Been busy last few days but this is too good to pass up a response to... 777 whhy not post the actual FINRA release and skip the opinion? ZeroHedge is either intentionally trying to lie/slander or they are just a bunch of idiots that don't have a clue about the markets. I say ZH because Business Insider just pulls articles from ZH and 777 uses them because it seems more legit. Look at the bottom of the article he posted and you will see "via zh". Here is the FINRA release:http://www.finra.org/Newsroom/NewsReleases/2010/P121951 NASDAQ reported Trillium to FINRA: "This conduct was initially referred to FINRA by NASDAQ's MarketWatch Department." (quoted from FINRA release above) Also, ZH clearly does not understand Quote Stuffing, neither does 777: http://seekingalpha.com/article/225204-trillium-wasn-t-quote-stuffing Seeking Alpha is slightly more legit than ZH. Trillium was not charged with Quote Stuffing - actually they weren't charged with anything, no one admitted any guilt, they just paid the fines and moved on. 777 why not read and try to understand the crap spam/troll us with?
PM me... I'll send you my resume an anonoymous poster quoting a legit source is still providing a legit source. I never offered my opinion outside saying you have no clue what you are talking about and I think ZH has no clue - other than that I provided a FINRA link and a valid, widely accepted source (seeking alpha) for ET'ers to read.
Agree that ZH misused the term quote stuffing on this one. Nonetheless, layering/market manipulation is illegal. About time someone gets fined for stuffing the book with fake size.
Now, let's get rid off the useless noise making bots, aka quote stuffing algos. MAN 1, BOT 0. http://www.nanex.net/FlashCrash/FlashCrashAnalysis_CBID.html BATS Improvement: CancelBot is Dead (or in hiding). Publication Date: 09/15/2010 In our Initial Flash Crash Analysis and in our continued Crop Circle Of The Day page, we have shown an algorithm originating from the BATS exchange, which we deemed "CancelBot". As of 09/15/2010 it has been two complete weeks that we have not seen a single occurrence of CancelBot. This is unusual in that CancelBot was an algo easy to find and ran on every trading day. It appears the BATS exchange has done due diligence, ferreted out this algo and put a stop to it. We will continue to monitor for signs of a CancelBot reappearance but as of this writing we are happy to report CancelBot is no more. Below is a small description of CancelBot and a few images demonstrating it's behavior: -------------------------------------------------------------------------------- CancelBot CancelBot is unique to the BATS exchange. CancelBot can effect the ask side and the bid side simultaneously or independently. On the bid side it will start near the BBO, cancel the order, drop the order by a penny, cancel the order, drop the order by another penny and again cancel the order, until the price reaches 0.01 and the sequence starts back near the BBO. Conversely the ask side of the algorithm raises the price by a penny until a level is hit when it reverts back to near the BBO. CancelBot generally runs from 100 quotes to 1,000 quotes a second. http://www.nanex.net/FlashCrash/CancelBot/CancelBot1.png
I've read the Nanex study a few times. It is bias in my opinion (but whatâs that worth). Personally (and this is my opinion), I think quote stuffing is total bullsh*t. What happens is that HFT shops see an increased volume in a stock and they step it up as well. I know a lot of guys in that space and none of them want to cloud up and confuse the quotes by spamming the world with quotes. The way that Quote Stuffing has been portrayed by the media is similar to a DOS (Denial of Service) attack on a website. Thatâs malicious, not productive. Layering on the other hand, has been around since the advent of electronic trading and has been in use for arbitrage strategies since the advent of ETFs and Cash vs. Futures strategies (as I understand it, Iâm sure someone could correct me as to when this all started but it was way before I started trading). Take ticker SPY for example. One could conceivably put out orders across all of the S&P500 (or the SPYâs basket of stocks) in an effort to trick traders (or MM algo or exchangeâs matching engines) into thinking that there is demand in a certain direction (this is what Trillium was doing). Once the stock moves that way, you pull your orders and the stock goes back to where the true market demand is. Layering =/= quote stuffing as layering is placing ONE or very few orders pretty far up/down the book (non marketable orders) while what I understand Quote Stuffing to be is spamming an order book so much that certain matching engines and traders canât keep up. Iâm not sure what Quote Stuffing accomplishes other than to me it sounds similar to a DOS attack. Many firms have inferior hardware and therefore the traders accuse HFT of quote stuffing while the flaw is within the firmâs quote and execution servers, not the MM or exchange matching engines. A minimum of 50ms posted order rule seems fine â I canât even execute that fast so for me to send and order followed by a cancel, as fast as my hardware will allow â Iâd be well within that limit. Increased volume yields increased number of quotes. People tend to forget that you canât produce a quote â only an order. So when people send marketable orders they possibly turn into a quote for a brief amount of time but very quickly turn into a print. When people send non-marketable orders they turn into quotes. So quote stuffing really means that firms are sending non-marketable orders at a rate that is considered to be unacceptable or manipulative â to someone. Where does it end? Regulating âquote stuffingâ is really regulating the ability to place orders⦠which is a slippery slope. Also, my execution is considered pretty fast by many (people like 777 accuse me of being HFT) but by most HFT standards Iâm very very slow. Many shops within the HFT space are in the mili/micro second space so 50ms is light years to them. The other thing about an order cancel time limit/delay is that for people in the micro/nano space it will be VERY VERY easy to just look at the order book and know which orders have been out there for 50ms and make a fairly accurate guess as to which orders will be pulled. The traders will always find yet another way to game the system â and my fear would be that yet again, it will be at the expense of the retail shareholder. Itâs another one of those âbe careful what you wish forâ type things. The Nanex article does a very good job of pointing out things like OMG there were 5000 quotes in under a second â but they donât disclose the stock, they donât disclose what the average volume is, average size of the book and other pertinent information that would make the study more objective and less bias, in my opinion. 5000 orders per second is not unreasonable in Citigroup for example. However, if a stock that trades 100k shares a day all of a sudden experiences massive volume and massive orders â well thatâs a different story and probably much closer to why Trillium got fined for Layering. The Nanex Study, in Recommendation #1, recommends that they want quotes to be time-stamped and disseminated in such a way that time stamps are part of the messages. I agree with them that this is a trivial implementation however I think they are missing the broader perspective. Be careful what you wish for⦠By adding time stamps you are going to be sending quotes out of order due to latency issues, etc. AND the quote messages will now contain MORE data. People like 777 donât realize that now they are going to need loads more computing power and raw speed to handle the data â traders will need to filter the data and put them in order (which takes time) as well as parse through larger messages which take up more bandwidth, RAM, Cache & CPU usage. In my opinion implementing Recommendation #1 would skew things even more against the little guy simply from a data-handling/processing perspective. Again, all of that is my opinion⦠take it for what itâs worthâ¦
It seems we pretty much agree on all that layering/quote stuffing thingy. Although until I'm proven otherwise I still think quote stuffing is a serious threat to the market stucture as shown by the nanex guys. I would have to read the articles again to see if they mentioned which stock was quote stuffed 5000/s but have a look at this page where they collect on a daily basis evidences of quote stuffing algos in different stocks stocks:http://www.nanex.net/FlashCrash/CCircleDay.html Lastly I definitely agree that whatever comes out of this HFT buzz, it will still be at the expense of the little guy - most of us make money off the dumb money but that doesn't mean we shouldn't look to rule out certain practises when they threaten the market strucutre. Anoter flash crash is the last thing we want, right? edit : not sure what you mean by By adding time stamps you are going to be sending quotes out of order due to latency issues, etc. AND the quote messages will now contain MORE data. I thought the order was times stamped at the time it is disseminated and the recommendation was to time stamp it as it enters the queue... so how could that change the size of the message? thanks for clarifying what you meant.
I don't accuse you of being hft, I accuse you of trying to justify every possible manipulation as a shill.
Iâm not sure I follow. Layering is bad I agree but the way you say âall that layering/quote stuffing thingyâ seems like you are still lumping them both in together and donât understand the difference. I donât think Quote Stuffing is a legit complaint â I think itâs just a complaint made up by people with outdated or inferior hardware who canât handle true market data feeds. Furthermore I think that Quote Stuffing is next to impossible to prove and that itâs a very slippery slope in terms of regulating who can place orders, how many orders and what types of orders. As I said earlier, Quote Stuffing is described similarly to a DOS Attack and therefore I agree that in concept it is a serious threat to the market â only in that it is vulnerable. I personally think the âNanex guysâ love all the attention they are getting from uninformed market watchers/participants. After all, they do sell market data feeds. The pretty pictures they put out are interesting to look at but donât really show anything other than the dayâs historical price action. Iâm going to call Nanex today and Iâve requested a trial/demo of their product via email. Iâd love to be proven wrong but they flat out admit on their website that they introduce latency into their data feeds at the origination and destination points. Iâve said all along that they have a bias against HFT and ultra-low latency and now its much more clear. Their product INTRODUCES latency therefore it has no traction (IMHO introducing latency has no business being in the markets at all). By hating on the low-latency players in the market they can distract the uninformed into thinking that their âcompressedâ data feed is an OK product because you get everything the big boys get in a smaller package blah blah blah⦠Only problem is that by introducing latency to compress/decompress the data they are essentially selling you HISTORICAL data and passing it off as real-time market data. To me, thatâs fraud, but Iâll wait and talk to them before I start alleging things like that. Iâll make a point to ask. They have not responded to my emails (and I donât expect them to) but Iâd imagine that based on their apparent bias and the products they offer â they probably cherry picked for marketing purposes and donât want to disclose that because it will further discredit them. (again just a hunch/opinion) These pages have been all over the internet since some clown at Zero Hedge misinterpreted them and posted them on a blog. Again, all those pretty pictures show is historical tick data, charted to look pretty. They use straight/jagged lines and right angles to make it seem like someone is manipulating the markets, etc. when in actuality itâs just traders doing what they have always done. One of the programmers I work with did the same thing for the historical data in the S&P pit 15 years ago and he made it look worse â and that was well before the times of HFT and all manual/voice trading. Also, those pretty pictures donât demonstrate Quote Stuffing or Layering, they may show someone walking up/down the price of a stock but there is nothing provided to show any type of Quote Stuffing or Layering. It canât really be avoided â Telling a market participant that they have to be in/stay in the markets is just as bad/dangerous as telling someone that they canât be in the market or canât place orders (reference to Quote Stuffing). If you were to tell MMs and Liquidity providers that they are not legally allowed to step away when the market plummets then they would just introduce wider spreads as âinsuranceâ against something happening again. All that does is hurt the little guy. You cannot tell a market participant when they can be in/out of the markets, it is solely their choice and should never be regulated. (but yes, itâs a bad thing and no one needs another day like that) Iâm going to be honest and say that I donât fully understand exactly where timestamps are assigned to events, nor do I know if there is an exchange-wide standard, or of exchanges can assign timestamps wherever they want and however best suits them. Iâm going to take my best shot at it but know that this is just how I understand it so if someone knows better please correct me or fill in the blanks. There are two parts to quote/data delivery: Execution/Matching Engine and the Data/Quote Dissemination Engine/Server. An order hits the execution/matching engine, is processed and then sent to the data/quote dissemination engine/server where it is then forwarded to all subscribers of that exchange. Most of the top exchanges are in the nanosecond space so assume for the time being that there is zero latency between the matching engines and the dissemination servers â for all I know some exchanges might have one computer doing both tasks which truly would put the latency at zero. What I do know is that due to latency, outside prints, errors, bad trades, busted trades, market makers/specialists adding executions late â or backdating a trade to âaboutâ when it happened, etc. the matching engines donât always send data to the quote dissemination engines in the exact order that they happened. Since the data coming out of the matching engine is not in the exact order that it happened, I donât know if the matching engines assign CORRECT timestamps to the data (meaning that as the dissemination servers receive the data the timestamps are out of order) or if the dissemination servers receive raw non-time stamped data and apply INCORRECT times to the quotes because the quote dissemination server would be applying sequential timestamps to the data as it receives it â except the data is out of order. If I understand it correctly, I think that the matching engines apply the correct timestamp to the data as it sends over to the quote dissemination servers. This means that the quote dissemination servers are receiving the CORRECT data with the CORRECT timestamps â just out of order. I think that this is the bone of contention that people are bitching about⦠the data is out of order, therefore its bad, if timestamps arenât in correct order then how can we parse the data, blah blah⦠(I pay for raw market data and âscrubbedâ historical data where things are put into order for me after the fact so my back tests are more accurate, so I donât really know and it does not affect me.) I *think* that what people want is to have the quote dissemination servers timestamp the data as it receives it, giving the market participants a more âtrue pictureâ of the data flow and when it actually happens. If I understand all of this correctly, my gripe with time stamps at the quote servers is that they will be incorrectly time stamping the trades. Even though the quote servers would be time stamping as they receive the data, it is known that the data is out of order, therefore, in my opinion, the argument for quote servers applying timestamps is also an argument for degradation of data. With regards to the size of the message, I assume that somewhere in there they will need to add a tracer as to when the print actually occurred vs. when it was disseminated. I could be all wrong about this â but thatâs how I understand it. LOL well when you say things like âheâs one of themâ it doesnât seem like it. Accuse me of whatever you want â Iâll keep accusing you of being an ignorant troll and not understanding the markets enough to comment in your own thread.
accuse all u like. 99% of the news flow is in my favor, and only the pitiful pleadings of chat room shills on the other side. Dude asking you for an opinion on this is like asking a NAMBLA member to chaperon the school outing.