Equities & futures

The classic arena the canon was written about: lit exchanges, Reg NMS / MiFID II, colocation, maker-taker rebates. The speed game is a consolidated oligopoly, but research, stat-arb and execution edges still exist.

The idea

What is the equities and futures arena, in microstructure terms?

It is the mature, regulated, lit arena: shares and listed derivatives trading on registered exchanges under Reg NMS (US) or MiFID II (EU), with deep order books, small ticks, maker-taker fee schedules, fragmentation across many venues consolidated into a national best bid/offer, and session structure: opening and closing auctions, halts, and circuit breakers. The sandbox above lands on this arena's preset and emphasises its competitive reality.

Intuition first: this is the arena the textbooks describe and the canon was written about, namely Reg NMS equities and listed futures (CME, Eurex, ICE). The limit order book, price-time priority, the spread, queue value, signed flow: all of it was modelled here first. So the maths is most mature and best-validated in this arena; what makes it hard is not the model, it is the competition and the barriers.

The defining structural features: lit, regulated venues with mandated transparency and a consolidated tape; deep, fragmented books (the same stock on a dozen venues, with the NBBO assembled across them under the Reg NMS Order Protection Rule); small ticks and maker-taker rebates; and session structure, an opening auction, a closing auction (where enormous volume concentrates), intraday halts and LULD circuit breakers. Futures differ from equities in important ways (central clearing, a single primary venue per contract rather than Reg NMS fragmentation, different fee models) but share the lit, regulated, deep, fast, contested character, and the same canonical maths.

Why can't a newcomer compete on speed?

Because lit-equity and futures speed trading is a decade-deep infrastructure arms race won by a handful of firms with colocation, FPGA/ASIC hardware, microwave/laser links between datacentres, and the capital to keep buying the last microsecond. The edge is not a formula you out-think; it is a path you out-spend. A newcomer cannot match the fixed cost or the latency tier.

Intuition: latency trading on lit venues is latency arbitrage and queue-position racing, and both are winner-take-all in microseconds. Beating the field by 2 microseconds wins nearly every contested fill; losing by 2 microseconds wins none. So the game collapses to "who has the fastest path", and the fastest path costs tens of millions in colocation and FPGA hardware and inter-datacentre microwave links. You do not out-think that; you out-spend it, and a newcomer cannot.

The economics are brutal for the speed game: enormous fixed cost (colo, hardware, data, connectivity, staff), razor-thin per-trade margin, and a winner-take-most structure where only the fastest few are profitable. This is the capacity / alpha-decay reality that makes the arena a utility, not an open field. This is the deliberate contrast with crypto and prediction markets: there, the colocation/FPGA/prime-broker barriers are absent or cheap, so a small team can start. Here they are the whole game. The portability thesis (markets hub) is exactly this point: the maths is identical across venues, but the institutional barrier is highest here.

Who dominates, and why?

A small set of principal trading firms dominate lit-equity and futures market making and latency trading in 2026: Citadel Securities, Jane Street, Virtu, Optiver, IMC, Jump Trading, Hudson River Trading, DRW and a few peers. They dominate because the game rewards scale, speed and capital: more venues covered, faster paths, deeper pockets to absorb the fixed cost and the tail risk.

These firms are the liquidity providers of the modern lit market; Citadel Securities and Virtu alone handle a large share of US retail equity flow via payment-for-order-flow arrangements. They are not "the banks"; they are specialist electronic market makers and principal trading firms, and they won the arena by industrialising the canonical market-making maths and out-investing everyone on infrastructure.

Why scale wins: market making is a law-of-large-numbers business, with thin per-trade margins, vast trade counts, diversification across thousands of names and venues. A firm covering more instruments on faster infrastructure has lower variance and higher capacity than a small entrant, so the advantage compounds. The taxonomy of who is who is on market participants. The honest implication for the reader: you are not going to displace these firms in lit-equity speed market making. That is not defeatism; it is the same reason you do not compete with a utility on price. The right response is to apply the maths where they are not, which is the whole point of the markets layer.

Where do edges still exist for a newcomer?

Away from the speed race. Research-driven statistical arbitrage (better signals, not faster wires), execution edges (slicing large orders to beat impact), and corners the giants ignore: illiquid names, niche futures, specific event setups. None of these require winning the microsecond war; they require a genuinely better idea, honest backtesting, and discipline.

Research-driven statistical arbitrage. The classic pairs trade is competed away in liquid equities, but relative-value mean reversion at horizons slower than the speed game (where the edge is a better signal, not a faster wire) is still a research problem an independent can attack. The competition is on idea quality, not latency.

Execution edges. Trading a large position well (slicing it to minimise market impact via VWAP/TWAP/POV or Almgren–Chriss) is a genuine skill that does not require FPGA-grade speed. The edge is not paying away the spread and impact, which is a research-and-discipline problem. The corners the giants ignore: illiquid small-caps, niche futures, specific scheduled-event setups, and instruments where the fixed cost of coverage is not worth it to a firm optimising for scale. Small capacity, but real, and a place a focused small team can have an edge precisely because the incumbents will not bother.

The honest framing: these edges are narrower than the open frontiers of crypto and prediction markets, because the arena is so well-researched and so contested. But "you cannot win on speed" is not "you cannot make money"; it is "make money on research and execution, not latency". For most independents in 2026, though, crypto and prediction markets remain the more realistic starting arenas; see is HFT still profitable in 2026.

Worked example

A simplified lit-equity market-making round on a synthetic large-cap, mid = 100.00, tick = 0.01, maker rebate 0.20 bps, taker fee 0.30 bps, deep book, session-bound; illustrative, as of a 2026 worked snapshot. You quote bid 99.99 / ask 100.01, a one-tick half-spread each side, tight because the book is deep and competitive. But the queue at the touch is 5,000 lots deep and you are at the back: at this arrival rate, thousands of lots must trade or cancel before you fill.

Balanced case (if you fill both sides). Buy 100 at 99.99, sell 100 at 100.01, plus a 0.20-bps rebate on each posted fill. Round-trip gross is roughly the two-tick spread plus about 0.40 bps of rebate, but the per-round-trip edge is tiny (a couple of basis points), so the business only works at enormous trade counts across thousands of names. That scale is exactly what a newcomer lacks.

Adverse-selection case. The seller who hit your bid was an informed firm; the mid gaps to 99.96 right after. You bought at 99.99 something now worth 99.96, a $-3.00$ mark on the 100 lots before you react. In a deep, fast lit market the informed flow is often faster than you, so you are systematically picked off, paying the adverse-selection cost that the incumbents' speed and fair-value quality minimise and yours does not.

The closing auction. An enormous share of the day's volume prints in the closing auction, a single uncrossing where the speed game pauses and the game is sizing and pricing your interest correctly. This is one of the lit-market corners where a research edge, not a latency edge, can pay, and where a newcomer is on more equal footing. The lesson: the model is textbook (the same A–S quoting as crypto and Polymarket), but in this arena the queue, the speed and the scale (not the formula) decide the P&L, and all three favour the incumbents. The numbers are illustrative and synthetic; real equity/futures spreads, ticks, rebates, queue depths and auction mechanics vary by venue and instrument, so check the venue spec and the Reg NMS / MiFID II rules, as of 2026. Educational only, not investment advice; no P&L is promised.