HFT through the years

Quote from EliteTraderNYC:

Wheres the chart?
More...
Sorry I don't know what happened to it. I tried to embed it. Instead, here is the link:

http://www.nanex.net/aqck/2804.HTML

You can probably use it to discern useful statistics as well. The open and close are particularly interesting. But the purpose of the chart is to show just how frenetic the markets have becomes since even as late as 2007.

Note also that it is High Frequency QUOTING that is the concern, not trading.
 
Id like to think someone in the gov would do something about all the BS, but considering we are still stuck with TBTF, QE and other band aids nearly 5 years after I crisis Im not holding my breath for any changes. However, I think the exchanges have huge operational risk and black swan risk due to too much tech and complexity. One big event and the regulators will have to dues something and the exchanges shares could plummet instantaneously/
 
Quote from JamesVU2000:

Id like to think someone in the gov would do something about all the BS, but considering we are still stuck with TBTF, QE and other band aids nearly 5 years after I crisis Im not holding my breath for any changes. However, I think the exchanges have huge operational risk and black swan risk due to too much tech and complexity. One big event and the regulators will have to dues something and the exchanges shares could plummet instantaneously/
More...

The government is regulating in the name of lobbyist who have access to money and are interested in more regulation as a high barrier of entry to prevent competition and transparency. You want more of it?
 
a) of course is it unjustified to modify an order 1000 times vs one execution. It clearly shows that the core interest of the party behind the order is NOT to seek an execution at a fair price and that the intend is NOT to provide or absorb risk. Yet it will be very hard to abolish such practice as the very parties that abuse the system pay their dues directly to their own industry regulator. Absurd but very true.

b) So how do you delineate trading and investing. I claim there is no such delineation other than the holding period difference. Stocks are not zero sum, simply because overall they reflect market perceived company valuations and in the short term, even millisecond time frame, a hft house moves prices around that a large investor may be seeking to build or reduce his long-term holding.

The vicious marketing trick of hft participants is that they claim hft reduced spreads and added liquidity while the truth lies in the fact that spreads came in because of other industry initiatives, increased trading volume (up to 2008), decimalization, better access by trading participants,...

Quote from vicirek:

Agree with a) and my response was only to one participants claim that cancelling 1000 orders per one fill is unjustified - how do you know which cancellation is "good" or "bad"

On b) we are talking different things; I talk about trading where money is being made or lost from available pool of liquidity and not from the investing perspective. Investing and trading are separate things.
More...
 
Forget Nanex, it was just shown recently that Nanex's clock is totally off and most all their measurements are not even worth the toilet paper they are printed on.


Quote from nitro:

Sorry I don't know what happened to it. I tried to embed it. Instead, here is the link:

http://www.nanex.net/aqck/2804.HTML

You can probably use it to discern useful statistics as well. The open and close are particularly interesting. But the purpose of the chart is to show just how frenetic the markets have becomes since even as late as 2007.

Note also that it is High Frequency QUOTING that is the concern, not trading.
More...
 
Quote from CalVolibrator:

Forget Nanex, it was just shown recently that Nanex's clock is totally off and most all their measurements are not even worth the toilet paper they are printed on.
More...

it's not nanex's clock that was off [for the fed data analysis piece the did that you're referring to]. it's that they use SIP timestamps for their analysis, and they don't factor into the equation the SIP latency.

basically, the SIP is where all the exchanges send their direct feeds to, which calculates the public NBBO, but it takes time for it to process. ~1-2 ms during normal activity, and more during heightened activity. nanex was making the incorrect assumption that the SIP timestamps were reliable and reflected actual trade/quote times.

the only way nanex could have done a reliable study on this, is if they had machines colo'd at each exchange, were pulling each direct feed, had a syncrhonized GPS time clock across each location, and were creating a picture from that.

the only thing they were able to prove, is lack of reliability/resolution of the SIP timestamp. also, they were able to prove that it's probably better to make sure your data is good before crying fraud from the rooftops. would have saved them some embarrassment.
 
Quote from CalVolibrator:

a) of course is it unjustified to modify an order 1000 times vs one execution.
b) So how do you delineate trading and investing.

The vicious marketing trick of hft participants is that they claim hft reduced spreads and added liquidity while the truth lies in the fact that spreads came in because of other industry initiatives, increased trading volume (up to 2008), decimalization, better access by trading participants,...
More...

a) yes it is but how to differentiate between market participants in real time if free and equal access to markets is very important

b) investing = income from investment, dividents etc; trading is when you go to the Exchange to exchange your holdings for money or acquire new holdings regardless of holding period.

c.. good point
 
Did a bit of reading up on this issue, and I happily stand corrected, kudos to you. Pretty accurate analysis and summary. Thanks!


Quote from onelot:

it's not nanex's clock that was off [for the fed data analysis piece the did that you're referring to]. it's that they use SIP timestamps for their analysis, and they don't factor into the equation the SIP latency.

basically, the SIP is where all the exchanges send their direct feeds to, which calculates the public NBBO, but it takes time for it to process. ~1-2 ms during normal activity, and more during heightened activity. nanex was making the incorrect assumption that the SIP timestamps were reliable and reflected actual trade/quote times.

the only way nanex could have done a reliable study on this, is if they had machines colo'd at each exchange, were pulling each direct feed, had a syncrhonized GPS time clock across each location, and were creating a picture from that.

the only thing they were able to prove, is lack of reliability/resolution of the SIP timestamp. also, they were able to prove that it's probably better to make sure your data is good before crying fraud from the rooftops. would have saved them some embarrassment.
More...
 
Id take anything Nanex says with a grain of salt. They strike me as attention seekers who make sensational headlines to extend their perceived 15 minutes of fame. If Eric would take his blinders off, perhaps he'd see that a good deal of his nonsense is explainable. His self proclaimed expertise in HFT is occasionally laughable.

I should probably unsubscribe from Eric's Twitter feed. The witch hunts and look-at-me tirades are getting old. However, I find it amusing when he gets so irritated when people other than the Zero Hedge crowd don't take him seriously.

The rejection of their proposal in favor of a much more expensive Midas is a prime example of what regulators (among others) really think of them.
 
You must log in or register to reply here.
Back
Top