In January 2018, HC Insider published a white paper by Etienne Amic, veteran energy trader and founding partner of CommodiTech Ventures, on the emergence of Commodities 2.0. Almost three years later, we find out from Amic what the current situation looks like as Covid-19 escalates the adoption of technology in the commodities industry.
In January 2018, when I first wrote about the coming of Commodities 2.0, Apple, Amazon, Google and Microsoft were racing to trillion-dollar valuations. Technology was starting to pervade all aspects of our lives as consumers and citizens, but the adoption of technology in the commodity trading industry was still lagging behind other sectors.
More than two years hence, the situation is similar, except that everything has become significantly bigger. The tech behemoths’ valuations are now comfortably above US$1 trillion and, in the case of Apple, well above $2 trillion, with Covid-19 continuing to force the adoption of technology on companies. While the gap of adoption in commodity trading remains, some technical hurdles have been cleared, paving the way for an acceleration of technology penetration in trading processes.
The picture is emerging of an industry that is immune to outright disruption by outsiders thanks to its low margins and very high level of expertise. Barring the acquisition of a trading house by Amazon, commodity trading is unlikely to know the fate of telecommunications, brick-and-mortar retail, or media. Its natural path is to have fewer white-collar workers augmented by a lot more software and automation.
This is good news for the large traders. The same is true for service providers such as trade finance banks or inspection companies, all the more so since the former have been hit by losses related to fraudulent activities.
As science fiction writer William Gibson famously observed: “The future is already here. It’s just not evenly distributed yet.” So, what are the visible seeds of that future for Commodities 2.0 and how will things be different when these seeds are the new normal?
There are three technological vectors currently making headway into commodity trading: the cloud, big data/machine learning, and blockchain. Each one of these trends feeds the acceleration of the other two.
“With this onslaught of analytics, traders could be forgiven for feeling that a lot of their proprietary activities have become a little too visible to the outside world. That is where the privacy-preserving technology of blockchain comes in.”
At first glance, cloud usage already seems overwhelming in corporations. Research and advisory company Gartner has calculated that, at the end of 2019, 90% of corporations were using cloud computing in one form or another, with most using a combination of hosted and public cloud. Still, 50% of companies continue to run applications in traditional, on-premises contexts, and Morgan Stanley estimates that, despite being a decade old, enterprise SaaS/cloud-only accounts for 20% of the workflows or technology workloads of large corporations (see image below).
That share is predicted to jump to 45% by the end of 2022. In other words, cloud adoption is just starting. Although there are no sectorial figures for commodity trading, anecdotally, every software publisher in commercial CTRM, terminal operations or ship management, seems to be in the process of webifying their application.
The great thing with the move to the cloud is that it creates a powerful incentive to re-architect applications into micro-services – collections of individual functional building blocks that each do one thing and do it very well.
As the technologist Tim O’Reilly remarked in his book WTF?: What’s the Future and Why It’s Up to Us, in a 21st century mindset, web services are the workers and their programmers are their managers. Each day, these ‘managers’ take in feedback about their workers’ performance, measured in real-time data from the marketplace, and if necessary, they give feedback to the workers in the form of minor tweaks to the algorithms. At an e-commerce site, one electronic worker helps the user find possible products that might match his or her search. Another shows information about the product. Yet another presents a web form requesting payment and validates the input. Another worker creates an order and associates it with the customer’s record. Another constructs a warehouse pick list to be executed by a human or a robot.
In an earlier generation of computing, these actions might be taken by a single monolithic application. Now it will be microservices communicating with each other through application programming interfaces (APIs). The resulting applications are much more open than before. When the move to the cloud is complete, all the parties in a physical commodity transaction — traders, brokers, inspectors, terminals, trade finance banks and port agents — will be able to exchange data with their partners much more easily than today, in digital formats such as XML and JSON.
Big data and the use of machine learning for analytics are fast becoming ubiquitous. Three years ago, tracking real-time waterborne oil flows and monitoring onshore inventories were still considered somewhat speculative technologies. Today, they have decisively crossed the credibility and adoption thresholds, with Vortexa, Kpler and Kayrros leading the way.
While the fine tuning of their machine learning algorithms has definitely helped improve the accuracy of predictions, a fundamental underlying reason for the success of these applications has been an increase in the proportion of “hard data” ingested by the algorithms, for example, actual fixtures for ships or inspection reports for cargoes.
“There are three technological vectors currently making headway into commodity trading: the cloud, big data/machine learning, and blockchain.”
This trend will only accelerate as the physical service providers to the commodity trading industry and the customs authorities move their own applications to the cloud and build APIs that can feed the data pipelines of the analytics firms directly.
There is still a lot more value left to be extracted from the automation of analytics. Today, the Signal Ocean Platform is striving to optimise fleets of tankers and dry bulk vessels algorithmically, with a view to improve returns for shipowners. OilX has also developed a broad perspective on fundamentals and builds what it calls “the world’s first digital oil analyst”.
Soon, commodity prices will be mixed with fundamentals data in the algorithms and it will be possible to answer causal inference questions such as, was it the price differential providing the trigger for the flow, or did the movement of cargoes cause the move in the price differential?
Similarly, calculating the historical supply and demand for ships on a given route, based on all the potential movements that could have satisfied the demand at a given location, is a computational task that is probably one or two orders of magnitude bigger than the current calculations on realised commodity flows. That too is now within reach of modern analytics. It is poised to change the way freight traders think about their art.
A natural next step will also be to mix publicly available data and proprietary data, in effect giving access to data pipelines and models for private insight by clients.
Beyond analytics, machine learning has started to be used routinely for the automated reading of logistics documents such as bills of lading, certificates of origin or barge receipts for further consumption by platforms. The imminent next step is the use of federated learning on blockchain, allowing competitors to improve the accuracy of their automated reading by sharing the learning with their peers, but not the underlying documents. Artificial intelligence and machine learning are only beginning to emerge from their formative stage – and the peak of the hype cycle – into a period of more practical, efficient development and operations.
With this onslaught of analytics, traders could be forgiven for feeling that a lot of their proprietary activities have become a little too visible to the outside world. That is where the privacy-preserving technology of blockchain comes in. Distributed ledgers may have acquired somewhat of a bad name in commodity trading — and wrongly so.
In their original hype phase, they were touted as solutions to problems which did not exist. Why use a distributed database in European gas and power when there is an independent network operator providing a perfectly fine nominations system, which is furthermore compelled by law to make aggregated data available to all while protecting individual traders’ nominations. Similarly, the tokenisation of commodities such as oil may be intellectually appealing, but it does not work in practice (just yet!) given the lingering complexities surrounding bills of lading.
On the other hand, where blockchain makes a ton of sense is in the post-trade processes of decentralised commodity supply chains, such as physical oil or agricultural commodities. That is the reason why companies like VAKT or Covantis were created by consortia of traders in the first place.
The logic is easy to understand. If the industry was left to its own devices, a physical trader would have to connect its (preferably cloud-based) systems to all the inspectors, all the terminals and all the trade finance platforms it uses — each with a different API. That is a lot of connectors to build and maintain. Also, all the other traders would have to re-build the exact same interfaces, creating wasteful duplication for all. If, on the other hand, a digital infrastructure with standardised APIs is available, a trader will be satisfied that it only has to build one connector to the platform.
Beyond efficiency, a trading house will also be focused on protecting its proprietary transaction data from prying eyes. Its relationship with terminals, surveyors, brokers and other traders or clients is the core of its business and there is a history of platforms reselling clients’ data back to them.
By marrying these two competing objectives — centralisation of workflows and privacy-preservation — blockchain finds it full potential. As a bonus, smart contracts are able to perform simple tasks on the digital infrastructure to augment human employees —providing a new kind of ‘software worker’ that is able to receive and forward nominations, or confirm a payment instruction or an invoice when data is cross-validated by all the parties to a transaction.
The three technological trends mentioned above have the potential to feed each other strongly, and to some extent they already do. The transition to the cloud creates an opportunity for easier trader-to-trader, trader-to-terminal, or trader-to-inspector digital communication by opening up applications to the standard web payloads — as opposed to unstructured emails which are harder for machines to read.
Analytics allows insights to be gained from data, either to optimise in-house operations or for trading purposes. In turn, it acts as an incentive to digitise operations and create open APIs to easily extract its inputs. In markets which have long resisted digitisation, the decentralised supply chains of oil, agriculture, and metals, blockchain underpins a cloud-based, privacy-preserving digital infrastructure which requires the standardisation of workflows and reference data. This in turn leads to reduced transaction costs and an opportunity to usher in new electronic platforms – or accelerate the development of the existing ones.
The venture capital firm Bessemer Venture Partners has observed a renaissance of B2B commerce across multiple industries, driven by four underlying drivers:
These four factors are very relevant to today’s commodity markets and I cannot wait to see how a younger generation of entrepreneurs and traders will reshape them to fit their needs and aspirations. The job of the current builders will be to make sure that they have an underlying digital infrastructure that they can connect to and where they can offer new services of their own invention.