Things got interesting at the TradeTech FX US bank and non-bank panel, which delved into the key differences between third party and bank algos and brought together perspectives from both ends of the buy versus build spectrum on one stage.
Speaking to the most important points of consideration, Vittorio Nuti, head of the segregated execution businesses at Deutsche Bank’s FX and LD divisions, stated that one of the main factors is of course cost (including of maintaining), but that the most important aspect is performance, which “has historically been extremely hard to judge”.
“The most important thing is to determine what the client wants. When you’re speaking to someone you must understand what they really want to achieve – is it a rival price, is it speed? – different workflows might have different purposes.
“[…] Those are the sorts of conversations that you try and achieve and then cater and change the algorithms so that it fits the profiles. That’s why we really push for customisation.”
Mary Leung, global head of client algos at State Street Global Markets suggested that customising can be hard for third party providers, due to “different clients having slightly different flavours of how they want to execute something,” adding that “if you go to the vendor they will have built a product that can cater to most of the client needs, but not the specifics.”
Leung went on to add that this was the driving factor as to why State Street had built their own algos – able to be customised – and believes that the fact that the firm tries and tests their own algos before offering out to the market “offers clients a level of reassurance”.
Danielle Caravetta, director of global sales at Pragma Securities, agreed with Leung to an extent around the importance of algos being very heavily tested before deployment, also highlighting the value of algos being used internally first in order to “kick the tyres”.
However, Caravetta went on to suggest that market players “should all focus on what they’re good at,” suggesting that sell-side firms are not the experts in this field.
“No offense but when I think of Deutsche Bank and when I think of State Street, I don’t think of algorithmic trading. I think of liquidity, market expertise, risk transfer, and all the other things that a large bank can provide. I’m not necessarily thinking about technology.
“[…] We think that whether you’re a bank or a hedge fund you should be partnering where you can with specialists in the space.”
Ben Klixbull, head of Americas, e-liquidity management, global products at XTX Markets suggested that ultimately everyone does things differently, highlighting that despite bank algos and vendor algos clearly doing things differently, it comes down to what is quality execution, and how that is defined.
Klixbull added: “Each of us have a different flavour to what we’ve built […] trading in your own name, through a vendor’s algos is very different than using a broker’s algos because you’re constrained by the sources of liquidity that that broker has access to versus the sources of liquidity you can access in your own name.
“This is to show that it always has to come back to performance and cost of execution for the strategy.”