6 ms·

Semi-Joke: Will/Could they fund this with a high frequency trading strategy?

frazar0 8y ago
Interesting point. However, I believe frequency trading relies on connections with extremely low latency, while satellite links are characterized by high Round Trip Time.
- mhandley 8y ago
  Geostationary satellites have high latency because they're so far up (36000 km). But dense low earth orbit satellite constellations can give lower latency than fibre over longer distances if you relay from satellite to satellite, because the speed of light in a vacuum is 1.5 times faster than in glass.
  - vasco 8y ago
    Low orbit is less of a vacuum
  - LeonM 8y ago
    That's why HFT's have switched to using RF communication years ago, as it has less delay than optical transmission over fibre. Most of they use radio towers, not satellites.
    - ls612 8y ago
      But even NYC to Chicago you have to bounce off several RF base stations due to the curvature of the earth, and each station will add latency. One LEO satellite can probably get LOS to both NYC and Chicago at the same time.
      - derekp7 8y ago
        Couldn't you build analog relay stations that add 0 latency if they are for a dedicated (non-switched) connection? Where the receiving antenna pipes the RF through an analog amp and sends it out the transmit side (possibly using filtering to clean up the raw signal)?
CoolGuySteve 8y ago
Hybernia from NYC to London is about 55ms whereas I remember reading somewhere these satellites would have a 30ms ping. So probably not for cross-Atlantic propagation once you add all the switching and hops, but maybe cross-Pacific might be doable since propagation delay in space is lower than in fiber.
usrusr 8y ago
If they do actually beat microwave towers, HFTs will be their first and most enthusiastic customers. SpaceX would not even need any specific strategy for the money to come from HFT. For biggest revenues, add steeply tiered pricing for the best latency, maybe even a bidding process. If they tried to keep those gains in-house and did their own HFT I would not be surprised if exchanges then found ways to define rules that weaken HFT as a whole. Just a gut feeling that you need to be part of the culture surrounding the exchanges to be in that game and keep winning.
- dx034 8y ago
  You don't even need to beat microwave. NYC-Tokyo or London-Tokyo don't have microwave links (and won't have anytime soon) so using it for these very long distances could be worth it.
gizmo 8y ago
Realistically, no. High frequency trading is a niche and not hugely profitable market. Not really worth disrupting when there is a much bigger opportunity in delivering internet access to billions of people and many billions of devices.
- hedora 8y ago
  You forget that there is no force in the universe as powerful as compound interest.

Interesting point. However, I believe frequency trading relies on connections with extremely low latency, while satellite links are characterized by high Round Trip Time.

mhandley 8y ago
Geostationary satellites have high latency because they're so far up (36000 km). But dense low earth orbit satellite constellations can give lower latency than fibre over longer distances if you relay from satellite to satellite, because the speed of light in a vacuum is 1.5 times faster than in glass.
- vasco 8y ago
  Low orbit is less of a vacuum
- LeonM 8y ago
  That's why HFT's have switched to using RF communication years ago, as it has less delay than optical transmission over fibre. Most of they use radio towers, not satellites.
  - ls612 8y ago
    But even NYC to Chicago you have to bounce off several RF base stations due to the curvature of the earth, and each station will add latency. One LEO satellite can probably get LOS to both NYC and Chicago at the same time.
    - derekp7 8y ago
      Couldn't you build analog relay stations that add 0 latency if they are for a dedicated (non-switched) connection? Where the receiving antenna pipes the RF through an analog amp and sends it out the transmit side (possibly using filtering to clean up the raw signal)?

CoolGuySteve 8y ago

Hybernia from NYC to London is about 55ms whereas I remember reading somewhere these satellites would have a 30ms ping. So probably not for cross-Atlantic propagation once you add all the switching and hops, but maybe cross-Pacific might be doable since propagation delay in space is lower than in fiber.

usrusr 8y ago

If they do actually beat microwave towers, HFTs will be their first and most enthusiastic customers. SpaceX would not even need any specific strategy for the money to come from HFT. For biggest revenues, add steeply tiered pricing for the best latency, maybe even a bidding process. If they tried to keep those gains in-house and did their own HFT I would not be surprised if exchanges then found ways to define rules that weaken HFT as a whole. Just a gut feeling that you need to be part of the culture surrounding the exchanges to be in that game and keep winning.

dx034 8y ago
You don't even need to beat microwave. NYC-Tokyo or London-Tokyo don't have microwave links (and won't have anytime soon) so using it for these very long distances could be worth it.

gizmo 8y ago

Realistically, no. High frequency trading is a niche and not hugely profitable market. Not really worth disrupting when there is a much bigger opportunity in delivering internet access to billions of people and many billions of devices.

hedora 8y ago
You forget that there is no force in the universe as powerful as compound interest.