Trading & Capital Markets · Chapter 4 of 8
The US market: fragmentation, Reg NMS and the NBBO
Sixteen exchanges, dozens of dark pools, and a rulebook built to make a fragmented market behave like one. This is where American market-structure jobs come from.
The American equity market is structurally the opposite of the Indian one. India concentrates liquidity in one dominant venue per segment; the US deliberately fragmented itself and then built a regulatory framework to stitch the pieces together. Almost every distinctively American market-structure job — smart order routing, best execution analysis, market data engineering, venue analysis — exists because of that choice.
The venues
- Exchange groups. NYSE, Nasdaq and Cboe each operate several separate exchange licences with different fee models, plus IEX, MEMX, MIAX and LTSE. The total sits in the mid-teens and changes as new venues launch.
- Alternative trading systems and dark pools. Venues that do not display quotes. Orders are matched without pre-trade transparency, typically at the midpoint. Used heavily by institutions trying to move size without signalling.
- Wholesalers / internalisers. A small number of very large market-making firms that fill retail orders from their own inventory. This is where most retail volume actually executes.
- Single-dealer platforms and systematic internalisers for institutional flow.
Reg NMS: the rules that make it hang together
Regulation NMS, adopted in 2005, is the framework. The parts worth knowing:
| Rule | What it requires | Practical consequence |
|---|---|---|
| Order Protection Rule (Rule 611) | You may not execute at a price worse than a protected quote displayed on another venue | Every venue must know every other venue's best quote in real time. This is what forces the whole market to be interconnected |
| Access Rule (Rule 610) | Fair access to quotes, and a cap on the fee to access a protected quote | The cap is what makes the maker-taker rebate model workable |
| Sub-Penny Rule (Rule 612) | Minimum quoting increments | Amended in the 2024 rulemaking to allow finer increments for tightly-quoted stocks |
| Market Data Rules | Consolidated data must be made available | Created the SIPs, and with them a permanent argument about their cost and speed |
The NBBO — national best bid and offer — is the highest bid and lowest offer across all protected quotes. It is a computed object, not a thing any single venue publishes, and computing it is a real engineering problem: you are consolidating feeds from many venues with different latencies and clocks, and the answer changes thousands of times a second in an active name.
The two-speed data problem, which is the heart of most US market-structure argument
There are two ways to see the market. The SIPs — the consolidated tapes — aggregate every venue's quotes and trades and publish one consolidated view. Everyone can buy it and it is comparatively cheap.
Or you can take direct feeds from each exchange and build your own consolidated view. This is faster, because you skip the aggregation hop, and considerably more expensive — you pay each exchange separately, plus the connectivity and the engineering.
The consequence is a market where some participants see the NBBO before others do. Whether that is a fair-access problem or simply the cost of speed is the argument that has driven a decade of SEC rulemaking, litigation over exchange market-data fees, and the founding of IEX — whose speed bump exists specifically to neutralise the advantage.
You do not need a view. You need to be able to explain the mechanism, because it is the single most common US market-structure interview question.
Maker-taker, and payment for order flow
Most US exchanges pay a rebate to the participant who posts a resting order and charge a fee to the one who takes it — typically fractions of a cent per share, with the taker fee capped by Rule 610. Some venues invert this, paying takers instead.
This creates a genuine conflict: a broker choosing where to route can be influenced by the rebate rather than by the client's execution quality. Best-execution obligations exist to counter it, and demonstrating compliance is what a large part of the transaction-cost-analysis industry does.
Payment for order flow is the retail analogue: a wholesaler pays a broker for the right to execute its customers' orders. It funded zero-commission retail brokerage in the US. It is banned in the UK and the EU has moved to phase it out, while the SEC ultimately pursued disclosure and execution-quality reporting rather than prohibition. Being able to state that divergence — same practice, three different regulatory answers — is a strong signal that you have actually followed the subject.
Post-trade: DTCC, and the netting that makes it possible
NSCC nets the day's obligations through continuous net settlement. If you bought 1,000 shares of a stock and sold 900, you settle a net 100 — and across the whole market the netting eliminates the overwhelming majority of movements that would otherwise have to happen. DTC then moves the net positions as book entries. Most shares never move at all.
The US moved to T+1 in May 2024. The interesting operational consequence was not domestic: it was that foreign investors lost most of their window to fund trades in dollars, because the FX market's own settlement conventions did not compress with it. That mismatch is an ongoing operational cost and a recurring interview topic for anyone in global operations.
0DTE, and the change in the options market
Zero-days-to-expiry options — contracts expiring the same day — grew from a curiosity to a very large share of S&P 500 index option volume, enabled by the exchanges listing expiries on every weekday. The parallel with India's weekly index options is close, and so is the regulatory unease. The structural question in both markets is the same: does very short-dated option activity amplify intraday moves through dealer hedging, and the honest answer is that the evidence is contested.
Check yourself