On Latency, Local Vnet issues, HFT, Shape of Trading

VNET and other carriers lagging behind as others move to 2,5, and even 10 leaves Erie vulnerable and guarantees no securities firm can emerge from our current infrastructure.

We need more fiber, faster. Local government - and to be fair, as I have it on good authority - some are working on this, and kudos to them. However, members of local government and other related parties need to do whatever they can to expedite the fiber build-out and make the process as easy as possible for Vnet to have expanded successful expansion. As for VNET, they shouldn’t delay waiting for the best terms, this is a mission-critical priority for our area, and its time has come. You can call it the fiber of Erie, but with so few able to access fiber, let alone business class fiber speeds beyond 1Gbps, that rings somewhat hollow. Vnet should put the pedal to the floor in rolling it out and let us (citizens) know if there are things we can do to pressure the local government into action. Alternatively, we could use our wallets to remove roadblocks to make this work. It’s hard to overstate how big of a deal having lit fiber is for a community, especially one as challenged and shrinking as ours. Some top-tier cities have quasi-fiber or no fiber at all. However, they do boast a “Multi-Gig” Coax, which is not, I mean, not even close to the same as multi-gig fiber. When I first learned of Vnet’s plans some years ago, I got excited as it was one of the first signs that something could happen here with technology and innovation.

Businesses that rely on ultra-low latency and bandwidth will flock to the area when fiber comes. Then, those already here will improve their performance due to a faster network and excel ahead. Finally, new and unknown players will quickly emerge with something potentially out of the left field that none of us see coming. That breakthrough could rewrite the book for Erie. Brilliant technologists and college students with computer science and engineering training are hyper-aware of where the best fiber connections can be found. They aggressively hunt out low-latency systems. Well, let’s put them on notice. We have it and support their mission to make something incredible with it.

This also extends to finance. When you trade stocks, your orders are bundled, transmitted, and compete with countless other market orders when they are submitted to the exchange to complete a buy or sale. Still, your order is placed in 45 out of 81959 other orders. That order is primarily set by latency. If breaking news hits the wires and you want to buy 100 shares of the SPY500 at 414.35, and you think you have it nailed, knowing that the information that just broke will drive the SPY up another 50 basis points. However, in your excitement, you missed the message on your trading platform that your order could not be executed. The security price moved beyond the range established by your trading order. Why? Because amply capitalized institutions and well-funded professional traders lease space in the same data centers as the market centers themselves, and as a result,  their orders reach the market-making servers before any other order, especially one coming from far away with lousy latency and poor connection.

Think of it this way: a trader in Sioux Falls sees mispricing in the stock of Apple – every time the stock has dropped 5 points, rallied back, fell another 5, and then rallied back again, a 25-point drop quickly followed it. So, you have a trading thesis in place. Let’s say another trader living in New Jersey sees this pattern who lives a block away from the actual NYSE data center where trading happens (yes, the NYSE is in NJ). He will immediately fire off an order to sell Apple at the top of the rally taking as many shares as possible going short. Our friend in Sioux Falls might place a trade right around where the price was when the trader in New Jersey placed his order. However, because of his proximity to the NJ Market Center, the NJ trader got filled and reaped the reward, while the Sioux Falls trader missed out on the trade. By the time the data trickled all the way to his platform, the move was already over. This result was solely due to the logistics of information and speed. 

Fiber might not have completely fixed this issue, but it would have helped make it a more even fight. Let’s say everyone has fiber. Your information will be more up-to-date if you are on the server where the good, or stock in our example, is being bought or sold. But Fiber makes it a much fairer fight. I can tell you, as someone conducting business in Erie in the capital markets, latency DOES impact me, it DOES hold me back, it HAS cost me, and with regards to the large datasets involved in my work, the bandwidth available to me via Spectrum creates a severe bottleneck. I want to stay in Erie, but I don’t know how long I can be on equal footing with my global peers without technology. That’s a very sad state of affairs, as I prefer the opposite scenario: I want to start a capital markets firm here in Erie and encourage traders to relocate here to make good money, live in a laid-back environment, and have an incredibly low cost of living relative to where they’ve been. Still, none of this will be possible without something as simple as reliable multi-gig fiber. Other issues might prevent my dream of trying to turn Erie into a capital market hub from coming true, but this is the one that has my attention for now.

Widely unknown by those outside the finance industry, the NYSE no longer operates on the “NYSE.” The NYSE of Wall Street fame is nothing more than a showroom for CNBC. Very little actual trading takes place down there. It’s pretty sad. I once upon a time interned there, there was still a few trading pits, but this was in ‘07, and those years were really the last days of a proper stock exchange complete with trading pits and shouting brokers. The rowdy pits of the Exchange symbolized a once-booming era, which is now essentially over. A new symbol of the following generation can be seen in the iPad and other high-tech devices carried around the exchange floor - as if the last of the remaining market makers needed a reminder that a $700 Apple device ultimately cost them their job. That job, in most cases, was likely inherited and handed down from grandfather to father to son.

Just one last passing shot at making clear that we’re in a new era of digital engagement in the financial markets: the current digital generation has taken the reigns. It runs the show while also enjoying a wide-open horizon with no threats to speak of. Of course, that’s not entirely true, as one major danger lurks beyond the bounds of current financial practices. We will revisit this once the first or second revolutionary AI systems emerge. Leaders lead until they don’t - feeling confident today and believing you’ve reached the top of the atlas is a dangerous attitude. AI has a lot in store for the financial markets, and how many laypeople can fully participate at the high end? Regardless of its direction, increasingly, large banks are colluding to drive well-funded trading systems to prey on market centers. The power of HFT is incredible, and I shudder to think about what will be possible once these large AI language models tap into financial markets and start training on that data.

All I can pray for is a gentle reign filled with fortune for all and a low entry barrier for those wanting to get involved. But before we go all AI, it’s worth acknowledging the change that’s taken place already. A once-significant era is no more; a new era of digital engagement has surpassed the highest peaks of the previous boom. For those looking, there are many signs of the emergence of a new digital age: electronic trading, the new frontier. As it happens, most US stock market activity occurs in New Jersey, and all the major market venues are stationed in a roughly straight line headed south down New Jersey. You could easily take a car ride up the coast of New Jersey and drive by each one. This information would likely surprise many (including those in New Jersey), but how many people knew we had over 15 stock market exchange centers? With all that said, despite the proliferation of electronic trading, the reality is that if you want to get close to the beating heart of wall street and finance, lower Manhattan is still the place to go. Exchange Place is a sight to behold, and for this writer, the NYSE evokes a sense of patriotism, pride, and optimism about the future of this country. Once you see the REAL NYSE, a drive down the Jersey Pike has more context. That ride will take you by all the “real” stock exchanges or, maybe more accurately, market centers. Still, you would never know it from looking at the buildings - windowless, foreboding buildings with barbed wire, spotlights, and enough armed guards to put down a small revolution.

The destinations for market orders are the “trading venues.” The venue is nothing more than a server on a rack and many other servers working in tandem, processing vast amounts of data and routing trading orders from every direction possible, finding and matching buyers and sellers worldwide in real-time while maintaining an accurate pricing history and quotation function. In short, trading venues exist on servers and their network equipment racks. If I’m an intelligent quantitative trader and created an HFT (high-frequency trading, a style of investment trading where orders are executed lightning-fast for tiny gains of a few pennies, done by knowing what orders are being sent to the stock market exchange where trading is taking place and, by using their high-speed technology, can get an order in to buy or sell the stock before you, forcing you to pay a premium to buy or sell the stock from the HFT trader. This strategy is highly competitive amongst the best trading firms on Wall Street, very expensive to execute, and highly profitable) or another algorithmic strategy that I thought was going to make huge profits, I would absolutely pursue seeking financing from wealthy friends and savvy lenders familiar with the practice to raise funds, funds which I would then use to secure myself a well placed, low latency server, close to the market center where my product of choice trades, which issues the buy and sell orders based on the signals my algorithms generate and execute in the very same datacenter, likely just a few racks over. The only problem is that the cost of leasing a server in one of the data centers where market centers are held easily runs into the very high thousands to tens of thousands. This should not be a surprise when you understand how much money can be made with a trading system with a winning strategy and next to zero latency. With the right software running on your server, you effectively have a right to print money. Going further, some market centers credit you for “providing liquidity” on certain exchanges by buying and selling securities at the “bid” or the “offer,” a process that can be executed in such a way that the market participant barely takes on any investment risk and solely captures the market center rebate for providing liquidity. So theoretically, you could have a program that constantly hunts for stock listings with low current liquidity. By routing an “offer” to that exchange, the trader is issued a rebate for “providing liquidity,” even if the trade order had no economic value other than simply filling another trader’s order. This filled order can be immediately liquidated, neutralizing any market direction risk and isolating the rebate as the profit driver. Certain groups run programs around the clock solely to capture rebates. Sadly, from what I understand, the practice can be pretty profitable. 

My type of trading is straightforward. Once I begin, my orders go to the servers. They are instantly matched with a corresponding buy or sell order, resulting in trade execution. Now, all that’s left is closing the trade. Assuming the market rallies .50%, selling the market and pocketing the difference would be reasonable. That, in short, is the essence of an outright trade. However, much more happens after this point when it comes to trading, especially for institutions, but the issue for this example is that the trade itself is over at this juncture. You can leave it to your broker and clearinghouse partners to ensure the money from the trade is in the right place and that any newly owned stock finds its way to your custody. That’s a job for your broker and the clearing agent.

Regarding latency, it doesn’t get much better than integrating your trading server with the network, real estate, equipment, and data pipes that the market exchanges you are trying to trade on use. I am not exaggerating when I say inches and milimeeters mean life or death, or 1ms latency vs. 3ms latency means one survives and the other is toast. Businesses maintain and operate in a fiercely competitive environment where they constantly push the envelope on the time constraints and mechanics of clearing and trading processes for the markets so that they can push their orders to market much faster, sometimes in the 1-2ms range. Even more perversely, due to the high-frequency trading, the nature of some of these players in the market, and the geographic discernment of exchanges across servers around the country, lead to sophisticated HFT traders seeing your order coming down the market pipeline as your order hits market center aftermarket center and that HFT trader bids the price up a touch by buying some shares of the stock you are trying to buy, knowing full well your order is coming, which in turn forces you increase your bid, and of course, when you do, the HFT trader liquidates the trade in a microsecond, capturing the value remaining between the price of the security paid before your order arrives and the new prevailing market quote after your order executes and drives the stock up a tick or more. HFT traders are totally content with capturing a tick to a few ticks at most (the smallest movement a stock can make is a tick, or more simply, it’s a print of a recorded trade on the market, a minimum of $.01). With reasonable capitalization, for the HFT trader/system, that tick could be worth anywhere from tens of dollars to far more. But what supercharges them is that they don’t take bathroom breaks. They can do it all day long throughout the trading hours, and as its all algorithmic, a computer identifies incoming orders, trades in front, allows the incoming order to lift the stock (after it bought shares for slightly less before the order hit the exchange), and liquidates the position, all in the matter of mere seconds, at most.

You make these small incremental trades enough times a day across enough markets, and you will start to make serious money. All that is required is incredibly low latency and some good programming. The odds are that an HFT trader is likely an algorithm, not a real trader. However, it’s very plausible that a trader overlooks and manages the algo, but the actual thing doing the trading is likely computer generated. In a given scenario, say that algo pulled that stunt mentioned above 10000 times daily. He bought a stock before another person’s order could reach the exchange and then sold it immediately after the order hit the exchange and lifted the stock price by a penny or two. It might seem like a little nuisance, someone sniping a penny or two from your gains, possibly just a cost of doing business. Well, if he made that trade 10000 times a day, with each HFT trade capturing two ticks (two price second by second price movements), or $.02 per share, with a total of 500 shares per trade, our HFT trader is grossing $100,000 a day. With roughly 252 trading days a year, our HFT trader would have yearly profits of $25,200,000. And that’s just the start!

HFT traders are well funded, so they could jump in front of the line coming in, given their latency advantage and ability to see order flow coming from further out exchanges, and slide an order in, say 500 shares at AAPL, to make a few pennies. In this scenario, Apple is at $155.55, and an HFT sees an order coming for $155.58; using HFT, they would quickly buy Apple at $155.56, and knowing your order is coming, sell it to you at $155.58, capturing 2 cents on the trade. That occurred in a few seconds, so 2 cents might not seem small. Still, when you realize the whole trading day is 6.5 hours, that means there are 23,400 seconds that the HFT can operate in, leaving more than enough time to reach the 10000 trades from above, especially when you remember that over 6,000 stocks are trading at any given time, each with their own 23,400 seconds of opportunity. For the entire day, this quiet HFT operator could be snipping a penny or two every other tick for the day. Just some quick math, let's say that our HFT trader made up .002% of Apple's trading volume (66,485,773). At .02 per trade, with 132,971.546 shares traded, that would leave him with 265 opportunities to scalp 500 shares for .02 cents. With that small percentage of total volume, this HFT system is still generating $2,650 a day in microsecond transactions. Not the end of the world, but I can assure you that the scale and volume of HFT is far more than .002%.

The only way we can compete and keep this market honest and fair is to ensure all market participants work from a dedicated fiber optic line with very low latency and substantial bandwidth. So you can see the opportunity for this type of parasitic trading. One thousand micro trades a day seems very small, but if our algorithm or trader trades around 1000 times daily, each trade profiting $300 dollars, then our modest HFT trader makes about $75,600,000 a year. Realistically, HFT systems trade more than 1000 times daily, and most HFT traders are concentrated at the same firms. Firms that can afford astronomical networking costs. Firms that could afford to lay fiber lines from New York to Chicago on their own dime to engage in this type of trading. Now you can understand why firms like Citadel and others have only had 4 losing trading days in the last year.

Think of it this way: you are trading in an open market. Still, the guy with the fastest ability to connect to all market participants at the exchange sees your trade order, whether buy to open or sell to close, and can choose to front run your order to buy or sell, putting you instantly in the red. This is, at its core, high-frequency trading. Sometimes even seeing the market aggressively taking the other side of a trade before you know what’s happening is a sign of HFT, leaving you to fight for scraps after the move has primarily been made. Those with low latency and high bandwidth are out celebrating. At the same time, you’re still stuck at work trying to figure out how the perfect trade fell apart. Could it be the technology or, more specifically, the speed of my network? Yes. It absolutely can.

My apologies: I went too far into HFT and Latency and the impacts on trading. Still, entrepreneurs and businesspeople in Erie need access to this beautiful technology. Whatever is in the way must be pushed aside; the people want it. Vnet should want it. I can only imagine the revenue of having a fiber monopoly on a city of 100,000 people and a metro area of far more. The local government likes it (it provides an excellent service with the added benefit of being something to brag about for the region, after all), so I’m at a loss regarding the hold-up. This should be considered a top priority. I’m very familiar with neighborhoods that have offered to pay for the installation costs of laying & installing the fiber to get access to it. So, a great product, check. A business with a profit motive, check. A firm should rationally want to expand and increase profit, check. Customers willing to subsidize the cost of building out fiber to their homes, for which they will pay hefty fees for many months, check. That’s many checks. So, with all that, I’m trying to figure out what I’m missing. I would love to know what Vnet is thinking. To be fair to them, I have not seen anything written. But if customers are begging to pay for installation costs of the very fiber they will be charged heavily for.. then I’m sorry, failure to roll it out rapidly is a business failure. Given all the above, I can’t think of a good reason for a hold-up.

How would you handle the following circumstances if you sold custom cars and had engineers? I was willing to bring tires, a chassis, an engine, an exhaust, everything else that comes with building a car, wages for the labor it would take to assemble it, and their benefits. Would you push this order off and ignore it? Or would you clear your schedule and say, “You know what? We can get this going very quickly. Will you be ready for pickup in three days?”

I want to ask for transparency. I have been requesting a timeline on when my home would be eligible for fiber for years, with a vague promise that it’s coming soon. Well, it’s been soon, so I want to see where we stand, how close Vnet is to expanding its footprint, and when a realistic timeframe is to anticipate having Vnet offerings. Is any more information needed from a neighborhood to show broad buy-in to the offering? What type of capital commitments is Vnet looking for? Long story short, I need that low latency. Yesterday. And I’m happy to do whatever I can to finish this quickly. As mentioned, it is sincerely desired, wanted, and necessary for many business models, including for this writer. Show me where the delay is or what is holding things up, and I will focus all my energy on helping remedy the delay. Unless a bank collapses, I must return to work, but I’d like to make some progress between now and the next bank collapse.