1. HFT firms have definitely tried to perform HFT communication on HF bands before, see [1].
2. IIRC there have been some undecipherable transmissions in ham HF bands that were suspected to be associated with HFT, but I can't find a source right now.
3. While you don't have any additional latency from routers and serialization delays in the path across the pond, you need to keep in mind that long-range HF communications rely on bouncing the signal on the F1/F2 layers of the ionosphere (an atmospheric layer with lots of charged particles/ions that act as a mirror to HF signals). If you're comparing HF transmission with a hypothetical straight-line fiber connection (assuming it the propagation speed in the fiber is comparable to the atmospheric propagation speed [2]), the fiber connection would win.
Additionally, bandwidth is an extremely scarce resource in HF bands (compared to cellular networks or other common VHF/UHF systems). There's a reason people still use CW (morse code) or even PSK31 [3] instead of SSB (voice) [4] in HF bands: You need around 500 Hz for CW signals, around 3000 Hz for SSB signals. PSK31 only needs 31.25 Hz. My point is: While you might beat latency of fiber connections and routers along the path, the throughput is going to be quite low for HF transmission.
[2] Since you mentioned copper: Copper can be quite a bit slower than fiber depending on cable type, since the propagation speed of signals in copper cables depends on the velocity factor. Wikipedia gives a few examples: https://en.wikipedia.org/wiki/Velocity_factor
[3] PSK31 stands for Phase Shift Keying at 31 Bd. The short explanation is that PSK31 is a bandwidth-optimized text transmission mode.
[4] SSB stands for Single Sideband. The short explanation is that SSB is a bandwidth-optimized AM voice mode.
1. HFT firms have definitely tried to perform HFT communication on HF bands before, see [1].
2. IIRC there have been some undecipherable transmissions in ham HF bands that were suspected to be associated with HFT, but I can't find a source right now.
3. While you don't have any additional latency from routers and serialization delays in the path across the pond, you need to keep in mind that long-range HF communications rely on bouncing the signal on the F1/F2 layers of the ionosphere (an atmospheric layer with lots of charged particles/ions that act as a mirror to HF signals). If you're comparing HF transmission with a hypothetical straight-line fiber connection (assuming it the propagation speed in the fiber is comparable to the atmospheric propagation speed [2]), the fiber connection would win.
Additionally, bandwidth is an extremely scarce resource in HF bands (compared to cellular networks or other common VHF/UHF systems). There's a reason people still use CW (morse code) or even PSK31 [3] instead of SSB (voice) [4] in HF bands: You need around 500 Hz for CW signals, around 3000 Hz for SSB signals. PSK31 only needs 31.25 Hz. My point is: While you might beat latency of fiber connections and routers along the path, the throughput is going to be quite low for HF transmission.
[1] https://spectrum.ieee.org/wall-street-tries-shortwave-radio-...
[2] Since you mentioned copper: Copper can be quite a bit slower than fiber depending on cable type, since the propagation speed of signals in copper cables depends on the velocity factor. Wikipedia gives a few examples: https://en.wikipedia.org/wiki/Velocity_factor
[3] PSK31 stands for Phase Shift Keying at 31 Bd. The short explanation is that PSK31 is a bandwidth-optimized text transmission mode.
[4] SSB stands for Single Sideband. The short explanation is that SSB is a bandwidth-optimized AM voice mode.