Article · 2026-08-11 · 6 moments

Software engineering at a proprietary trading company: Optiver

A shift from a focus on latency to building better AI models, owning the full stack from applications to building custom hardware, very different incentives to most tech companies in play, and more ✦ AI generated

Read original ↗

GUEST The Pragmatic Engineer GUEST Alex Itkin HOST Gergely Orosz

Claim◆

At Optiver, the shift from reducing latency to building better AI models is complete: 'latency is the floor' and AI information models have become the key differentiator, with the company now investing substantially more in building better models than in lowering latencies.

Optiver describes how trading has evolved from winning on pure latency to competing on AI models. Slow models with fast triggers send signals while fast edge models make realtime decisions, and the company now invests more in model building than latency reduction.

transcript

The Pragmatic Engineer: Today, latency is the floor, and AI models are becoming a differentiator. Gone are the days of having lower latency than the competition allowing for arbitrage opportunities to make risk-free profits. Instead, information models are becoming a differentiator: slow models with a fast trigger sending signals to execute trades, and fast models running at the edge of the network making trade decisions realtime.

explains mechanism · 1gives example · 2

Claim◆

Optiver's 'build and own' culture makes engineers responsible for their projects as if they were CEOs, with no concept of throwing work over a wall to QA, and ownership is baked into the interview, onboarding and daily work alike.

Optiver's ownership culture treats engineers like CEOs of their projects - designing, building, testing, deploying and monitoring. There is no QA handoff, new hires get real owned projects from day one, and within a year they become the experienced person in their domain, with ownership stressed in interviews and onboarding.

transcript

The Pragmatic Engineer: Optiver runs on an ownership culture, with the principle that the best engineers take work personally and care deeply about Optiver's systems, decisions, and outcomes. Leaders want engineers to treat their projects as if they were CEOs of a company, and be responsible for design, build, rollout, shipping, or support. There is no notion of throwing work over the wall to a QA team.

explains mechanism · 1

Context

Optiver's business model is fundamentally different from most tech companies because it has no external customers: its own business is the customer, which removes external deadlines and pressures but places high value on personal motivation and a 'build and own' ownership culture.

Optiver is a proprietary trading firm ('prop shop') that trades only its own capital, unlike investment banks or hedge funds. This means no external customers and no client-driven deadlines, with engineers operating on an ownership model where they design, build and own systems end-to-end.

transcript

The Pragmatic Engineer: No external customers. Usually, companies have consumer customers (B2C), business customers (B2B), or both. But not trading houses like Optiver, where their own business is the customer. This is a different reality: there's no external deadlines and related pressures, but personal motivation to improve is highly valued.

Mechanism

Optiver's engineering organization shifted from a fragmented regional 'unblock yourself' culture (1986-2020) to global platforms built for the whole company (2020-present), which was necessary because fragmentation into different regional technologies and duplicated services became too costly.

Optiver's tech went through two eras: regional systems where local teams moved fast and 'blocked themselves out' of duplication, then global platforms built for scale. The downside of fragmentation and duplication grew to outweigh the speed benefit, driving a consolidation started around 2023.

transcript

The Pragmatic Engineer: Regional systems ("unblock yourself": 1986-2020): internal systems and platforms were built to serve local needs, such as building support for a market. Systems built exclusively for the US, Europe, or Asia were common. Global platforms ("build for the whole company": 2020-present): Optiver recently started to build new systems to work globally across their platform. This global focus is also why the company is investing a lot more in its platform engineering arm.

Data

The trading evolution across four eras - from pre-electronic trading floors (pre-1990s), to the first wave of electronification (1990s), to automated trading (late 1990s-2015), to quantitative trading with ML (2015-present) - has weeded the market, so that today only a handful of truly large firms remain because serious prop shops must invest hundreds of millions in research clusters.

CTO Alex Itkin frames trading's evolution in four eras, from face-to-face trading floors to today's data-driven quantitative era. Each era weeded out players who failed to make the next leap, and today only a handful of big firms remain because serious research clusters require hundreds of millions in investment.

transcript

Alex Itkin: According to Itkin, competition has got tougher over time, while the number of serious players has decreased. Today, there are only a handful of really big firms, and one reason for this is cost: investment in research clusters – which serious prop shops all do – requires hundreds of millions of dollars.

explains mechanism · 1

Claim

Ultra-low latency is no longer a competitive moat in itself at trading firms; as competitors squeezed performance out of their systems, the focus shifted to fine-tuning trading strategies, and HFT now evolves faster than other industries because profitable strategies don't last long.

Optiver's fastest trading system operates in the sub-nanosecond realm, yet even this ultra-low latency is no longer a moat by itself. Since competitors have matched the performance, the competitive edge has moved to fine-tuning strategies and building better models, and HFT evolves faster than other industries.

transcript

The Pragmatic Engineer: However, in this niche, even ultra-low latency is no longer a competitive moat in itself. As competitors have squeezed performance out of their systems, focus has shifted towards fine-tuning of trading strategies. Today, Optiver invests substantially more in building better models than it does in lowering latencies.

Highlight slides

Latency Is the Floor, AI Models Are the Differentiator✦ from: At Optiver, the shift from reducing latency to building better AI models is complete: 'latency is the floor' and AI information models have become the key differentiator, with the company now investing substantially more in building better models than in lowering latencies.Optiver's 'build and own' culture✦ from: Optiver's 'build and own' culture makes engineers responsible for their projects as if they were CEOs, with no concept of throwing work over a wall to QA, and ownership is baked into the interview, onboarding and daily work alike.How AI Models Drive Trades✦ from: At Optiver, the shift from reducing latency to building better AI models is complete: 'latency is the floor' and AI information models have become the key differentiator, with the company now investing substantially more in building better models than in lowering latencies.

Related episodes

Video · BG2 Pod

The SpaceX IPO, Fable 5, AI Capex Update & Market Check w/ Gavin Baker, Andrew Fox & Clark Tang

→

Video · Syntax

Do You Read The Code? ⟡ HTTP Gets a NEW Method ⟡ Codex Gets Sol Ultra ⌁ Syntax Weekly ⌁

→

Video · a16z Podcast

The New Rule for Picking AI Winners | The a16z Show

→