I can see insider trading modules being sold as an upgrade.
But... why? An insider has information which can be used once, and in a very low-frequency fashion, for maximum profit. Insiders don't need HFTs, because by definition they have an information advantage over everyone else.
How about an explanation on how HFT is invulnerable to destructive interference? Then follow up with how it is impossible for HFT to create, even accidentally, stock market chaos?
Why is this needed, when it was never needed for other types of speculators or market-makers in the past? You are seeking perfection in a system which has never been perfect.
And frankly, your" Just don't trade" silliness, is just stupid. I have funds in the stock market, and a lot of people do.
It's possible to minimize trading even with funds in the stock market. Pick funds which generate very low transaction costs (i.e. those which trade very little). Rebalance your funds on a low-frequency basis, maybe quarterly at most. You will find very little competition with HFT if you do this, because your holding periods will be so vastly different. A home-office day trader, OTOH, is practically getting in line for a sucker-punch.
When you trade on an exchange all the quotes, prices, volumes and trades are public.
...except the quotes that cannot be displayed because they lock the price shown at another exchange. Or orders that can execute at subpenny increments, but must be ranked at penny increments. Both of these contribute to a "dark pool" effect even on the lit exchanges.
I've also heard it claimed that the mechanisms used in high speed trading would be illegal if done by people instead of by computers.
That's absurd. On the contrary, regulators expect that all quotes are bona fide and tracked by an audit trail. No high-speed "shenanigans" will have any impact on a thorough post-trade audit. Where did you hear this claim?
The impression I got was that it involved making bids that you didn't intend to execute, so that you clogged up the system until it got to the price where you were willing to buy, but I'm not at all sure I understood this correctly.
The term "quote stuffing" comes to mind. That is abuse of the trading platform, effectively a DoS attack against the exchange and other participants. Definitely a bad thing. However, it has no intrinsic impact on the price that the "stuffer" can execute.
They have [become much more consistent], but they systematically favor larger players. Smaller players can't even afford the sort of trading that can get rolled back.
If you're referring to capital limits or the fact that small players can go bankrupt before they accidentally push a price through the limit, then I agree with you but I don't really see an alternative. Traders who can weather larger losses will always win.
So why did they cancel those trades in the recent incident? Why not keep all trades?
They canceled trades in the recent incident when the price exceeded predetermined bands. Why do they cancel trades at all? Well, to be honest I'm not exactly sure... I suppose there's some intent of "protecting investors" or "maintaining confidence in the markets". But anyone writing an automated trading system could have programmed in the rules to halt when trade prices exceeded the bands, so at least in the recent incident the operational risk could have been managed.
No I meant the cases where lots of trades got rolled back/cancelled.
Name a recent event. Really, I don't follow the news that closely, and as far as I know the trade-bust rules have become much more consistent as of the last few years.
As I stated in my post, the existing cross-exchange price rules on trade cancellation were created in response to this event. So it's not a very good example of unfairness, because, well... the rule-makers agreed with you. The unpredictability of bust-or-no-bust made hedging decisions impossible for everyone involved.
Specifically, what I'm objecting to is the practice of extracting 'book information' from the market by making a bunch of entries to probe for this information (which is not generally available to the trading public) and then canceling the orders.
I'm curious what sort of information you expect HF traders can see which is both non-public and also doesn't require trading. The closest that I can think of is "flash orders" which were briefly implemented on several exchanges, but quickly shut down. Otherwise, all that I can think of as "probing" is receiving (your own) executions before the public feed shows them. Executions naturally entail more risk and cannot be canceled the way orders can, so this doesn't seem particularly abusive or fraudulent.
This cancellation could be implemented by simply dropping the network connection or utilizing an error handling process intended for correcting errors in placed orders.
If the exchange supports cancel-on-disconnect functionality, it is typically done on a "best-effort" basis. That means that there is very little speed advantage available, the cancels are never guaranteed, and there is a substantial penalty for receiving executions after the requisite reconnect (hedging calculations require a known-executed quantity).
a HF trader continually probes the market and steps in to buy 100 shares at $2.00 and sell at $2.01 just before my order. So I get my shares at $2.01
If the HF trader swept all available liquidity at $2.00, your market order would indeed jump to $2.01. If 100 shares is all that's available, then you should have placed a limit order for $2.00 with the expectation that a thinly-traded stock will have higher volatility and you would probably execute at some point during the day. In any case, the only way to "probe" for a market order is to be the other side of it, because aggressive orders are only broadcast after they trade. So this entire example is either pure fiction, or a brokerage front-running a customer order (which is fraudulent, but is orthogonal to HFT).
But that's not the way HFT works. If I make money, I keep it. If I'm about to lose money, I drop the network connection. That's fraud.
That's not how HFT works, either. If you send orders that get executed, you are on the hook for them regardless of the state of your network connection. It's your problem to reconnect and re-download the executions which the matching engine has generated (and automatically sent to your clearer, who probably has a more reliable network connection anyhow).
b) your transactions get rolled back (see the HFT rollbacks).
Oh, you mean like the Knight Capital debacle where everyone got a free showing of "HFT market-maker gone wild"? The one where they lost over $400 million on 150 stocks? Trades were busted on only a handful of stocks based on preexisting price rules (IIRC, rules set in place to combat the uncertainty of ad-hoc or politically-motivated busting in response to the May 6 Flash Crash). All other trades stand.
Now, if Knight gets a "Wall St. Bailout", then you have yourself an argument... against bailouts, not trade busts.
What you are describing is a limit order. Some people use the terms "bid" and "ask" to describe "buy" and "sell" orders that rest in the order-book before passive execution, in which case "limit" is implied.
Somewhere in there is a simple and highly robust protocol struggling to get out.
Yeah, they call it OUCH. FIX is the "enterprise" solution, in that it requires an enterprise architect to craft the specification and it is still vague enough that some of the exchanges end up including errata in their own client specifications (section 5.4).
I don't think it's necessarily safe to assume that it wasn't a human erroneously thinking they were "resubmitting" their order to the client software, and the client software dutifully generating a new unique client order identity each time.
Oh, I totally agree. I can't imagine many trading UIs come with the "sends duplicate orders" feature; especially those developed in-house. However, owing to the speed of modern trading systems, I would hope that more UIs would be developed that limit in-flight (unacknowledged) orders.
If that was (say) your own mother who could just watch some of her savings evaporate (and your inheritance) purely and only because of so-called "technical errors"
I don't mean to defend NASDAQ at all (and IANA Financial Advisor), but my personal opinion is that (1) mom and pop should be treating IPO day like a day at the casino, not a rational investment; and (2) mom and pop shouldn't expect to open and close a position in a single day. They aren't day traders or scalpers, and they are sure to get burned by "overtrading" an IPO.
No, I want $1000 dollars of FB. If that's 40 shares at the time I place the order and 20 shares at the time the order is processed 4 hours later because Nasdaq screws up, I get 40 shares and a bill for $2000.
Allow me to introduce you to the limit order. You want to buy $1000 of FB and your screen/broker/google-finance says that it's $25/share? Send your limit order for 40 shares @ $25. If the price jumps up (regardless of the reason) you won't get filled and you'll have to try again later, but at least you're not stuck with a mystery bill for your purchase. If you sent a market order for 40 shares... well... I hope it was a good learning experience.
That depends on what the protocol spec says. If NASDAQ acknowledged receiving the order but didn't confirm that the order was placed, it's entirely possible that the right thing to do was send the order again under the assumption that it didn't get filled.
The OUCH specification says, "All Inbound Messages may be repeated benignly." If UBS sent multiple (identical) orders with the same Order Token, then they would (should?) have been fine due to filtering of duplicates. However, perhaps their client software didn't actually re-send identical messages (due to the rarity of this type of system outage, and laziness) but may have generated unique Order Tokens. Whoops, that's multiple orders, sent in a panic without thinking about the associated obligation.
They used to make this same argument about Greek and Latin: sure, you may not actually use these skills, but they teach critical thinking and build character. It was nonsense then and it's nonsense now.
The very little Greek that I learned in one college course was all about word-stems. I use it regularly as an aid to interpret new words or to help recall rarely-used words. The "critical thinking" and "character-building" arguments for foreign languages/words do sound like nonsense, but it is ignorance to believe that the information is not useful. It takes a lazy mind to ignore all of the Greek and Latin heritage in English.
I mean, do I "understand" (a+b)^2 == a^2 + 2ab + b^2? I can't say I really do.....
Here's a geometric interpretation which may actually help your understanding. Start with the equation area(rect) = w * h. That is, the area of a rectangle is the product of its two dimensions. Trivially, the area of a square is w*2. Now, take a square and cut it along both axes at the same distance from one side, and you should have two squares and two rectangles, with symmetry across the "hypotenuse" of the original square.
Now call the width of one square a and the width of the other square b. That means that the rectangles are each a x b. The area of the four rectangles is a*a + 2*a*b + b*b which is the same as the area of the original square, (a+b)*(a+b).
It's easier with a picture; or better, a piece of paper so you can crease the "hypotenuse".
Map of the Problematique sounds awful, which sucks because the song is otherwise such a good high-energy vamp. Also, there is a Guitar Hero-based "remaster" of Knights of Cydonia floating around the tubes which is purported to sound better than the CD release.
Muse is a big step up from Glee, but then so is a white-noise generator. It's not a very useful bar to set.
there are "sides". Specifically, they is "evidence" and "belief".
Calling the scientific consensus "truth" is begging the question. The scientific method is akin to credit for work shown rather than credit for correct answers (i.e. faith), so the honest argument is already over when "truth" starts getting thrown around. But based on the rest of your post, I suppose I'm preaching to the choir.
Of course if their root key was compromised they're fucked, but that one is deep in the goblin-caves of Moria and the only place it'd come out would be in the pocket of a hobbit.
Slashdot Mirror
But... why? An insider has information which can be used once, and in a very low-frequency fashion, for maximum profit. Insiders don't need HFTs, because by definition they have an information advantage over everyone else.
Why is this needed, when it was never needed for other types of speculators or market-makers in the past? You are seeking perfection in a system which has never been perfect.
It's possible to minimize trading even with funds in the stock market. Pick funds which generate very low transaction costs (i.e. those which trade very little). Rebalance your funds on a low-frequency basis, maybe quarterly at most. You will find very little competition with HFT if you do this, because your holding periods will be so vastly different. A home-office day trader, OTOH, is practically getting in line for a sucker-punch.
That's absurd. On the contrary, regulators expect that all quotes are bona fide and tracked by an audit trail. No high-speed "shenanigans" will have any impact on a thorough post-trade audit. Where did you hear this claim?
The term "quote stuffing" comes to mind. That is abuse of the trading platform, effectively a DoS attack against the exchange and other participants. Definitely a bad thing. However, it has no intrinsic impact on the price that the "stuffer" can execute.
If you're referring to capital limits or the fact that small players can go bankrupt before they accidentally push a price through the limit, then I agree with you but I don't really see an alternative. Traders who can weather larger losses will always win.
They canceled trades in the recent incident when the price exceeded predetermined bands. Why do they cancel trades at all? Well, to be honest I'm not exactly sure... I suppose there's some intent of "protecting investors" or "maintaining confidence in the markets". But anyone writing an automated trading system could have programmed in the rules to halt when trade prices exceeded the bands, so at least in the recent incident the operational risk could have been managed.
Name a recent event. Really, I don't follow the news that closely, and as far as I know the trade-bust rules have become much more consistent as of the last few years.
As I stated in my post, the existing cross-exchange price rules on trade cancellation were created in response to this event. So it's not a very good example of unfairness, because, well... the rule-makers agreed with you. The unpredictability of bust-or-no-bust made hedging decisions impossible for everyone involved.
I'm curious what sort of information you expect HF traders can see which is both non-public and also doesn't require trading. The closest that I can think of is "flash orders" which were briefly implemented on several exchanges, but quickly shut down. Otherwise, all that I can think of as "probing" is receiving (your own) executions before the public feed shows them. Executions naturally entail more risk and cannot be canceled the way orders can, so this doesn't seem particularly abusive or fraudulent.
If the exchange supports cancel-on-disconnect functionality, it is typically done on a "best-effort" basis. That means that there is very little speed advantage available, the cancels are never guaranteed, and there is a substantial penalty for receiving executions after the requisite reconnect (hedging calculations require a known-executed quantity).
If the HF trader swept all available liquidity at $2.00, your market order would indeed jump to $2.01. If 100 shares is all that's available, then you should have placed a limit order for $2.00 with the expectation that a thinly-traded stock will have higher volatility and you would probably execute at some point during the day. In any case, the only way to "probe" for a market order is to be the other side of it, because aggressive orders are only broadcast after they trade. So this entire example is either pure fiction, or a brokerage front-running a customer order (which is fraudulent, but is orthogonal to HFT).
I hear the term "resting order" used a lot.
That's not how HFT works, either. If you send orders that get executed, you are on the hook for them regardless of the state of your network connection. It's your problem to reconnect and re-download the executions which the matching engine has generated (and automatically sent to your clearer, who probably has a more reliable network connection anyhow).
Oh, you mean like the Knight Capital debacle where everyone got a free showing of "HFT market-maker gone wild"? The one where they lost over $400 million on 150 stocks? Trades were busted on only a handful of stocks based on preexisting price rules (IIRC, rules set in place to combat the uncertainty of ad-hoc or politically-motivated busting in response to the May 6 Flash Crash). All other trades stand.
Now, if Knight gets a "Wall St. Bailout", then you have yourself an argument... against bailouts, not trade busts.
What you are describing is a limit order. Some people use the terms "bid" and "ask" to describe "buy" and "sell" orders that rest in the order-book before passive execution, in which case "limit" is implied.
Yeah, they call it OUCH. FIX is the "enterprise" solution, in that it requires an enterprise architect to craft the specification and it is still vague enough that some of the exchanges end up including errata in their own client specifications (section 5.4).
Oh, I totally agree. I can't imagine many trading UIs come with the "sends duplicate orders" feature; especially those developed in-house. However, owing to the speed of modern trading systems, I would hope that more UIs would be developed that limit in-flight (unacknowledged) orders.
I don't mean to defend NASDAQ at all (and IANA Financial Advisor), but my personal opinion is that (1) mom and pop should be treating IPO day like a day at the casino, not a rational investment; and (2) mom and pop shouldn't expect to open and close a position in a single day. They aren't day traders or scalpers, and they are sure to get burned by "overtrading" an IPO.
Allow me to introduce you to the limit order. You want to buy $1000 of FB and your screen/broker/google-finance says that it's $25/share? Send your limit order for 40 shares @ $25. If the price jumps up (regardless of the reason) you won't get filled and you'll have to try again later, but at least you're not stuck with a mystery bill for your purchase. If you sent a market order for 40 shares... well... I hope it was a good learning experience.
The OUCH specification says, "All Inbound Messages may be repeated benignly." If UBS sent multiple (identical) orders with the same Order Token, then they would (should?) have been fine due to filtering of duplicates. However, perhaps their client software didn't actually re-send identical messages (due to the rarity of this type of system outage, and laziness) but may have generated unique Order Tokens. Whoops, that's multiple orders, sent in a panic without thinking about the associated obligation.
The very little Greek that I learned in one college course was all about word-stems. I use it regularly as an aid to interpret new words or to help recall rarely-used words. The "critical thinking" and "character-building" arguments for foreign languages/words do sound like nonsense, but it is ignorance to believe that the information is not useful. It takes a lazy mind to ignore all of the Greek and Latin heritage in English.
Here's a geometric interpretation which may actually help your understanding. Start with the equation area(rect) = w * h. That is, the area of a rectangle is the product of its two dimensions. Trivially, the area of a square is w*2. Now, take a square and cut it along both axes at the same distance from one side, and you should have two squares and two rectangles, with symmetry across the "hypotenuse" of the original square.
Now call the width of one square a and the width of the other square b. That means that the rectangles are each a x b. The area of the four rectangles is a*a + 2*a*b + b*b which is the same as the area of the original square, (a+b)*(a+b).
It's easier with a picture; or better, a piece of paper so you can crease the "hypotenuse".
Map of the Problematique sounds awful, which sucks because the song is otherwise such a good high-energy vamp. Also, there is a Guitar Hero-based "remaster" of Knights of Cydonia floating around the tubes which is purported to sound better than the CD release.
Muse is a big step up from Glee, but then so is a white-noise generator. It's not a very useful bar to set.
Calling the scientific consensus "truth" is begging the question. The scientific method is akin to credit for work shown rather than credit for correct answers (i.e. faith), so the honest argument is already over when "truth" starts getting thrown around. But based on the rest of your post, I suppose I'm preaching to the choir.
I don't believe this for a minute. You didn't misspell his name as "Osama" even once.
I love it! I remember seeing "happy meal ethernet" references in an internship in a UNIX support team.
FTFY. The root key wants to be found.