Microwave And Millimetre-Wave Trading Networks

Fiber-optic cable between Chicago and New York runs along railway lines and highways, curving around towns and terrain, and light inside glass travels at only about two-thirds the speed of light in a vacuum. A trading firm that needs to react to a price move in Chicago's futures market by trading the corresponding stock basket in New York cares about every one of those microseconds, because a rival who gets there first captures the arbitrage and leaves nothing for anyone behind. The fix isn't a faster cable — it's skipping the cable. A microwave or millimetre-wave link sends the signal as a radio beam through open air, hopping from tower to tower in a nearly straight line, and air is close enough to a vacuum that the speed advantage over fiber is real and measurable.

Towers instead of trenches

A microwave network is a relay chain: dish antennas mounted on towers spaced perhaps 30–50 km apart, each one receiving the signal and re-transmitting it toward the next tower in line. Millimetre-wave links use even higher frequencies, which carry more bandwidth over shorter hops but are more sensitive to rain and fog. Both approaches trade off against fiber's reliability — a heavy storm can degrade or briefly drop a microwave link, so firms often run microwave as the fast primary path with fiber as a fallback that never quite loses the race but never gets rained out either.

The commercial value of a marginal microsecond on these routes is large enough that firms have built literally dozens of competing microwave paths on the Chicago–New York corridor, sometimes just to shave a single-digit number of microseconds off the previous fastest route, because in latency arbitrage second place gets nothing.

Worked example: converting distance into an edge

Chicago to New York is roughly 1,200 km by a reasonably direct microwave path versus roughly 1,300 km along the fiber routes actually available. Signal speed in air is about $3 \times 10^8$ m/s; in fiber it's about $2 \times 10^8$ m/s (roughly two-thirds of light speed, due to the refractive index of glass). Time for each path:

$$t_{\text{microwave}} = \frac{1{,}200{,}000 \text{ m}}{3 \times 10^8 \text{ m/s}} \approx 4{,}000\ \mu s,\qquad t_{\text{fiber}} = \frac{1{,}300{,}000 \text{ m}}{2 \times 10^8 \text{ m/s}} \approx 6{,}500\ \mu s .$$

That's a gap of roughly 2,500 microseconds — 2.5 milliseconds — one-way, which sounds tiny until you remember that a firm racing a rival by even 5 microseconds on a link that's already been optimized down to a few milliseconds captures the trade essentially every time information first appears in one market and needs acting on in the other.

What this means in practice

Microwave and millimetre-wave links matter most for strategies that trade the same or closely related information across two geographically separated venues — index arbitrage between Chicago futures and New York equities being the textbook case. Because the advantage is measured in single-digit microseconds and the routes are public knowledge, firms treat their exact tower paths and link providers as closely guarded competitive information, and the arms race has largely reached a point of diminishing returns where further speedups cost enormously more than the last one bought.