SpaceX Starlink Will Not Be for High-Frequency Trading

The SpaceX Starlink satellite network will not for high-frequency trading. It is for the productivity of tens of thousands of finance people in each finance city pair. Latency makes each communication people and machines, machines and machines slower. It is a million tiny cuts per person. Each finance person costing an average of about $200K to $1+ million including overhead.

However, the money lost is from getting in orders behind other traders during market rushes. It is not time waiting and salary lost. It is getting the information a bit later, making a decision and then being at a different spot in the order execution line. You are reacting to a stock moving rapidly higher or rapidly lower. Where are you in line to buy and sell? The Black Friday checkout line at Walmart starts filling up and did you get there ten seconds earlier or ten seconds later. Do you get door-buster pricing or do you not?

The high-frequency trading is taking no market risk. They are shaving the bid and offer gaps. They are taking some money every second by capitalizing on inefficiency in the market.

The speed of light is faster in space than it is through a fiber cable. It is nearly twice as fast. Cities more than 2000 miles away will be able to communicate faster. This is not bandwidth speed but latency or the initial lag at the beginning of each new transmission or data message.

A 5% improvement in lowered delays was worth more than $100 million per between New York and Chicago. New York has major stock exchanges and Chicago has commodities trading.

Latency is like a smaller version the gains going from dial-up modem to ISDN line gains. The 10-second boot-up at the start of the dial-up connection was not latency per se but it was an obvious delay penalty.

Once the data is moving then bandwidth determines the speed.

A dial-up connection had a latency of around 90-110 millisecond. An average ADSL connection is around 30-50 milliseconds. However, those are latency tests to the nearest servers. There is also passing information around the network.

High-frequency trading is nanosecond stuff. They are co-located with the exchanges.

The way to get rid of high-frequency trading is to collect all trades and settle once per second or once per minute. There is then almost no advantage for being nanoseconds faster.

The SEC and the Exchanges need to level the playing with top of the second settlement. This shift would not be that difficult. It is not happening. Certain people and companies keep making billions. It is a corruption tax on the investment returns of everyone else who is not gaining from high-frequency trading.

18 thoughts on “SpaceX Starlink Will Not Be for High-Frequency Trading”

  1. Starlink is not high-frequency trading. Those are nanoseconds and microseconds. The uplinks to low earth orbit have too many delays for that. I tried to explain it is about share movements over hundredth and tenths of second.

    Tax on trading. Not relevant if it did happen. And will not happen because this does not realistically reflect politics and money. Any laws will be ineffective. Likely designed to not work.

    Phased array on roof shingles. This is overly complicated. Too expensive. Will not happen. I could parse it out but it would be boring.

  2. Problem is that the satellite would have to be set up for this as in having an dedicated link for this, bandwidth is not the issue this is just signals not even voice.

    It depend a lot on how the routing and down linking work on starlight.
    I suspect routing is pretty slow as its very dynamic, you will have to switch between satellites as they move and you sometimes have to send signal between two satellites depending on orbits while other cases one satellite could do both.

  3. Each satellite covers an area of about 1300 miles. The distance between NY and Chicago is about 800 miles.

    The big boys would take long-term leases on huge amounts of bandwidth.

    With multiple sats always in view, there would be, I imagine, little if any any need for star to star routing of traffic.

    NY to London would require a couple of jumps.

    Not arguing for or against, just thinking about it.

  4. I wondered about how an over reliance on ultra-high speed trading might expose Starlink to a sudden cut in revenue if a talked-about tax on such trading came around.

    As far as cost of residential antennae goes, how much of a leap would it be for Musk to add a phased array feature as an option to his solar roof shingles?

    The three rules of Musk: integration, integration, integration.

  5. I don’t see where, besides HFT, there is sufficient value for customers to be willing to shell out big bucks for a low latency, space-based system. A human being simply can’t take advantage of a latency of “only” 50 milliseconds, as opposed to 350 milliseconds. It is also far from clear to me that program trading which is not HFT (which, as I see it, is what’s left) can extract enough value from low latency to justify big purchases.

  6. High frequency trading is nowhere needed, productive or smart.
    It is just scraping of some gold of the coin someone else has worked for.

  7. high frequency trading is no-where smart or producing or inventing…. its just scraping-off some gold of the coin someone else has worked for.

  8. Correct its about computer trading, and yes stock trading is pretty much an zero sum game, stock prices tend to raise overall but short time trading is zero sum.

    Its however so lucrative they are willing to buy special hardware like dedicated cables with minimal routing to reduce latency.
    Now reason this probably don’t work for starlink is that starling will involve lots of routing in space, the package will jump from satellite to satellite before getting sent to an ground station. It can easy become many jumps and some risk your package will take an detour if capacity is reached. Say you use starlink in New York, but as bandwidth is pretty capped on the east coast your signal is routed to an satellite over the mid west before sent to an ground station.

    For this kind of use you probably want higher orbits to minimize the numbers of jumps. Have it as an feature on Iridium might make more sense.

  9. We all know it’s main function will be to transmit porn. Add neuralink to the equation and online sex with feelings transmission You can’t get lag!

  10. Interesting, in the articles so far I hadn’t seen anything about the antennas. That might be worth an article of its own.

  11. It is not about getting into line it is the ability to see other orders and then to place your order first. It is also NOT about human trading it is about trading algorithms with a super fast computer and fastest line. Make no mistake these guys skim and front run orders. This money must be TAKEN from someone! It comes from the traders and YOUR retirement fund. The high frequency traders claim to add lubrication to the market but the only people needing lubrication is the retirement funds and the average traders.
    There once was a fee for each transaction. If that were brought back then the thousand trades a day would no longer make sense and a fairer market would happen.
    Shorting is where you “borrow” someones stock, leave an IOU and sell it into the market. Basically you are betting that the stock will go down. Then the shorter is tempted to make sure the stock will go down to make a profit sometimes in the order of a 100 million dollars. Lets say there are 100,000 shares in the market then someone borrows 10,000 and sells them into the market. The market now has a supply of 110,000 shares. Basic supply and demand curves say that more supply will drive down the market. If this excess supply is used as a weapon the price can be driven “spiked” down rapidly causing others to sell due to many reason including panic and “stop loss” orders. Chipotle shows how far they will go.

  12. Starlink will be for the mass market when the phased array antennas come down in price. Kymeta antenna costs $40,000 each. However it is based a lot upon something flat panel TV display technology. At least mostly the same production lines to my understanding. They should be able to get down to $200-500 each. Maybe towards $100 each. We can make 32 inch flat panel TVs at that price. I say that it should happen but it has not. There are also a dozen other companies working to crack the cheap flat antenna problem. Then the satellites can whiz by in low earth orbit and the antennas can receive, track and switch. Like pointing at different trucks passing on a highway.

    The money differential is mainly when a stock is falling say 30% in a day, but not gapping down or moving up rapidly but no gapping up. This is when the trader can see it start to move and get the order made. The 30 millisecond faster refresh of the ticker and the 30 second faster order placement would matter by say 2% or 5%.

    You are right it may not be humans. It could be computers.

    It could also be the traders are running and managing their own algorithmic trades or settings. If a hundred traders want to put in the trade when the stock hit $25. Then they putting in the order and executions could range from $25 to $28 on a fast moving stock. Where it was moving in 20 minutes from $20 to $35.

    They may not have set a stop loss order with the exchange because they wanted more dynamic control.

  13. Still think Starlink is going to be for the mass market, not just finance guys.

    If it’s a human doing the trade, I don’t think a few milliseconds makes a lot of difference. And if things are that tight, you’re going to be physically close to the order book anyway.

  14. I wanted to talk about the time value of information and the ongoing unfair playing field. Like how the Rothschilds made massive profits by knowing days in advance of everyone else that Napoleon had lost at Waterloo. This does not related to shaving 30 milliseconds from 60 milliseconds.

Comments are closed.