Interesting!

But getting high bandwidth out of shortwave would be quite a challenge and need either extremely high SNR (high power and higher gain antennas) or a wide slot in the spectrum. The latency of any message is the time-of-flight plus duration of the message (bits/baud).

They must be sending very simple messages on this, e.g. "S&P 500 moved up by more than 0.1%". You don't need to be fancy if you're 10ms ahead of everyone else.

At the typical speeds of shortwave data systems, that "simple message" of 34 characters encoded in 6 bits+ 1 bit parity would take almost 800 msec to arrive.

No doubt it's tightly compressed in the most efficient way possible. And yet the maximum baud rate for shortwave data seems to be on the order of 300 baud. That's 3 msec per bit.

Anyway, where did you get you limit from? Remember that they are not broadcasting, it's directional communication (Chicago - London).

From a number of different sources online that talk about creating digital data communications (packet radio, RTTY, etc.) at various wavelengths. Shortwave seems limited to about 300 bps because the properties of the atmospheric medium suck (especially during the day). The problem is known as group delay and is caused by the ionosphere's reflection of the signal being extremely frequency sensitive. Attempts to send wide-band on shortwave are distorted by this. It's VHF and UHF that can support much higher bandwidths but they are line-of-sight.

I wonder if all this relatively small scale stuff shows that HFT trading is just not that lucrative. For example, if you really wanted trans-atlantic microwave, you could presumably create a network of, say 120 ships spaced 25 miles or whatever.  Say $120 million up-front for the ships, another $100 million/yr for crews, maintenance, and microwave. Doesn't seem that much if the profits were really there. Apple could spend two days of profits and do that ...

Anyway, where did you get you limit from? Remember that they are not broadcasting, it's directional communication (Chicago - London).

From a number of different sources online that talk about creating digital data communications (packet radio, RTTY, etc.) at various wavelengths. Shortwave seems limited to about 300 bps because the properties of the atmospheric medium suck (especially during the day).  The problem is known as group delay and is caused by the ionosphere's reflection of the signal being extremely frequency sensitive.  Attempts to send wide-band on shortwave are distorted by this.  It's VHF and UHF that can support much higher bandwidths but they are line-of-sight.

So do those limits apply equally to directional shortwave radio as much as to broadcast shortwave?

Yes. The sorts of people doing shortwave data do use directional antennas and sophisticated equipment to improve the gain. Shortwave radio enthusiasts both compete on long-distance transmissions (called DXing) and work with disaster relief and civil defense authorities to provide emergency communications. It's a rich person's hobby that attracts technically-minded people. It's also in military use (NATO RTTY operates at only 75 or 100 baud). If high-speed data on shortwave were possible, these hackers or the military would have created it.

I'll ask my cousin. He does research in high bandwidth low frequency by using a virtual wave.

Profits are getting smaller and the maintenance costs are getting higher every year. I personally find this process quite fascinating but i guess only the big players will survive because the small ones wont have the money to maintain the network system. HFT though, is what traditional makret making has turned into. So, it will never die, but the profits will be lower and lower right?

One question: Is working for a Quant Hedge Fund or a Prop trading (non market making) firm more stimulating and lucrative than a HFT firm? Ofc it depends, but what would you choose and why?