7 ms·

Weirdest thing I've read in a long while. The Author's comments refer to an internet forum thread[0] in German which discusses a German news article from this

by Jommi 8y ago

Weirdest thing I've read in a long while. The Author's comments refer to an internet forum thread[0] in German which discusses a German news article from this April.[1]However, it turns out this article is an April Fools text. Just digging deeper into the whole blog you can find something that can only be described as a fanfiction representation of "Flashboys", but in the context of this "Shortwave trading".[2] Is this whole thing a troll? Is this real? What on earth is this? [0]http://funkbasis.de/viewtopic.php?f=39&t=45900 http://funkbasis.de/viewtopic.php?f=39&t=45900 [1]https://www.heise.de/newsticker/meldung/Boersenhandel-beschert-Kurzwellenfunk-ein-Comeback-4008891.html https://www.heise.de/newsticker/meldung/Boersenhandel-besche... [2]https://sniperinmahwah.wordpress.com/2017/06/26/network-effects-part-ii/ https://sniperinmahwah.wordpress.com/2017/06/26/network-effe...

Jommi 8y ago
Seems like it is real: https://www.six-group.com/en/home/media/releases/2018/20180503-12H-start.html https://www.six-group.com/en/home/media/releases/2018/201805...
- dx034 8y ago
  This is microwave based. Those lines are well known by now and there are known routes within Europe (most notable Frankfurt<->London) and the US (NYC<->Chicago). However, microwave only works over short distances (FRA<->LON has multiple towers to cover the distance). Crossing the atlantic isn't possible. Shortwave could deliver a similar speedup over long distances. I'm a bit sceptical though. Don't the waves have to bounce several times to get to the target? I'm not sure how much of the ~30% potential speed up vs. fibre you still get.
  - jhpankow 8y ago
    I'd suspect the latency improvement is control over the entire path and not specifically about the propagation speed/delay. Getting your data across a long haul undersea fiber requires aggregating it with everyone else's data at a central point before it gets sent out optically. This adds latency. Even with the propagation variability of the wireless path you might be saving hundreds of nanoseconds.
    - dx034 8y ago
      I doubt that this is a major factor. The extra delay (light speed vs. fibre speed) for signals across the Atlantic is at least 10ms each way, a few hundred nanoseconds for aggregation shouldn't make a difference there.
    - bobvan13 8y ago
      There is some latency savings because the radio path is far straighter than the fiber path, even considering the ricochetting off the ionosphere and the earth. But the bulk of the latency savings comes from the fact that radio waves move at the full speed of light while photons through fiber only move at about 2/3 of that speed. So yeah, it works out to about a 10 ms savings on shortwave compared to fiber.
  - bobvan13 8y ago
    Yes, it has to bounce, but the number of times it bounces is dictated by the takeoff angle, as is the path length of each bounce relative to ground traversed. You aim for takeoff angles around 12 degrees. Since cos(12 degrees) ~= 0.98, there's not much path length added by the bouncing. So you still end up WAY ahead of fiber. I'll detail a stacked-curtain, log-periodic antenna used at another Chicago-area site in a forthcoming post.
- antb123 8y ago
  Yeah it started in Chicago. At my old shop we had a semi-exclusive for Toronto - Chicago route which is used for ETF market making. The main issue is that the network has a lot of packet loss due to weather condititions.
nielsole 8y ago
The linked article [1] appears to be built around a word-play, where "Shortwave"-trading and "High frequency"-trading are ambiguously used as "Hochfrequenz"handel
baxtr 8y ago
German speaker here. The sentiment on heise.de comments is that this was an April’s Fools joke
- ubiyubix 8y ago
  The majority of Heise forum participants are wrong then.
  - baxtr 8y ago
    Maybe
avip 8y ago
Long-time "sniper in mahwah" follower here. The author is without the slightest doubt an elite tech reporter with in-depth understanding of RF tech. He always provides an overwhelming body of evidence to back his writing.
- wglb 8y ago
  But why not then listen and figure out what frequencies are being used? That skill would be well within the "elite" category, no?
  - pbhjpbhj 8y ago
    Eavesdropping laws?
  - bobvan13 8y ago
    The license applications list a range of frequencies where operation is allowed. Indeed, many SWLers have documented such transmissions, e.g. https://www.youtube.com/watch?v=ySzpz8LWrjA https://www.youtube.com/watch?v=ySzpz8LWrjA The trouble is that when you do listen, you either hear test signals as recorded on YouTube, or you just get pseudorandom noise because it's encrypted.

by Jommi 8y ago

Jommi 8y ago

Seems like it is real: https://www.six-group.com/en/home/media/releases/2018/20180503-12H-start.html https://www.six-group.com/en/home/media/releases/2018/201805...

dx034 8y ago
This is microwave based. Those lines are well known by now and there are known routes within Europe (most notable Frankfurt<->London) and the US (NYC<->Chicago). However, microwave only works over short distances (FRA<->LON has multiple towers to cover the distance). Crossing the atlantic isn't possible. Shortwave could deliver a similar speedup over long distances. I'm a bit sceptical though. Don't the waves have to bounce several times to get to the target? I'm not sure how much of the ~30% potential speed up vs. fibre you still get.
- jhpankow 8y ago
  I'd suspect the latency improvement is control over the entire path and not specifically about the propagation speed/delay. Getting your data across a long haul undersea fiber requires aggregating it with everyone else's data at a central point before it gets sent out optically. This adds latency. Even with the propagation variability of the wireless path you might be saving hundreds of nanoseconds.
  - dx034 8y ago
    I doubt that this is a major factor. The extra delay (light speed vs. fibre speed) for signals across the Atlantic is at least 10ms each way, a few hundred nanoseconds for aggregation shouldn't make a difference there.
  - bobvan13 8y ago
    There is some latency savings because the radio path is far straighter than the fiber path, even considering the ricochetting off the ionosphere and the earth. But the bulk of the latency savings comes from the fact that radio waves move at the full speed of light while photons through fiber only move at about 2/3 of that speed. So yeah, it works out to about a 10 ms savings on shortwave compared to fiber.
- bobvan13 8y ago
  Yes, it has to bounce, but the number of times it bounces is dictated by the takeoff angle, as is the path length of each bounce relative to ground traversed. You aim for takeoff angles around 12 degrees. Since cos(12 degrees) ~= 0.98, there's not much path length added by the bouncing. So you still end up WAY ahead of fiber. I'll detail a stacked-curtain, log-periodic antenna used at another Chicago-area site in a forthcoming post.
antb123 8y ago
Yeah it started in Chicago. At my old shop we had a semi-exclusive for Toronto - Chicago route which is used for ETF market making. The main issue is that the network has a lot of packet loss due to weather condititions.

nielsole 8y ago

The linked article [1] appears to be built around a word-play, where "Shortwave"-trading and "High frequency"-trading are ambiguously used as "Hochfrequenz"handel

baxtr 8y ago

German speaker here. The sentiment on heise.de comments is that this was an April’s Fools joke

ubiyubix 8y ago
The majority of Heise forum participants are wrong then.
- baxtr 8y ago
  Maybe

avip 8y ago

Long-time "sniper in mahwah" follower here. The author is without the slightest doubt an elite tech reporter with in-depth understanding of RF tech. He always provides an overwhelming body of evidence to back his writing.

wglb 8y ago
But why not then listen and figure out what frequencies are being used? That skill would be well within the "elite" category, no?
- pbhjpbhj 8y ago
  Eavesdropping laws?
- bobvan13 8y ago
  The license applications list a range of frequencies where operation is allowed. Indeed, many SWLers have documented such transmissions, e.g. https://www.youtube.com/watch?v=ySzpz8LWrjA https://www.youtube.com/watch?v=ySzpz8LWrjA The trouble is that when you do listen, you either hear test signals as recorded on YouTube, or you just get pseudorandom noise because it's encrypted.