=1 600w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-1.png?resize=768%2C371&amp;ssl=1 768w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-1.png?resize=800%2C386&amp;ssl=1 800w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-1.png?fit=829%2C400&amp;ssl=1 829w" sizes="(max-width: 829px) 100vw, 829px" />
However, the system doesn’t always work the way it should. Particularly in the Morningstar era we routinely come across examples like this:
<img class="alignnone wp-image-2311 size-full" src="https://insigai.io/wp-content/uploads/insights/bad-data-image-2.png" alt="bad-data-image-2" width="900" height="164" srcset="https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-2.png?resize=200%2C36&amp;ssl=1 200w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-2.png?resize=300%2C55&amp;ssl=1 300w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-2.png?resize=400%2C73&amp;ssl=1 400w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-2.png?resize=600%2C109&amp;ssl=1 600w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-2.png?resize=768%2C140&amp;ssl=1 768w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-2.png?resize=800%2C146&amp;ssl=1 800w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-2.png?fit=900%2C164&amp;ssl=1 900w" sizes="(max-width: 900px) 100vw, 900px" />
This date is clearly incorrect: the 31st of December is the financial year end not the publication date (which is usually months after the financial year end). The same thing happens with Sage, where we can see the correct date from this RNS:
<img class="alignnone wp-image-2312 size-full" src="https://insigai.io/wp-content/uploads/insights/bad-data-image-3.png" alt="bad-data-image-3" width="900" height="119" srcset="https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-3.png?resize=200%2C26&amp;ssl=1 200w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-3.png?resize=300%2C40&amp;ssl=1 300w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-3.png?resize=400%2C53&amp;ssl=1 400w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-3.png?resize=600%2C79&amp;ssl=1 600w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-3.png?resize=768%2C102&amp;ssl=1 768w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-3.png?resize=800%2C106&amp;ssl=1 800w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-3.png?fit=900%2C119&amp;ssl=1 900w" sizes="(max-width: 900px) 100vw, 900px" />
But again on the NSM the ‘publication date’ is in fact the financial year end:
<img class="alignnone wp-image-2313 size-full" src="https://insigai.io/wp-content/uploads/insights/bad-data-image-4.png" alt="bad-data-image-4" width="900" height="74" srcset="https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-4.png?resize=200%2C16&amp;ssl=1 200w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-4.png?resize=300%2C25&amp;ssl=1 300w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-4.png?resize=400%2C33&amp;ssl=1 400w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-4.png?resize=600%2C49&amp;ssl=1 600w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-4.png?resize=768%2C63&amp;ssl=1 768w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-4.png?resize=800%2C66&amp;ssl=1 800w, https://i0.wp.com/insigai.io/wp-content/uploads/insights/bad-data-image-4.png?fit=900%2C74&amp;ssl=1 900w" sizes="(max-width: 900px) 100vw, 900px" />
The NSM is not the only party to blame, sometimes the company fails to carry out its obligations under the Listing Rules properly. If we look at Sage again, for 2018 and 2017 they did not publish an RNS when they released the annual report meaning there is now no extant public record of when that document was published. Some other FTSE 100 and FTSE 250 companies that have similarly failed at least once include Wood Group, Inmarsat, Arm, Sports Direct, Amlin, Polyus Gold, Imagination Technologies, International Power, and Aberdeen Asset Management.
This should serve as a reminder that (1) official sources are by no means immune from mistakes and (2) that something is a legal or regulatory requirement does not guarantee companies will follow it. On a more positive note the FCA’s recent decision to bring the NSM back in-house appears to have made a significant difference to the quality meaning we may have less of these issues in future.

bad-date-in-official-places-the-uks-national-storage-mechanism

Bad data in official places: the UK’s National Storage Mechanism

Our ESG Research Director Diana Rose recently spoke with Pensions Expert on how how machine learning and AI can help investors assess ESG standards.
To read her thoughts in full, use the link below:
<a href="https://www.pensions-expert.com/ESG/Machine-learning-can-help-investors-tackle-fast-fashion-ESG-issues?ct=true">https://www.pensions-expert.com/ESG/Machine-learning-can-help-investors-tackle-fast-fashion-ESG-issues?ct=true</a>

machine-learning-can-help-investors-tacklet-fast-fashing-esg-issues

Machine learning can help investors tackle fast fashion ESG issues

In the latest commentary piece, featured in ESG Investor, Diana Rose explains how artificial intelligence can expose evidence of empty promises and hold companies to account.
You can read her piece in full using the link below
<div class="video-shortcode">
<blockquote class="wp-embedded-content" data-secret="ZBQqWovgRs"><a href="https://www.esginvestor.net/esg-scoring-systems-are-giving-way-to-intelligent-analysis-of-disclosures/">ESG Scoring Systems are Giving Way to ‘Intelligent’ Analysis</a></blockquote>
<iframe class="wp-embedded-content" sandbox="allow-scripts" security="restricted" style="position: absolute; clip: rect(1px, 1px, 1px, 1px);" title="&#8220;ESG Scoring Systems are Giving Way to ‘Intelligent’ Analysis&#8221; &#8212; ESG Investor" src="https://www.esginvestor.net/esg-scoring-systems-are-giving-way-to-intelligent-analysis-of-disclosures/embed/#?secret=oK7zPxjmmQ#?secret=ZBQqWovgRs" data-secret="ZBQqWovgRs" width="600" height="338" frameborder="0" marginwidth="0" marginheight="0" scrolling="no"></iframe></div>

esg-scoring-systems-are-giving-way-to-intelligent-analysis-of-disclosures

ESG scoring systems are giving way to intelligent analysis of disclosures

When I co-founded Insig AI three years ago with my colleague Warren Pearson, we did so with a purpose to transform the asset management industry through AI and machine learning. This mission was based on multiple conversations we had with industry leaders where it became apparent to us that while many wanted to get the benefits of machine learning and AI, they simply did not know how to go about it.
Through companies like Citadel and BlackRock, we have seen just how successful machine learning and AI can be. Few in the industry, however, have their resources – to spend big and build teams who can create bespoke in-house solutions – and the challenge for them becomes how to keep up. This is where Insig AI’s suite of “out of the box” products come in. Our solutions allow investment professionals to see, interact, and experience their data in a way they never have before.
A core way in which we do this is through data visualisation. When we take a client’s data and apply our machine learning and AI tools to it, we are giving them the ability to interpret it through multiple lenses in a fraction of the time traditional tools would take. The success our clients are enjoying shows just how beneficial these solutions are. One client’s fund has so far outperformed by 30 percentage points, while another has experienced a 25% reduction in operational costs.
It’s important to note that these results are not reached by machine learning and AI alone, but rather augmenting the portfolio managers’ decision-making ability and judgement. The tools generate insights that may otherwise have gone unseen. Ultimately, however, investment decisions continue to remain with the portfolio manager. One of the major benefits AI and machine learning products gives to asset managers is speed in processing and investigating data. When once testing a hypothesis would once have taken five days to investigate can now be done in a matter of minutes.
As many asset managers know, their first instincts on an investment can often be far from the mark. Machine learning and AI products expedite this discovery process, so when they are wrong – they know it fast. This allows the user to move on to their next hypothesis rapidly. The success here lies in being able to investigate multiple theories at once, in order to find and validate the right path forward. During the discovery process, we also know that when weighing a decision, it is equally important to know the data and decision-making behind it. Therefore transparency is a core pillar of our company and our products are not “black boxes” – you can drill down every step of the way.
As new trends and industry sentiments emerge the ability to be flexible will be as important as ever. As the global asset management spend on AI is set to reach $8.3bn in the next five years, technology will continue to play an ever-increasing importance.
To this end, we are already developing a product called Insig ESG, set to launch in the third quarter of this year, which will develop our clients ESG strategy and execution. While our future products pipeline will look to continue to enhance the ability to mitigate risk, reduce costs, deliver better returns, and deliver products and services more efficiently. The transformative ability of AI and machine learning on Asset Management is no longer the domain of a few, and Insig AI is the perfect partner to help deliver it, now and in the future.

ai-offers-unparalled-insights-to-the-asset-management-industry

AI offers unparalleled insights to the asset management industry

As the demand for sustainable investing continues its rapid growth, so too does the importance of reliable ESG credentials and analysis. The uncomfortable truth is that not all ethical investments are what they appear to be, and it’s easy to lose sight of key ESG issues in decision-making. The challenge for investment managers, therefore, becomes how to truly see the wood for the trees.
At its heart, the rise of ESG is a response to demand for companies to do what’s right. If this seems hard to pin down, it is because it is – what’s right for one corporate now may not be right for others in the future. To find consensus and gain perspective on the big picture, while also gathering the detail that really matters, are real challenges that threaten ESG’s credibility. They are also where AI and machine learning can play a crucial role.
Traditionally, ESG credentials are based on the findings from a human analyst, meaning they can be skewed by bias and assumptions. These findings are then prone to being highly processed by obscure methodologies to deal with the challenge of aggregating disparate data. Meanwhile, to add a further challenge, the goalposts of what matters to whom are constantly shifting.
Furthermore, investment analysts are facing an ever-increasing amount of complex data and information relating to ESG. Gathering reliable data points, unpicking methodologies and reading between the lines of Sustainability Reports can seem like a monumental task. Going deep is one thing, but after all that hard work, data can still end up meaningless unless seen in the wider context of a decent benchmark.
This is where the ratings agencies such as MSCI and Sustainalytics stepped in to do a lot of the hard work of research, input, benchmarking and scoring. Asset Managers breathed a sigh of relief. But it soon emerged that a disparity in approaches was resulting in a lack of consensus and transparency in the outputs.
This was recently highlighted by analysis by State Street Global Advisors, as part of an 18-month due diligence process which looked at 30 ESG data providers. It found that ratings aligned on only half of the companies covered. As SSGA noted; “in choosing a particular provider, investors are, in effect, aligning themselves with that company’s ESG investment philosophy.” One thing has become clear: a simple ESG rating is no longer enough for asset managers to hang their hats on; proprietary analysis is still crucial in the quest for evidence and transparency to underpin their confidence in ESG investment decisions.
This is where AI and machine learning can add real value to the ESG analyst’s task as well as the investment manager’s confidence. When it comes to drilling deep into a company’s ESG disclosures, for example, machine learning can do in minutes what would take a team of people days. And as for going wide, AI is perfectly equipped to filter, organise and benchmark huge volumes of diverse data in ways an analyst can then digest and explore. AI can do a lot of the hard work, without introducing the same human bias and assumptions that can creep into aggregation methodologies.
Natural Language Processing (NLP) is a great example of AI’s ability to go wide, by harnessing the content of hundreds of thousands of sources for relevant data. NLP can find and contextualise text that is not in a structured format and make sense of it. The result is that relevant insights from live, ‘real world’ sources can be organised, ready for analysis alongside more formally reported information. This provides a new level of context and richness to the picture of an ESG investment opportunity. NLP can also act like an ‘ear to the ground’ in a shifting landscape of ESG materiality and risk, identifying and highlighting emerging trends that may affect a company – beyond what they’ve reported or are even aware of.
While we are still in the early stages of AI’s application to ESG, some key factors are set to rapidly accelerate its use. On climate change, for instance, it’s no huge surprise that some companies have been getting away with greenwashing – for example talking about ‘Net Zero’ without any strategy to get there. AI has the ability to help bring this to light, and public, political and regulatory scrutiny is on the rise; as Raba Foroohar recently pointed out in the FT “individual companies are being held explicitly responsible for the risks of global warming”.
These increasing standards and scrutiny are both overdue and necessary, making it an exciting time in the world of ESG. But the speed at which things are moving presents challenges for both companies and investors, and there’s a risk that transparency, evidence and ultimately credibility get compromised. As a partner to human decision-making in a complex and at times perplexing world, the power of AI is starting to be harnessed for ESG; if done right, it will add both quality and that all-important sense of perspective.

seeing-the-wood-for-the-trees-with-the-help-of-ai

Seeing the wood for the trees with the help of AI

The recent high-profile acquisition by Philip Morris International (PMI) of Vectura, a health tech company that produces inhalers, has been dominating headlines. Heated debate has been spurred as to whether such an apparent conflict of interest should be allowed, or whether it’s a sign that ESG momentum is biting and business is being forced to pivot for the better.
On one side, there is a sense of moral outrage that a company should profit from solutions to problems it contributed towards. There are also serious concerns expressed by health academics who believe the acquisition will allow Big Tobacco to wield undue influence on UK health policy. This response seems to boil down to a question of whether businesses with damaging legacies deserve to thrive now our eyes are open to their impact – are they beyond becoming sustainable.
On the other side of the coin is the more pragmatic view that these influential companies have a role to play in the transition away from harmful practices. This side argues that reactive divestment in bad news assets is not the answer; washing your hands of poor ESG stocks merely allows them to be snapped up by the less discerning. Ultimately, this leads to our environment and society being no better off.
So while PMI seems to have pulled off a highly strategic move that’s probably key to its survival and the whole thing makes us deeply uncomfortable, it begs the question: Does the motivation really matter if the overall outcome is positive? Sustainability is fundamentally about the long term view – how will we look back at this moment in its evolution?
It’s impossible to say, but critically, the first part of answering this question is to ensure that the outcome is positive and to do that, companies need to be held to account in fulfilling their ESG promises. It is the same logic when it comes to greenwashing. If a company sets long-term targets for emissions, ethics and equality but nobody checks in on whether they’re delivered upon, the company enjoys all the benefits of sounding virtuous without ever having to deliver or face the consequences.
Part of the problem is that littering reports with generic wording, vague long-term targets and catch-all terms has become dangerously easy. So much so that the EU has introduced new anti-greenwash rules forcing the asset management industry to categorise and qualify their ESG claims. And the EU is not alone, with the UK Competition and Markets Authority (CMA) recently putting business “on notice”. In a release published just last month, the CMA warned businesses they have until the New Year to make their environmental claims comply with the law. Further, it stated it would be carrying out a full review of misleading green claims early next year and “stands ready to take action against offending firms”.
Some see the greenwashing trend as threatening the whole ESG movement; others see it as teething problems that won’t hold back the wave of change. Again, what’s key to getting it right is to enforce accountability and responsibility. Consumer pressure, regulation and convergence of standards clearly have a role to play, but it’s still a tricky space for discerning investors. Assessing credibility and tracking delivery to targets is made challenging by the sheer volume of reporting resources published, and genuine disclosures can get obscured in confusion.
This is where we are starting to see the role technology can play in holding companies to account to having the impact they intend. AI and machine learning are emerging as practical tools that investment professionals can use to identify if corporate promises are being acted upon. Thoughtfully trained algorithms can do a lot of the heavy lifting of organising and processing vast volumes of corporate disclosures so that a researcher can more effectively and efficiently home in on metrics, read between the lines where things aren’t what they seem, and determine from the evidence whether words and actions match where it matters.
We are still in the early days of ESG and, as the PMI dilemma highlights, there are conflicts, trade-offs and strong arguments for principles and pragmatism that can pull in different directions. However, as the complexity we need to grapple with increases, so to do the tools and solutions at our disposal. It is my belief that ESG will create positive change over the long term, and AI and machine learning will play a valuable role in supporting us on the journey.

Insights