The Algorithmic Trading market is expected to reach US$31.49 billion by 2028. Algorithmic trading is a process of executing trading orders using an automated system. Algorithmic trading aims to use speed and accuracy that’s greater than a human to increase returns.
We spoke with one of our top algorithmic trading professionals to get their thoughts and analysis on some of the major questions on the algorithmic trading industry. Dive into what they had to say below.
Deep Technology and Algorithmic Trading Expert: Game Theory says that it was already about time. The data is flooding everywhere, the tech is affordable, the data scientists are more abundant, the news about its pros are all over the place… and that all ultimately reduces the risk of investing. Not launching algorithmic trading is becoming a signal of poor adaptation rather than over-ambition.
Algorithmic Trading and Software Developer: I think that the next big thing for the algo-trading field will be coming from the blockchain world which is still in the early stage of development. It might be a good idea for any company or algo-trading professional to continuously monitor what is happening with Decentralized Finance (DeFi), smart contracts, and Non-Fungible Tokens (NFT) topics, and think about how technologies in this space can be utilized. There are a lot of interesting technical innovations to keep an eye on.
Deep Technology and Algorithmic Trading Expert: Ideally, you should start by being guided by one of that handful of experts that I mentioned before. But not always we can reach those ideal experts. In those cases, we have to target optimal. And the following two steps are the optimal path from my personal experience and from what I have seen in the industry. Realistic and efficient.
First, you start standard, i.e. through Digitalization. You onboard data scientists and allow them to impress traders through the standard use cases of Machine Learning (features selection, graph theory, random forest, clustering, gigas of this, teras of that…). Then, you open up each one of your market access providers’ software and deploy the algorithms. By doing this you unlock the second step. You will, both, gain credibility through the delivery of new insights from data but you’ll also notice that you have started the whole thing in a disorderly manner. You will start wishing you had done things differently, through a bespoke middleware platform that exploited synergies, which you will eventually realize is far more complex to design and deploy than you think (it took SciTheWorld 5 years to evolve its platform).
Next, you need to align future efforts with current ones. You start devising not only your mid-long run but also that of your best competitors. You start to understand the difference between Digitalization and Algorithmisation – when the company is not just data-driven but it is becoming a platform that organizes new protocols across roles, upon data. As said, there aren’t many companies capable of providing such a platform in a bespoke manner with the credibility of the aforementioned handful of experts so, if you find one, you will take the chance to, in parallel, create a pet-project of Algorithmisation to start feeding the platform. This way, you keep the protocol of Digitalisation until Algorithmisation organically takes over.
Finally, you would have retrained your data scientists without friction – timely (as soon as possible to avoid change resistance) and pressureless (due to its pet nature at the beginning) – and you will start onboarding the rest of the roles around this crucial, bespoke platform. Every algorithmic project from that point onwards should meet the triplet “smarter, cheaper, further”.
Algorithmic Trading and Software Developer: The easiest way is to join a trading company that invests heavily in building internal trading systems and decision support systems using data science. One more option is to work for a financial or electricity exchange. Finally, of course, you can read some books, articles on the internet and probably learn by experimenting with cryptocurrency trading.
Deep Technology and Algorithmic Trading Expert: The major challenge remains subtle yet crucial in the overall industry discussion. It is fairly simple: who can do it? Who, internally and/or externally, has the merits across tier one machine learning, tier-one architecture design and tier-one trading to lead this transformation? Because it is precisely in that intersection where an optimal legacy can be born – and recall that legacies soon become burdens so you have to get them right no matter what. The bad news is that there are just a handful of professionals that have international recognition for the job.
Now, shall that fact stop it? No. Why? Because, in the end, you only need to be better than your competitors to win. So, while your competitors remain fairly basic, quite the overall state nowadays, you have the chance to jump in and find your way through via dynamic fine-tuning of your approach. In fact, that game-on pattern, very much linked to the startups’ culture, is probably here to stay.
Deep Technology and Algorithmic Trading Expert: First, it should incorporate your on-platform bespoke protocol. It takes time to decide how to click all the roles in one place. For that, it ought to seek maximal synergies exploitation: not only cross businesses/assets but, if possible, cross units (trading, research, asset management, wealth management, broker). Why? Their interaction unlocks the competitive advantages going forward (better products and new products & businesses). Then, it has to be fully actionable by the business experts – they have to have control, otherwise, the whole process can become untrustful. And then, several other features are key to grant the success of the venture: to follow an ‘Augmented Machines’ approach (where the expert provides robustness in areas where the machine is weak); to be fully compatible with third parties (IP protection and combination); to include cybersecurity by design; to become the data scientists lab (i.e. production equals testing); to become a training sandbox for traders and risk managers (so that timely chase by the latter can be granted); to serve as an audit tool for the algorithms.
This all can be achieved in one platform. Precisely because once you get the first of them correct, many of the rest become byproducts. For example, SciTheWorld’s use case for algorithmic audit was awarded as Best Innovation in Simulation at Cognition X; and its use case for 1-year prop-trading reached a return of 48% during 2020/2021.
Deep Technology and Algorithmic Trading Expert: The budget tends not to be a major problem on this challenge as success fees are certainly natural in this arena – easy to measure, justify and share. The minimal costs for the smallest budgets shall consider that. At the very least, you need to make sure that your provider does not lose money providing the service – architecture, maintenance, improvements, consultancy etc. shall all be covered. The largest figures are those of the licensing fees which are sunk costs and hence, can be negotiated towards aggressive success fees.
Algorithmic Trading and Software Developer: The way Intraday electricity is traded is very similar to how shares or currencies are traded in the financial markets. There is still a huge gap though as the technology is better utilized in the financial world. On the other side, it means a big advantage for companies with financial trading knowledge who want to explore energy trading.
Intraday liquidity is steadily increasing year over the year on NordPool, EPEX, etc. because of the green energy shift and government support. More energy producers and trading companies are trying to figure out how to trade smartly. Some of them trying to buy external algo-trading solutions to compensate for lack of knowledge in the trading space, while others are investing heavily into building their own solutions and hiring algo-trading professionals.