Reminds me of how compact motors and wireless telegraphs were combined so that small, fast, motorboats could meet cargo vessels outside New York and telegraph back their cargo to some trader on Wall Street...
Later, DHL was formed to fly bills of lading by commercial flights from the West coast ports to Hawaii ahead of the ships arriving. So by the time they arrived the cargo could be offloaded without delay.
Interesting read - I had not realized that DHL had started in the US, let alone in San Francisco. Not having dealt with them directly much, I had thought they were a European company. Not that it's the same company after being bought out.
And: land's price-inelasticity means that a land value tax (LVT; property, exclusive of improvements) must decrease its exchange value. Cook County's reaal estate tax includes improvements, and hence isn't a perfect LVT. But it approaches the concept.
Meh, they need to use gravity waves; not affected by dielectric constant. Maybe a spinning mass that is frequency modulated. I just need to patent that and be funded by some wealthy traders.
Seriously though, I’m surprised they are not using direct path, deeply buried air-line coax. Has to be TEM propogation or else group delay is too high, so rectangular waveguide won’t work. Circular may work; I need to brush up on my EM.
Eh, too much energy required for gravitational waves. But I think that particle accelerators and giant neutrino detectors have been at least half seriously considered. I think that, although the impulse pretty much travel at c and can tunnel though anything, the lenght sampling period of the impulse required to distinguish it from noise would defeat the latency advantage.
Too late (patenting that gravity idea, that is) - John D. Kraus suggested precisely that in an appendix to his magnificent book on all things antennas, aptly named 'Antennas'.
I do believe he concluded it was not the most practical of communication schemes, though. :)
(Edit - corrected name of author; I had originally called the poor guy Erwin Klaus)
My apologies; a double typo -KRAUS. (Courtesy of my SwiftKey keyboard app), first name John. (Erwin Kraus is an acquaintance of mine; no idea how that short circuit occurred.)
On the off chance you were looking for a book on antennas which gives you lots of ideas without getting too carried away with maths, Karl Rothammel's 'Antennenbuch' is probably the one to beat - he's basically nicked every antenna idea ever published and put it in a massive compendium. As an added bonus, it will teach you German if you don't know it already.
To ease your feet gently into the water, you could do a lot worse than picking up a copy of the 'ARRL antenna book' - a blend of theory and practice, giving ideas for a lot of different designs while explaining a bit of the theory behind their function.
Signals interests me in general (top to bottom— including hard and softwares), and I'm trying to get around to deeper physics and maths study around it, but there's only so much time in a day.
If you have any other recommendations, I'll gladly hear them— there's a link to my contact in my profile.
I've just spent a while looking for my old signal processing textbooks, but there was too much piled on top of them (I've moved house three times since I last worked with signal processing... :)
I remember two introductory books which were quite good - Kamen/Heck's 'Signals and Systems' - tells you a bit about the tools of the trade, laplace and fourier transforms, transfer functions, filter design &c - the book was accompanied by a number of MatLab scripts which let you visualise how manipulating properties of a filter affected the output; most useful.
Proakis & Manolakis' Digital Signal Processing was also quite good; both books assume you know your way around engineering mathematics - series, integrals mostly - but in particular Kamen/Heck I remember provoking quite a number of insights from the text alone.
Again, the ARRL may be of some assistance - the radio amateur's handbook has a couple of chapters on the basics of filters, leaning very much against the applied end of things.
I hear good things about Owen's Practical Signal Processing, too, but haven't studied it myself yet; as you say, there's only so much time in a day...
Oh, and steer clear of anything with the phrase 'non-linear' anywhere in the title or blurb. Just don't go there.
Pitch them a series of boxes full of quantum entangled cats. The "driver cat" in jersey can pick the stocks while the "tail cat" in NY executes the trades. Forget Heisenberg, such uncertainties are irrelevant when thrown darts can best the market.
Markets stay synchronized via continuous arbitrage. US equity trading centers in New York while options trading centers in Chicago. Slowing the data transmission to the Cboe data center in New Jersey slows its transmission to Chicago, and thus provides increased opportunity for HFT traders to earn money. If you value fairness, this is a bad thing.
Bids should be made only through a turn-based system, with ample time to delete any advantage (or harm) from "data transmission" delays (1 minute for example).
Does that hold in a turn base system? Whether they played at the start of the minute or its end, they all knew the results of the last turn when they started bidding.
They might have external world info though.
Then again, you can always have the trades gathered during the turn period randomised in order and applied that way.
> This is only an issue if later players have knowledge of the action other players are taking during the turn in progress
You’re assuming full endogeneity. That is never the case. Others’ orders are just one input. Other markets’ books and orders are far more important, particularly for derivatives.
I don't have any issue with HFTs that respond only to external news source, the research being done in news sentiment analysis and how to incorporate that into buys and sells might not be great but the HFT players that I take issue with are those that look at price trends and try to snatch penny advantages out of a perceived momentum, that sort of market play is meaningless.
Perceived momentum? I think your idea of HFT strategies might be pretty far off the mark.
Guessing that the market will continue to go up or down (a.k.a 'perceived momentum') is just a gambling strategy used by day traders. There are no long term successful hft operations based on predict the future strategies like that.
> Does that hold in a turn base system? Whether they played at the start of the minute or its end, they all knew the results of the last turn when they started bidding.
Not quite. In a turn based board game, sure, because literally nothing relevant happens outside the context of the game. In real life, plenty happens outside the context of a hypothetical Chicago turn. Most relevantly, New York happens.
(Actually, it’s New Jersey. Land is very expensive in NY, and data centers are fairly large, so “New York” finance mostly happens in Secaucus, NJ.)
You'd have to time it in place-reveal sequences, so all orders received between 12:59 and 1:00 would only be revealed (along with the matches between buy and sell orders) at 1:00.
Then I’ll settle over the counter, directly with my counterparties, and disintermediate your marketplace. The misallocation appears to be those looking to introduce complexity to solve a non-issue.
Realistically, you won't because of the transaction costs. Fuzzing doesn't add complexity, it sacrifices precision in favor of robustness. If your trading strategy depends on millisecond timing, it's too fragile for principal-agent applications.
That would be a massive change to the American securities markets. Enough so that it might destroy them. “You can buy or sell your securities at competing venues” is a basic tent of finance.
HFT is just a technology for trading faster. It's not really all that much different from the open outcry system it replaces; the big change is traders feel each other out using computers rather than by waggling their fingers at each other. Access to the market is probably more equitable now than it used to be, since you can give more people direct access to the markets with electronic links. With pit trading, you were limited by the physical size of the pit.
If you're going to give people electronic access to the markets, there's not really any sense in saying they aren't allowed to use whatever computer technology they see fit to trade with that access, as long as they aren't actually manipulating the market.
As far as economic value, it has different uses to different players on the market. Speculators can take advantage of increasingly fleeting arbitrage opportunities, yes. But the flip side of that is that institutional investors can use it defensively, to make it harder to find and exploit those arbitrage opportunities.
That basic game of cat-and-mouse has existed for centuries, and, here, almost 2 decades into the 21st century, it seems absurd to me to say that, of all the places people use technology to do things faster, nobody should be allowed to do it in this one particular case. Which when you get down to it, is essentially what people are suggesting when they say that high frequency trading is bad.
If it stills seem unfair to you (and that's an entirely legitimate feeling), I'd submit that it isn't that that using technology is unfair in and of itself, so much as that technology has the potential to magnify the effects of unfairness that was already baked into the financial system to begin with.
HFT means several banks are competiting for your orders very aggressively allowing a tighter spread. So, you sell a bit higher, and buy a bit lower. This is highly beneficial.
HFT allows the faster trader to see that you're trying to buy (or sell), put their order to buy (or sell) in ahead of yours, then turn around fill your buy order at a higher price (or your sell order at a lower price).
HFT allows better-connected traders to front-run you. You don't get the best price because the HFT guy extracted risk-free profits for himself at your expense.
Front running and HFT are two different things. Front running implies that you have access to the orders or other information and the ability to get your order in ahead of others. This is at worst, illegal as you're trading on non-public information (if you get an earnings report early), and at best, bad for business as front running would be a "tax" on every stock being sold on an exchange. This basically means that the exchange becomes less valuable (because you have to pay more for that exchange) and other brokers have the ability to offer "lower" prices for the same stock at the same moment in time as they can say that they're not providing advanced information. Not to say that it doesn't happen, but there is some "magic hand of the market" involved.
Front running is accomplished by HFT, but not all HFT is front running. HFT is simply the ability to execute trades quickly on new information, like earnings reports, fed meeting notes, etc. HFT back in the 17th century would simply have been having the fastest couriers and horses to carry news, now it's having the fastest software and data connections. HFT has a couple beneficial sides too, as it tightens the bid/ask spread.
Basically don't confuse a technology that can be used for bad purposes, to be the problem, blame the brokers and the regulators that allow it to happen.
With the current rules in place this isn't illegal, it would be if you coerced the information through some out-of-band system i.e. "Hey bob tell me what you're going to do five minutes before you do it and there's a nickle in it for you" but currently HFT front running is taking advantage of market arbitrage anomalies and trend prediction along with quick response to news events.
HFT adds nothing to the markets of value[1] and lowers the value average players are able to extract.
[1] except liquidity which they usually don't add in a useful manner, they will tend to make highly liquid stocks more liquid while not adding any liquidity to stocks that are being harmed by illiquidity.
I like to think anything that adds liquidity to markets are beneficial. It prevents large swings as everything is always priced correctly at that moment.
Insurance is an apt analogy. Intuitively, if someone consistently makes money, they're just skimming without providing any value. People were astonished that Virtu made money almost every day, but wouldn't care if GEICO made money almost every day.
In statistical expectation, you are better off not selling your stock to an HFT market maker. They are only buying from you, on average, when the spread you pay is more than they expect the stock to move. You could wait and sell to someone else for a tiny bit more, on average.
In statistical expectation, you are better off not insuring your car. The insurance company writes the policy such that their expected payout over its lifetime is less than the premiums they collect.
But real people don't live in the world of maximizing expected value. The guy selling his stocks is doing it a couple times a year, and doesn't want to risk losing a few % if the market gaps down, just to make a tenth of a penny more in expectation. Nobody wants to take out another car loan if their shiny new SUV gets hit driving out of the dealership. Bearing risks people don't want is valuable.
So bizarrely, while these services may satisfy a need, consumers are only really happy when the provider loses. Nobody looks back on decades of crash-free driving reminiscing about paying those insurance premiums. Nobody likes to sell stock and see it tick up.
(And I'm hand waving assuming you're a randomly selected trader or driver. If you have inside information, trading with anyone is positive EV. If you drive drunk at 100mph but have no tickets or DUIs on your record, insuring your car is positive EV.
I'm also hand waving away comparative advantage. That better price an HFT market maker gets when turning over your trade may not be achievable for you. Maybe they can trade on more exchanges or predict prices better to get out at the right time. They do this all day so trading optimally is not a waste of their time, but it's probably a waste of yours. In that case, you may lose more and take more risk doing it yourself. I think that's closer to reality.)
> you are better off not selling your stock to an HFT market maker
You are also better off selling into a market containing at least on HFT market maker. Without them, spreads become fat. It is notable that the principle aggressors, at the political level and until recently, when HFT became an armchair economics term of art, against HFT were the old line market makers.
That's fair, but the reason isn't entirely due to bias. Few people outside the industry really understand market microstructure. It's a dry and complex subject. Legitimate criticisms are pretty subtle and get lost in the noise of "zomg guys with lasers front running Joe Average's 3 share GOOG trade!"
I no longer work in HFT. You can do everything better, smarter, faster than ever, and make less. It is one of the most brutally capitalist businesses: orders on exchanges are a pure undifferentiated commodity. The guy making the smallest spread fastest gets the trade. [1] Nobody cares if it's Virtu or Citadel or three guys in their garage. Think of a business like Wal-Mart's, but if someone finds a way to sell soap a half-cent cheaper, you sell 0 and go out of business.
And for the end users of the markets, that's great. Let these guys compete to make tighter markets or arbitrage prices sooner. HFTs play an intermediation role, helping other traders transfer risk immediately to lock in a guaranteed price, rather than waiting for someone else to trade with them. They're basically like CarMax or a grocery store: research what things are worth, buy at wholesale prices, sell at retail prices, and control inventory risk.
If it makes you feel any better, I don't work for an HFT firm. Instead, my job involves trading against them. To do that effectively, it's necessary to understand what they're actually doing and how they make their money.
I speak up here because it irritates me to see them maligned by people who know what they're talking about.
Those with short-term information are the ones who benefit most by increases in the time taken to disseminate information. Let's say some fund's AI parses and interprets a favorable article 25 ms before anyone else. They start simultaneously buying stocks in NYC while writing puts and buying calls in Chicago.
Moisture in NJ slows down news from Chicago to NYC by 2 ms.
Automatic market making algorithms in NYC, in the absence of evidence from the options markets, bet on mean reversion and don't adjust their bids and asks as quickly as they would had information arrived from Chicago.
In those 2 ms, two investors each sell 100 shares a penny cheaper than they would have otherwise to the institutional investor with the news-reading AI.
A retired teacher and traditional hedge fund manager are each $1 poorer and someone writing a newsreading bot is $2 richer than they would have been had NJ been drier.
The individual stories are pretty boring, but millions of times a day, a small short-term information edge is slightly more valuable. Sure, 99% of the time both the winners and the losers are institutional investors. However, if you just look at a retail investor (or some institutional investor representing mostly retail investors), they're almost never on the side with a short-term informational advantage.
On the one hand, it's 99% just institutions winning and institutions losing. But, when you zoom in on those 1% of cases where an unwitting retail investor got lucky or got unlucky, there's a bias in their luck due to their lack of valuable short-term information.
On the other hand, those 2 ms might also mean that if a big retirement fund is unloading $10 million in stock to a bank, that bank has an extra 2 ms to simultaneously start selling the stock and get short in the options market before the low-latency market makers adjust prices. Maybe this means the bank is willing to bid $2 higher for the retirement fund's block trade.
In any case, the actions of the HFTs have both positive and negative knock-on effects for ordinary retail investors, even if other HFTs account for the vast majority of the HFTs' profit margins. Sometimes the short-term information comes from the outside world, and sometimes it comes from institutional investors trying not to telegraph their moves on behalf of retail investors.
Yes, they are taking on risks by making these trades. Do insurance companies 'earn' money? Do bail bondsmen 'earn' money? Does a retirement account 'earn' money?
The definition that google returns for the word "earn" is: 1. obtain (money) in return for labour or services. 2.
gain deservedly in return for one's behaviour or achievements.
I think it's interesting to see the contrast in those 2 definitions. In the first place you concern yourself with whether or not a service has been performed. In the second place you concern yourself whether or not the money is deserved by way of the action/accomplishment.
I think this neatly separates the two kinds of thinking on this sort of topic. On the one hand, people are appalled that people are making a lot of money without providing any particularly useful service (in the case of HFT, I suppose we could argue that it increases liquidity, which is helpful in some ways, but the market had ways of dealing with that before HFT so I think it's not a super strong argument).
On the other hand, there are people who admire the cleverness, skill, impetus and daring of people who can find ways to make money without providing some kind of traditional service. To those people, the money has been earned by virtue of their ingenuity/initiative/luck.
I think the conflict here is that to many people, the world would be a better place if we valued #1 considerably more than we value #2 (possibly to the point of not valuing #2 at all). It is, of course, not the world we currently live in, but I can understand the sentiment.
Hft does provide a service: taking on financial risk. This is a hard concept for people not in finance to understand, but think of it like insurance if that helps you.
The typical HFT firm will have it's computers colocated inside the exchange and will be working the order books of instruments listed on that specific exchange.
There are some "complicated" smoke and mirror setups for eg where execution reports of the HFT are "captured" by some "big broker" on the street and HFT's trades are booked by the big broker, but again, the big broker's computer is also in this case colocated in the venue.
What is exactly slowing down HFT ? What is the overlap of instruments between these venues ?
Unfortunately there are over 50 ATS in the US, and although most people think the “stock market” is in New York, it’s really in Chicago (eminis run the world these days). Trading spooz against single names as HFT do often use microwave antenna transmission.
At least in the case of cash equity, as soon as a buyer matches a seller, the price of the instrument should be updated in real time.
Same applies to inserting orders in the order book, there should be zero delay, so that the exchange can halt to auction as soon as it detects out of the norm bids (or asks).
The radio route between NYC and ORD is quite a bit faster than the fastest fiber path, its about 8.5ms vs 13.1ms. They have much better as-the-crow-flies paths, and the transmission latency is lower.
HFT links use microwave frequencies. Water absorbs microwave radiation quite well (cf. microwave oven); as the water content (humidity) of air increases, so does microwave absorption. Conversely, signal power delivered to the receiver drops, and SNR decreases, making reception more difficult.
Water should intuitively have a lower velocity factor, too, so waves might actually propagate more slowly in high humidity air compared to dry air.
No, that’s not how it works. That’s a 1 dimensional view.
We will have more wild variations of weather, so some places will get much, much more humid and others will get much much drier. And maybe the same place will get wetter and drier in the same season. It will be disastrous, but in a wilder and more unpredictable way.
Absolutely correct. What's happening is the increase of total energy in a titanic, global chaotic system. By definition, and even in a purely mathematical model, you'll get more intense 'outlier' events in both range and statistical behavior. In real life, exactly the same thing happens and is happening.
I agree with you, weather is different during different seasons and other factors. No matter what the global average temperature is increasing, with that is the certainty that areas affected will have increased capacity for moisture in the air. As of 2017, this is the majority of the globe, will have greater capacity for increase humidity from previous years. I've provided data showing the anomaly over time is warmer for that majority of our planet.
A buddy of mine used to work on radio internet stuffs. He said water does all kinds of strange things to the signal and it's path. Transmitting over large bodies of water and in full sports stadiums can be problematic
