■ HIGH RISK ■ Finance
For liquid electronic markets, this already happened — algorithms quote the spreads and humans were mostly gone years ago. What remains human is block trading, illiquid instruments, and supervising the machines, and even that list is shortening.
“High-frequency trading algorithms don't panic during market crashes. They cause them.”
Our AI replacement risk score — how we score jobs
Market making is the business of quoting both sides — always willing to buy at the bid, sell at the ask — earning the spread while managing the terrifying inventory risk of being everyone's counterparty. The human version lived on exchange floors and dealing desks: reading order flow, sensing when a big seller was working, widening quotes into uncertainty, and hedging with instincts built over thousands of sessions. That version is essentially a museum exhibit for liquid markets.
Equities, futures, FX, and listed options are now quoted by firms like Citadel Securities, Jane Street, and Virtu, whose systems update prices in microseconds, manage inventory across thousands of instruments simultaneously, and hedge automatically. Machine learning increasingly sets the quoting logic itself — predicting short-term flow toxicity, adjusting spreads to volatility regimes — and the humans employed are quants, engineers, and risk overseers, not traders shouting quotes. Even bond markets, long the human holdout because instruments are heterogeneous and trade rarely, are electronifying: algorithmic pricing engines now quote a growing share of corporate bonds, and all-to-all platforms erode the dealer's information edge. Our 65 score is really two numbers averaged: near-100 for the floor-trader version of this job, much lower for the quant-supervisor version replacing it.
The durable human seats cluster where liquidity is lumpy and relationships price risk. Block trades that would move the market need negotiation and trust; distressed debt, exotic derivatives, and off-the-run instruments trade on judgment about situations no training set covers; and when volatility breaks the models' assumptions — as it periodically, spectacularly does — humans decide whether to keep quoting or pull the plug. Risk oversight is also regulator-mandated. The career advice writes itself: the job now requires either a quant skill set or a franchise of relationships, and preferably both.
Automatability: our editorial assessment of current and near-term AI capability
The main event already ran: electronic market makers displaced human quoting in liquid markets over the past two decades, and the current wave is mopping up — algorithmic pricing spreading through corporate bonds and other slower markets right now. By around 2030, expect human quoting to persist only in blocks, distressed, and exotics, with everything else supervised rather than traded. The remaining hiring is quants and engineers, a different occupation wearing the same badge.
In liquid markets, comprehensively. Equities, FX, futures, and listed options are quoted by high-frequency firms whose systems price thousands of instruments in microseconds — the humans there build and supervise models rather than trade. The transition happened over the past two decades. Today's frontier is fixed income, where algorithmic pricing is spreading through bond markets that once ran entirely on dealer phone calls.
Where liquidity is scarce and trust prices the trade. Block transactions need discreet negotiation so the market doesn't front-run the order. Distressed debt, exotic derivatives, and rarely-traded bonds require judgment about one-off situations models haven't seen. And in genuine crises, humans make the quote-or-withdraw call. These niches are real but narrow — and electronification keeps nibbling at their edges.
Yes, if you mean the modern version: quantitative researcher, trading-systems engineer, or risk specialist at an electronic market-making firm. Those roles are lucrative and in demand. The classic path — junior trader learning to read the tape and quote by feel — has essentially closed in liquid products. Enter through math and code, or through credit and blocks where relationships still trade.
They've contributed to some spectacular ones — flash-crash events have shown how quickly automated liquidity evaporates when algorithms simultaneously widen or withdraw, turning a dip into an air pocket. Regulators responded with circuit breakers and market-maker obligations. It's the job's central irony: the systems that quote more reliably than humans in normal times can vanish faster than humans in abnormal ones, which is precisely why human risk oversight remains mandatory.