Starlink : Markets and Marketing

[TorenAltair]

Not sure if it really fits in here. Please move it in case @mods

Germany talks about subsidising rural Internet access via satellite and other technologies. So talks are happening with Starlink about a 500 Euro per customer voucher.

both in German
https://www.computerbase.de/2021-06/starlink-und-co.-bund-foerdert-internet-per-satellit-und-richtfunk/
paywalled: https://www.handelsblatt.com/technik/it-internet/spacex-tochter-gutscheine-fuer-elon-musks-starlink-wie-der-bund-jetzt-endlich-schnelles-internet-aufs-land-bringen-will/27235686.html

[Ludus]

Might be worth just leaving them out of the service entirely. South Africa actually has pretty decent internet in urban areas, and the people in rural areas generally don’t have the money for a $100/month service anyway.
So no big loss from a revenue point of view.

That's no more true than the last 50 times someone said it. A Starlink terminal doesn't have to be for one person. A $40 wifi 9db omni could easily cover a small village.
 And not everyone bases their lives on maximum revenue.

My speculation is Starlink’s most serious revenue stream in the next few years comes from using a small but high priority fraction of the system for ultra low latency financial data between distant financial capital cities. If so they can prioritize pro Bono subsidized terminals connecting rural villages all over the world to balance huge profits in financial data and analytics with doing something important. Acting as an ISP so middle class people in the countryside can use zoom is a sort of mildly profitable middle ground.

[daavery]

My speculation is Starlink’s most serious revenue stream in the next few years comes from using a small but high priority fraction of the system for ultra low latency financial data between distant financial capital cities. If so they can prioritize pro Bono subsidized terminals connecting rural villages all over the world to balance huge profits in financial data and analytics with doing something important. Acting as an ISP so middle class people in the countryside can use zoom is a sort of mildly profitable middle ground.

starlink should be slower that dedicated fiber network WANS in almost every conceivable case. financial traders (institutional) already have dedicated fiber links in place

[launchwatcher]

My speculation is Starlink’s most serious revenue stream in the next few years comes from using a small but high priority fraction of the system for ultra low latency financial data between distant financial capital cities. If so they can prioritize pro Bono subsidized terminals connecting rural villages all over the world to balance huge profits in financial data and analytics with doing something important. Acting as an ISP so middle class people in the countryside can use zoom is a sort of mildly profitable middle ground.

starlink should be slower that dedicated fiber network WANS in almost every conceivable case. financial traders (institutional) already have dedicated fiber links in place

The speed of light in optical fiber is about 70% of the speed of light in air or vacuum.

Fibers are going to be higher bandwidth than starlink, but also higher latency.   

High-frequency traders care about latency, and have been building microwave relays because they're lower latency than fiber.

[daavery]

starlinks adds at least 1000kM plus any internal latency

[Robotbeat]

starlinks adds at least 1000kM plus any internal latency

That's not how the Pathagorean Theorem works. Over longer distances, the added distance can be less than 1000km. Secondarily, fiber often doesn't go in a strait line.

But agree the whole "Starlink is for HFT" thing is wrong.

[envy887]

starlinks adds at least 1000kM plus any internal latency

That's not how the Pathagorean Theorem works. Over longer distances, the added distance can be less than 1000km. Secondarily, fiber often doesn't go in a strait line.

But agree the whole "Starlink is for HFT" thing is wrong.

Is transoceanic HFT a thing? It's hard to build a surface microwave relay across the Atlantic, and Starlink can easily beat undersea fiber over that distance, especially once they have ISLs but even before that.

[gongora]

Starlink can easily beat undersea fiber over that distance, especially once they have ISLs but even before that.

uh, pretty sure before ISL's the Starlink data will be going over the undersea fiber.  Even with ISL's it's not going to be a free for all where any satellite can instantly transmit to any other satellite.

[freddo411]

starlinks adds at least 1000kM plus any internal latency

That's not how the Pathagorean Theorem works. Over longer distances, the added distance can be less than 1000km. Secondarily, fiber often doesn't go in a strait line.

But agree the whole "Starlink is for HFT" thing is wrong.

Why do you think "Starlink is for HFT" thing is wrong?   

Any system that allows for communication between two trading points that is faster than other traders can be (and I claim will be) used as an arbitrage advantage.    The physics of the starlink system shows that this *can* be true if SpaceX sells this service to some exclusive HFT customers.   SpaceX may or may not choose to sell specific HFT-useful service.   

I would agree that Starlink serves many other purposes and many other paying customers.   Not selling starlink for use in HFT is not a big deal.

* HFT = high frequency trading

[RAN]

Why do you think "Starlink is for HFT" thing is wrong?   

Any system that allows for communication between two trading points that is faster than other traders can be (and I claim will be) used as an arbitrage advantage.    The physics of the starlink system shows that this *can* be true if SpaceX sells this service to some exclusive HFT customers.   SpaceX may or may not choose to sell specific HFT-useful service.   

I would agree that Starlink serves many other purposes and many other paying customers.   Not selling starlink for use in HFT is not a big deal.

* HFT = high frequency trading

which HFT customers would benefit from starlink? I am under the impression that most high frequency traders colocate with the exchanges, ie. servers a few blocks from the New York Stock Exchange

While there is no single definition of HFT, among its key attributes are highly sophisticated algorithms, co-location, and very short-term investment horizons.

[freddo411]

Why do you think "Starlink is for HFT" thing is wrong?   

Any system that allows for communication between two trading points that is faster than other traders can be (and I claim will be) used as an arbitrage advantage.    The physics of the starlink system shows that this *can* be true if SpaceX sells this service to some exclusive HFT customers.   SpaceX may or may not choose to sell specific HFT-useful service.   

I would agree that Starlink serves many other purposes and many other paying customers.   Not selling starlink for use in HFT is not a big deal.

* HFT = high frequency trading

which HFT customers would benefit from starlink? I am under the impression that most high frequency traders colocate with the exchanges, ie. servers a few blocks from the New York Stock Exchange

While there is no single definition of HFT, among its key attributes are highly sophisticated algorithms, co-location, and very short-term investment horizons.

The key bit to understand this is that any mechanism that allows a trader to know information faster than other traders provides an advantage.   Starlink can provide information from NYC to London a few microseconds faster than other communication pathways.

Here is one example on how HFT can make extra fast communication profitable:   If it is known that the price for $X has risen in NYC, then traders in London can buy $X for the cheaper London price, and sell it in NYC for the higher price there.

[Robotbeat]

starlinks adds at least 1000kM plus any internal latency

That's not how the Pathagorean Theorem works. Over longer distances, the added distance can be less than 1000km. Secondarily, fiber often doesn't go in a strait line.

But agree the whole "Starlink is for HFT" thing is wrong.

Why do you think "Starlink is for HFT" thing is wrong?   

Any system that allows for communication between two trading points that is faster than other traders can be (and I claim will be) used as an arbitrage advantage.    The physics of the starlink system shows that this *can* be true if SpaceX sells this service to some exclusive HFT customers.   SpaceX may or may not choose to sell specific HFT-useful service.   

I would agree that Starlink serves many other purposes and many other paying customers.   Not selling starlink for use in HFT is not a big deal.

* HFT = high frequency trading

Because it's outcompeted by microwave towers and shortwave radio.

[AC in NC]

Here is one example on how HFT can make extra fast communication profitable:   If it is known that the price for $X has risen in NYC, then traders in London can buy $X for the cheaper London price, and sell it in NYC for the higher price there.

This isn't really an example of what HFT is all about.  Collocation as said above is the key.

[envy887]

Starlink can easily beat undersea fiber over that distance, especially once they have ISLs but even before that.

uh, pretty sure before ISL's the Starlink data will be going over the undersea fiber.  Even with ISL's it's not going to be a free for all where any satellite can instantly transmit to any other satellite.

They could cross the Atlantic with 1 (or maybe a few, for higher reliability) ship-mounted relay station, plus ground stations in St. Johns and western Ireland, and get from New York to anywhere in western Europe in 4 hops. This will probably beat fiber by a pretty good amount, since it's essentially a microwave relay with really, really tall towers.

I kind of doubt the network is set up to have a ground station push traffic form sat to sat, but I don't see any reason they couldn't add that if there was a market

This video has been posted before, but it's interesting re: relays...

[freddo411]

Here is one example on how HFT can make extra fast communication profitable:   If it is known that the price for $X has risen in NYC, then traders in London can buy $X for the cheaper London price, and sell it in NYC for the higher price there.

This isn't really an example of what HFT is all about.  Collocation as said above is the key.

Ask why co-location is key.   So the traders have access to stock price information a tiny fraction of a sec before others.

The same situation would exist between a trader in London monitoring prices in NYC using starlink, vs a trader in London using a fiber connection.

The key is faster access to market information;  not colocation per se.

[Redclaws]

starlinks adds at least 1000kM plus any internal latency

That's not how the Pathagorean Theorem works. Over longer distances, the added distance can be less than 1000km. Secondarily, fiber often doesn't go in a strait line.

But agree the whole "Starlink is for HFT" thing is wrong.

Why do you think "Starlink is for HFT" thing is wrong?   

Any system that allows for communication between two trading points that is faster than other traders can be (and I claim will be) used as an arbitrage advantage.    The physics of the starlink system shows that this *can* be true if SpaceX sells this service to some exclusive HFT customers.   SpaceX may or may not choose to sell specific HFT-useful service.   

I would agree that Starlink serves many other purposes and many other paying customers.   Not selling starlink for use in HFT is not a big deal.

* HFT = high frequency trading

Because it's outcompeted by microwave towers and shortwave radio.

Even across oceans?  Is short wave used today for HFT where there’s no land mass, then?

[freddo411]

starlinks adds at least 1000kM plus any internal latency

That's not how the Pathagorean Theorem works. Over longer distances, the added distance can be less than 1000km. Secondarily, fiber often doesn't go in a strait line.

But agree the whole "Starlink is for HFT" thing is wrong.

Why do you think "Starlink is for HFT" thing is wrong?   

Any system that allows for communication between two trading points that is faster than other traders can be (and I claim will be) used as an arbitrage advantage.    The physics of the starlink system shows that this *can* be true if SpaceX sells this service to some exclusive HFT customers.   SpaceX may or may not choose to sell specific HFT-useful service.   

I would agree that Starlink serves many other purposes and many other paying customers.   Not selling starlink for use in HFT is not a big deal.

* HFT = high frequency trading

Because it's outcompeted by microwave towers and shortwave radio.

That is an interesting premise.

[Scintillant]

starlinks adds at least 1000kM plus any internal latency

That's not how the Pathagorean Theorem works. Over longer distances, the added distance can be less than 1000km. Secondarily, fiber often doesn't go in a strait line.

But agree the whole "Starlink is for HFT" thing is wrong.

Why do you think "Starlink is for HFT" thing is wrong?   

Any system that allows for communication between two trading points that is faster than other traders can be (and I claim will be) used as an arbitrage advantage.    The physics of the starlink system shows that this *can* be true if SpaceX sells this service to some exclusive HFT customers.   SpaceX may or may not choose to sell specific HFT-useful service.   

I would agree that Starlink serves many other purposes and many other paying customers.   Not selling starlink for use in HFT is not a big deal.

* HFT = high frequency trading

Because it's outcompeted by microwave towers and shortwave radio.

Even across oceans?  Is short wave used today for HFT where there’s no land mass, then?

Apparently yes. The article below shows that satellite systems can at least achieve parity with shortwave, so Robotbeat is wrong here.

https://www.wsj.com/articles/high-frequency-traders-eye-satellites-for-ultimate-speed-boost-11617289217

Using subsea fiber, it takes about 33.5 milliseconds to send data from Chicago to London, using the fastest available links across the eastern U.S. and under the Atlantic. With shortwave, one provider, Raft Technologies, says it can radio data from Chicago to London in 29.6 milliseconds.

With satellites, it’s harder to put a precise number on latencies. That’s because the satellites in a network are constantly moving, so the length of the shortest path between two points varies over time. Moreover, many satellite ventures haven’t released data on the latencies they expect to achieve across specific routes. One that did—LeoSat—told prospective clients in a technical-specifications document seen by The Wall Street Journal that, around 70% of the time, it could send data from Chicago to London in 29 milliseconds or less.

[vsatman]

My speculation is Starlink’s most serious revenue stream in the next few years comes from using a small but high priority fraction of the system for ultra low latency financial data between distant financial capital cities. If so they can prioritize pro Bono subsidized terminals connecting rural villages all over the world to balance huge profits in financial data and analytics with doing something important. Acting as an ISP so middle class people in the countryside can use zoom is a sort of mildly profitable middle ground.

No!. The reason is very simple, when using inter-satellite links, the distance between satellites will constantly change and the total distance of the route will also change, that is, the channel will have different delay and different jitter all the time. This service is not needed for HFT

[Jarnis]

When inter-satellite laser links are a thing, we will see.

I predict with some confidence that if Starlink can provide consistently <29ms (even with variance, but with <29ms say > 80% of the time) link between the East Coast and London, there will be customers wanting to use that for the purposes of moving market data. And this being a relatively short hop over the Atlantic, if this one works, there will be other markets (Australia, Hong Kong, Tokyo) where ther will be specific customers who will pay a premium if they can get data from New York and Chicago with less latency than today's solutions.

And yes, HFT traders trading on New York or Chicago markets obviously place their hardware there as close to the exchanges as possible and these days microwave links are often the fiber-beaters over shorter distances like from Chicago to New York. This is about gaining information advantage over other traders in foreign exchanges further away from US East Coast which still is the "heart" of the international markets in a way. If you trade in Hong Kong and know about a price move in New York 10ms before all your competitiors, you can easily make money off that. Same can also sometimes be true the other way around - a price move in a foreign market can sometimes be turned into money trading in the US market, tho most of the time it is the US markets that "drive" prices worldwide.

But I don't see this being much of a thing prior to laser interlinks and it will be a niche (but potentially a very profitable niche) for very few customers. Not really the bread-and-butter of Starlink which still is getting non-terrible internet access to places that currently have nothing anyone could call "broadband" available (in my books, 100mbit but this varies from market to market). That is also why they can easily (always) ask for more than cable prices - if you can get cable, you are not the target market. If one day they run into a problem where they have spare capacity even after capturing huge chunk of the market outside the areas served by fiber, cable etc. they can then consider offering discounts in specific areas to compete with cable if they want. I'm not seeing this being the case ever, but we'll see.