Live data from Hacker News

Companies Pitch Shortwave Radio to Shave Milliseconds Off Trades

bloomberg.com

71–80 of 85 posts

Re: Companies Pitch Shortwave Radio to Shave Milliseconds Off Trades

#71
post #23
post #15

I remain convinced that this is Starlink's short-term customer if their promises are true. They are doing transmission via optical as the crow flies. If they're doing O-O, or even low latency O-E-O, they can beat transatlantic fiber. For HFT houses, that's potentially a major edge.

What happens in the global market if a single person consistently has a timing advantage over every other participant(because the barrier to building a global satellite network is too high/costly for anyone else)? Could Elon come to own the entire market, or is there some limit to the powers of arbitrage?

it'd be rather interestink.

but shortwave will be competitive with spacex, although at, of course, lower capacities.

for inter-market trading, sub-ms latency differences don't matter that much.

Re: Companies Pitch Shortwave Radio to Shave Milliseconds Off Trades

#72
post #70

Earlier quoted context omitted.

Starlink latency would be worse than shortwave. The advantage would be bandwidth, but I doubt that high bandwidth is really helpful for this application, especially if you have to give up a tiny bit of latency for it. After all, you can already have all the bandwidth you want with only slightly higher latency.

starlink latency could be faster than shortwave, once full gen2 constellation launches, and if starlink decides to care about this industry. ...disclosure: i'm the first one to build NY-Chi microwave, and Chi-Lon shortwave.

Cool! How do you figure? Doesn't shortwave take the great circle path, at lower altitude, making the path shorter?

Re: Companies Pitch Shortwave Radio to Shave Milliseconds Off Trades

#73
post #70

Earlier quoted context omitted.

starlink latency could be faster than shortwave, once full gen2 constellation launches, and if starlink decides to care about this industry. ...disclosure: i'm the first one to build NY-Chi microwave, and Chi-Lon shortwave.

Cool! How do you figure? Doesn't shortwave take the great circle path, at lower altitude, making the path shorter?

gen2 constellation is at 350km agl shortwave f-layer is at 200-400km agl depending on space weather

difference is not that big

Re: Companies Pitch Shortwave Radio to Shave Milliseconds Off Trades

#74
post #73

Earlier quoted context omitted.

Cool! How do you figure? Doesn't shortwave take the great circle path, at lower altitude, making the path shorter?

gen2 constellation is at 350km agl shortwave f-layer is at 200-400km agl depending on space weather difference is not that big

Sure, but it seems to me like the Starlink path will still be longer in almost all cases unless you're incredibly lucky and the satellites line up exactly along the path for a few seconds. And Starlink could have additional delays related to routing and possibly TDMA. What kind of latency issues need to be considered with shortwave?

Re: Companies Pitch Shortwave Radio to Shave Milliseconds Off Trades

#75
post #70

Earlier quoted context omitted.

Starlink latency would be worse than shortwave. The advantage would be bandwidth, but I doubt that high bandwidth is really helpful for this application, especially if you have to give up a tiny bit of latency for it. After all, you can already have all the bandwidth you want with only slightly higher latency.

starlink latency could be faster than shortwave, once full gen2 constellation launches, and if starlink decides to care about this industry. ...disclosure: i'm the first one to build NY-Chi microwave, and Chi-Lon shortwave.

It's also bandwidth right? What's the throughput on shortwave? I believe that Starlink promises mbps+. I imagine there are applications you can implement on higher bandwidth are more interesting?

Was Pilosoft always an HFT thing?

Re: Companies Pitch Shortwave Radio to Shave Milliseconds Off Trades

#76
post #73

Earlier quoted context omitted.

gen2 constellation is at 350km agl shortwave f-layer is at 200-400km agl depending on space weather difference is not that big

Sure, but it seems to me like the Starlink path will still be longer in almost all cases unless you're incredibly lucky and the satellites line up exactly along the path for a few seconds. And Starlink could have additional delays related to routing and possibly TDMA. What kind of latency issues need to be considered with shortwave?

a single orbital plane at 53' contains 7028 satellites (out of 40k total, but let's take one orbital plane)

basic math, "satellite density" or "average distance between satellites" if they were just at random position is ~200km (napkin math, might be off).

Re: Companies Pitch Shortwave Radio to Shave Milliseconds Off Trades

#77
post #46

Earlier quoted context omitted.

What's in Aurora? And why West Chicago in the original article? Amusingly, I grew up 2 towns over from Aurora and live 3 towns from Carteret now.

The CME datacenter is in Aurora, and the ICE datacenter is in Chicago proper.

Ah, cool. Is there a significant operation in West Chicago, too? From the article, I was still confused by why a shortwave tower would be located there. West Chicago is close to Aurora, I suppose, but if it's for CME data center, why not put the tower in Aurora?

Re: Companies Pitch Shortwave Radio to Shave Milliseconds Off Trades

#78

Earlier quoted context omitted.

I am not saying I would want to draft Goldman Sachs programmers into digging ditches. I have worked with ECU development and there are low hanging fruit to optimize for there if the effort, brainpower and money was put into it. I just wish that more effort was put into "low margin" classical industrial design than artificial top of the pyramids' algorithms, like banks and ads.

>I just wish that more effort was put into "low margin" classical industrial design than artificial top of the pyramids' algorithms, like banks and ads. Much of the development of high-performance network switches has been driven by HFT. I doubt Metamako would have been created if HFT wasn't around as a customer.

I feel like that is saying that "high-speed cameras would never have been developed if we hadn't needed them in surveillance planes to plan for bombing runs". Seems like a lack of creativity in coming up with alternate, "better" use-cases that would still create a demand for the same technology. High speed cameras are useful for, say, factory automation and crash-testing. High-performance network switches could be used in robotic tele-surgery and protein-folding supercomputers.

Re: Companies Pitch Shortwave Radio to Shave Milliseconds Off Trades

#79
post #76

Earlier quoted context omitted.

Sure, but it seems to me like the Starlink path will still be longer in almost all cases unless you're incredibly lucky and the satellites line up exactly along the path for a few seconds. And Starlink could have additional delays related to routing and possibly TDMA. What kind of latency issues need to be considered with shortwave?

a single orbital plane at 53' contains 7028 satellites (out of 40k total, but let's take one orbital plane) basic math, "satellite density" or "average distance between satellites" if they were just at random position is ~200km (napkin math, might be off).

Some folks wrote a paper based on the proposed constellation: https://dl.acm.org/doi/10.1145/3286062.3286075

The paper actually talks about trading: > Already there are new private microwave relay links between New York and Chicago[11], London and Frankfurt[1], and London and Paris. These links have relatively low capacity compared to fiber, but are of high enough value to the finance industry to be worth building new low latency links

In their paper, they say NYC / LON RTT will be between 58 and 66 ms -- this isn't much better than fiber. On the other hand, London to Joburg -- starlink has an advantage.

Re: Companies Pitch Shortwave Radio to Shave Milliseconds Off Trades

#80
post #79
post #76

Earlier quoted context omitted.

a single orbital plane at 53' contains 7028 satellites (out of 40k total, but let's take one orbital plane) basic math, "satellite density" or "average distance between satellites" if they were just at random position is ~200km (napkin math, might be off).

Some folks wrote a paper based on the proposed constellation: https://dl.acm.org/doi/10.1145/3286062.3286075 The paper actually talks about trading: > Already there are new private microwave relay links between New York and Chicago[11], London and Frankfurt[1], and London and Paris. These links have relatively low capacity compared to fiber, but are of high enough value to the finance industry to be worth building ne…

That paper is based on gen1 constellation, and predates gen2 (350km) announcement.
Post reply on HN