ODDO BHF fund uses artificial intelligence to identify nuggets in which to invest

[ad_1]

On December 18, 2018, ODDO BHF, a Franco-German financial group specializing in asset management, announced the launch of ODDO BHF Artificial Intelligence. This thematic fund, which required 18 months of research, aims to identify actions in companies of all sizes, on a global scale, and which develop or use artificial intelligence.

But ODDO BHF also decided to use artificial intelligence to better identify the nuggets of AI specialists around the world. It uses it to precisely characterize the investment theme and the most relevant companies linked to it, as she summarized at a conference held in Paris on February 12, 2020.

Find yourself among giants and a myriad of small actors

The development of this tool was born out of an observation. At a time when artificial intelligence techniques have an impact on all sectors of the economy, its ecosystem remains complex with "on the one hand the big technological players and on the other hand a myriad of small players specialized in fast growing niches"says Brice Prunas, global thematic equity fund manager at ODDO BHF Asset Management.

But technology companies are not only concerned. The fund also looks at companies directly involved in the AI ​​value chain or using AI to improve their own value chain. "There are so many actors, news, innovations and information to assimilate that we need the support of artificial intelligence to filter the market, identify important information and finally find the relevant titles.", continues Brice Prunas.

three filters combining AI and human validation

 6 main sectors affected by artificial intelligence according to Oddo BHF

– Health: the large pharmaceutical laboratories positioned on oncology and immunotherapy use AI and data from the human body to determine the most effective treatment for such a patient or to improve clinical trials; medtech, via radiotherapy and more precise identification of the areas to be treated, as well as robotics applied to surgery.

– Financial services: investment banks, which use AI to perfect analyzes, such as Morgan Stanley; insurers or credit organizations to improve customer scores, personalize offers.

– Consumer goods, in particular smart health connected objects, which represent an estimated market of 180 billion dollars. Companies specializing in wearables that thanks to AI offer new innovations.

– The Internet: in addition to Gafam, less known and very innovative companies, such as CarGurus (used vehicle sales platform), Splunk and Altricks (data prediction tools), companies specializing in cybersecurity (Chekpoint, Fortinet, Palo Alto or Cortex).

– Software: a sector in full revolution thanks to machine learning.

– Semiconductors: companies that collect data, the raw material of AI, and which are vectors for dissemination, with, among other things, the promises of the automotive sector, with an increase in the value of equipment, and of course those of 5G; companies like AMD that create value by launching innovative product cycles.

The company has developed a model that aims to provide investors with a full exposure to the companies that make artificial intelligence today. It identified 3,500 listed securities – out of a total of 40,000 securities worldwide – and six major sectors disrupted by artificial intelligence: health, finance, consumer goods, the Internet, software and semi -conductors (see box).

In a very simplified way, the selection is divided into three phases. The first studies so-called unstructured data – estimated at 80% of global data – or semantic data, scrutinized daily in the media. To determine the companies related to artificial intelligence, a lexical library was created, with 70 keywords divided into themes and sub-themes. For example, the Deep Learning category has a GPU subcategory and machine learning identifies the company Nvidia as one of the actors to analyze. Another criterion, emotions: the tool analyzes whether the articles are positive or negative. For example, the opening of a research laboratory relayed by the press will be considered a positive sign, unlike the departure of a financial director. Note, mentions on social networks are not voluntarily studied.

The second phase involves adding a layer of structured data – that is, corporate financial and market data – provided by the American media Bloomberg. 4 million data are filtered "by a proprietary multifactorial quantitative model"(baptized" Algo 4 ") which determines the three best titles of the day. Criteria which allow to exclude certain companies which lose money despite their valuation, and to keep only profitable titles. Netflix does not pass this selection, just like many companies specializing in autonomous driving.

The third phase is to have the list obtained by humans verified. The management team, made up of three people, who works with five data scientists, then proceeds to a corrective process, which removes 8 to 10% of the companies. Cognitive biases are studied, such as the over-representation of large companies, the artificial communication of entities, the reputational risk that can cause a fall in value, the appearance of a new competitor … Every 6 months, the keywords are redefined, with a rotation of 10%.

Buzz word versus real value creation

The methodology therefore makes it possible to sort and filter companies whose common point is, at very different levels of development, artificial intelligence. Become a real buzz word, artificial intelligence is today a victim of its success. According to an Infosys study published in 2019, 86% of companies declare having carried out artificial intelligence deployments at an intermediate or advanced stage. "But we have to determine if these developments really create value", warns Brice Prunas.

With the fund taking the MSCI World Net USD index as a benchmark, US companies make up 60% of the list, with 20% from Asian companies and 20% from Europeans. French companies, including STMicroelectronics, represent 1.5% of the total. "The United States represent the most important source for listed companies", says Brice Prunas.

In a context of strained economic relations with the United States, few Chinese companies appear there (Alibaba is an exception), despite their dominance over facial recognition technologies. On the other hand, there is a good representation of the Japan, very active in robotic solutions and semiconductors. Assets under management amounted to 80 million euros from institutional and private investors as of February 12. ODDO BHF specifies that its tool will be duplicated for other specializations. The company manages assets amounting to 59.8 billion euros.

Aude Chardenon

 @ChardenonA 

[ad_2]