Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.misc > #2282 > unrolled thread

High frequency trading and a 4ms faster cable

Back to article view | Back to comp.misc

Contents

  High frequency trading and a 4ms faster cable RS Wood <rsw@therandymon.com> - 2014-04-06 19:20 +0000
    Re: High frequency trading and a 4ms faster cable Randy Westlund <Vorzakir@invalid.invalid> - 2014-04-06 19:44 +0000
      Re: High frequency trading and a 4ms faster cable Dag-Erling Smørgrav <des@des.no> - 2014-04-06 22:21 +0200
        Re: High frequency trading and a 4ms faster cable Randy Westlund <Vorzakir@invalid.invalid> - 2014-04-07 01:11 +0000
      Re: High frequency trading and a 4ms faster cable Hils <hils@saynotospam.net> - 2014-04-06 21:23 +0100
        Re: High frequency trading and a 4ms faster cable Randy Westlund <Vorzakir@invalid.invalid> - 2014-04-07 01:05 +0000
          Re: High frequency trading and a 4ms faster cable Dave Garland <dave.garland@wizinfo.com> - 2014-04-08 15:36 -0500
            Re: High frequency trading and a 4ms faster cable Randy Westlund <Vorzakir@invalid.invalid> - 2014-04-09 05:51 +0000
              Re: High frequency trading and a 4ms faster cable Dave Garland <dave.garland@wizinfo.com> - 2014-04-09 16:24 -0500
                Re: High frequency trading and a 4ms faster cable Randy Westlund <Vorzakir@invalid.invalid> - 2014-04-10 17:40 +0000
      Re: High frequency trading and a 4ms faster cable Hils <hils@saynotospam.net> - 2014-04-06 21:39 +0100
      Re: High frequency trading and a 4ms faster cable Hils <hils@saynotospam.net> - 2014-04-06 21:39 +0100
        Re: High frequency trading and a 4ms faster cable RS Wood <rsw@therandymon.com> - 2014-04-07 08:57 +0000

#2282 — High frequency trading and a 4ms faster cable

Pretty spectacular article at the Guardian about Michael Lewis' new book
detailing how high frequency trading is rigging the stock market and why
investors are willing to plough hundreds of millions of dollars into a
cable that would shorten network latency between Chicago and New York by
4ms.

http://www.theguardian.com/business/2014/apr/06/michael-lewis-flash-boys-high-frequency-traders

[toc] | [next] | [standalone]

#2285

On 2014-04-06, RS Wood wrote:
> Pretty spectacular article at the Guardian about Michael Lewis' new book
> detailing how high frequency trading is rigging the stock market and why
> investors are willing to plough hundreds of millions of dollars into a
> cable that would shorten network latency between Chicago and New York by
> 4ms.
>
> http://www.theguardian.com/business/2014/apr/06/michael-lewis-flash-boys-high-frequency-traders

I've done some reading about this before.  HFT scares me.  We have
thousands of algorithms running in NYC and Chicago, all trying to
buy and sell as fast as they can to rise the waves of whatever
trends they see.  The scary part is how this causes emergent
behavior of a complicated system that we can never hope to
understand.  There are flash crashes [1] that happen from time to
time, and we don't know why.  Maybe someone's algorithm trying to
create a trend?  Maybe just a race condition inherent in the
interactions of the algorithms?

And then we get silly things like property near the NYSE being
significantly more valuable because it can ping the NYSE in 0.5 ms
less time than other property in NYC.

1: http://en.wikipedia.org/wiki/Flash_Crash

[toc] | [prev] | [next] | [standalone]

#2289

Randy Westlund <Vorzakir@invalid.invalid> writes:
> HFT scares me.  [...]  There are flash crashes that happen from time
> to time, and we don't know why.  Maybe someone's algorithm trying to
> create a trend?  Maybe just a race condition inherent in the
> interactions of the algorithms?

You can see the same thing happening on a smaller scale on amazon.com.
Some second-hand booksellers use bots that monitor their competitors and
set their own prices (on the same book) just a little bit higher.  The
idea is to stay competitive while avoiding a race to the bottom.  Once
in a while, two of these bots cross paths and some obscure genre
paperback ends up with a price tag of thousands of dollars or more.

DES
-- 
Dag-Erling Smørgrav - des@des.no

[toc] | [prev] | [next] | [standalone]

#2300

On 2014-04-06, Dag-Erling Smørgrav wrote:
> Randy Westlund <Vorzakir@invalid.invalid> writes:
>> HFT scares me.  [...]  There are flash crashes that happen from time
>> to time, and we don't know why.  Maybe someone's algorithm trying to
>> create a trend?  Maybe just a race condition inherent in the
>> interactions of the algorithms?
>
> You can see the same thing happening on a smaller scale on amazon.com.
> Some second-hand booksellers use bots that monitor their competitors and
> set their own prices (on the same book) just a little bit higher.  The
> idea is to stay competitive while avoiding a race to the bottom.  Once
> in a while, two of these bots cross paths and some obscure genre
> paperback ends up with a price tag of thousands of dollars or more.
>
> DES

I hadn't considered that, but this could explain something that
happened to me a few years ago.  I saw my favorite hardcover Latin
textbook being sold for less than $3, so I tried to buy a dozen of
them.  The next day, Amazon had cancelled my order for reasons not
specified in their email and there was no record of the transaction
anywhere on their site.  The price of the textbooks was back to
~ $15 each.

[toc] | [prev] | [next] | [standalone]

#2290

On 2014-04-06 20:44, Randy Westlund wrote:
> On 2014-04-06, RS Wood wrote:
>> Pretty spectacular article at the Guardian about Michael Lewis' new book
>> detailing how high frequency trading is rigging the stock market and why
>> investors are willing to plough hundreds of millions of dollars into a
>> cable that would shorten network latency between Chicago and New York by
>> 4ms.
>>
>> http://www.theguardian.com/business/2014/apr/06/michael-lewis-flash-boys-high-frequency-traders
> 
> I've done some reading about this before.  HFT scares me.  We have
> thousands of algorithms running in NYC and Chicago, all trying to
> buy and sell as fast as they can to rise the waves of whatever
> trends they see.  The scary part is how this causes emergent
> behavior of a complicated system that we can never hope to
> understand.  There are flash crashes [1] that happen from time to
> time, and we don't know why.  Maybe someone's algorithm trying to
> create a trend?  Maybe just a race condition inherent in the
> interactions of the algorithms?
> 
> And then we get silly things like property near the NYSE being
> significantly more valuable because it can ping the NYSE in 0.5 ms
> less time than other property in NYC.
> 
> 1: http://en.wikipedia.org/wiki/Flash_Crash

We won't see any regulation of this while law is made not democratically
but by a minority who, if they are not already oligarchs can easily be
bribed by oligarchs. Perhaps the most democratic response is to disrupt
the trading networks and systems by any means available.

[toc] | [prev] | [next] | [standalone]

#2299

On 2014-04-06, Hils wrote:
> On 2014-04-06 20:44, Randy Westlund wrote:
>> I've done some reading about this before.  HFT scares me.  We have
>> thousands of algorithms running in NYC and Chicago, all trying to
>> buy and sell as fast as they can to rise the waves of whatever
>> trends they see.  The scary part is how this causes emergent
>> behavior of a complicated system that we can never hope to
>> understand.  There are flash crashes [1] that happen from time to
>> time, and we don't know why.  Maybe someone's algorithm trying to
>> create a trend?  Maybe just a race condition inherent in the
>> interactions of the algorithms?
>> 
>> And then we get silly things like property near the NYSE being
>> significantly more valuable because it can ping the NYSE in 0.5 ms
>> less time than other property in NYC.
>> 
>> 1: http://en.wikipedia.org/wiki/Flash_Crash
>
> We won't see any regulation of this while law is made not democratically
> but by a minority who, if they are not already oligarchs can easily be
> bribed by oligarchs.

But what's the proper way to regulate this?  I don't think they're
doing anything wrong, but it's leading to undesirable behavior.
Regulation of HFT would be very difficult to do justly, I think.

> Perhaps the most democratic response is to disrupt
> the trading networks and systems by any means available.

You're not suggesting we start DDoSing them, are you?  Doesn't seem
like a good move for the world economy.

[toc] | [prev] | [next] | [standalone]

#2313

On 4/6/2014 8:05 PM, Randy Westlund wrote:
> On 2014-04-06, Hils wrote:
>> On 2014-04-06 20:44, Randy Westlund wrote:
>>> I've done some reading about this before.  HFT scares me.  We have
>>> thousands of algorithms running in NYC and Chicago, all trying to
>>> buy and sell as fast as they can to rise the waves of whatever
>>> trends they see.  The scary part is how this causes emergent
>>> behavior of a complicated system that we can never hope to
>>> understand.  There are flash crashes [1] that happen from time to
>>> time, and we don't know why.  Maybe someone's algorithm trying to
>>> create a trend?  Maybe just a race condition inherent in the
>>> interactions of the algorithms?
>>>
>>> And then we get silly things like property near the NYSE being
>>> significantly more valuable because it can ping the NYSE in 0.5 ms
>>> less time than other property in NYC.
>>>
>>> 1: http://en.wikipedia.org/wiki/Flash_Crash
>>
>> We won't see any regulation of this while law is made not democratically
>> but by a minority who, if they are not already oligarchs can easily be
>> bribed by oligarchs.
> 
> But what's the proper way to regulate this?  I don't think they're
> doing anything wrong, but it's leading to undesirable behavior.
> Regulation of HFT would be very difficult to do justly, I think.

Put a 0.1% transaction tax on every completed transaction, and the
problem would cure itself. And probably solve the national deficit in
the process.

Putting a 5s delay between bid and acceptance (or not) would probably
do the trick, too. (I'm guessing that would make it comparable to
manual floor trading.)

I don't see why either could be considered "unjust".

> 
>> Perhaps the most democratic response is to disrupt
>> the trading networks and systems by any means available.
> 
> You're not suggesting we start DDoSing them, are you?  Doesn't seem
> like a good move for the world economy.
> 

[toc] | [prev] | [next] | [standalone]

#2319

On 2014-04-08, Dave Garland wrote:
> On 4/6/2014 8:05 PM, Randy Westlund wrote:
>> But what's the proper way to regulate this?  I don't think they're
>> doing anything wrong, but it's leading to undesirable behavior.
>> Regulation of HFT would be very difficult to do justly, I think.
>
> Put a 0.1% transaction tax on every completed transaction, and the
> problem would cure itself. And probably solve the national deficit in
> the process.
>
> Putting a 5s delay between bid and acceptance (or not) would probably
> do the trick, too. (I'm guessing that would make it comparable to
> manual floor trading.)
>
> I don't see why either could be considered "unjust".

I don't see anything necessarily unjust about that, but I don't
think it would solve the problem -- it would just change the time
scale.

In either case, algorithms would have to wait longer between trades,
either because they needed the share value to increase more to cover
the cost of the tax, or because of the hard limit caused by the
delay.  But we'd still have thousands of algorithms interacting in
nondeterministic ways and inexplicable share values that don't
reflect market conditions, which I think is the real problem.

Come to think of it... By skimming some off the top, the tax would
reduce the total amount of money invested in the market and reduce
the returns each person gets from investing in companies, which is
not what we want.

[toc] | [prev] | [next] | [standalone]

#2334

On 4/9/2014 12:51 AM, Randy Westlund wrote:
> On 2014-04-08, Dave Garland wrote:
>> On 4/6/2014 8:05 PM, Randy Westlund wrote:
>>> But what's the proper way to regulate this?  I don't think they're
>>> doing anything wrong, but it's leading to undesirable behavior.
>>> Regulation of HFT would be very difficult to do justly, I think.
>>
>> Put a 0.1% transaction tax on every completed transaction, and the
>> problem would cure itself. And probably solve the national deficit in
>> the process.
>>
>> Putting a 5s delay between bid and acceptance (or not) would probably
>> do the trick, too. (I'm guessing that would make it comparable to
>> manual floor trading.)
>>
>> I don't see why either could be considered "unjust".
> 
> I don't see anything necessarily unjust about that, but I don't
> think it would solve the problem -- it would just change the time
> scale.
> 
> In either case, algorithms would have to wait longer between trades,
> either because they needed the share value to increase more to cover
> the cost of the tax, or because of the hard limit caused by the
> delay.  

I posit that it would make algorithm trading far less attractive,
because it would increase the chances of a bad outcome. With the
transaction tax, you need to be confident that the price will change
(in the direction you want) by enough to at least cover the tax. With
the delay, you need to be confident that the price will continue to
change in your direction during the delay.

But we'd still have thousands of algorithms interacting in
> nondeterministic ways and inexplicable share values that don't
> reflect market conditions, which I think is the real problem.

Indeed. But how do you make algorithm trading unattractive? Right now,
for smaller investors (everybody except the 1%) it's a shell game, an
unregulated gambling device. Some friction needs to be introduced into
the system to damp wild oscillations.

> 
> Come to think of it... By skimming some off the top, the tax would
> reduce the total amount of money invested in the market and reduce
> the returns each person gets from investing in companies, which is
> not what we want.

If _my_ return was reduced by 0.1%, I wouldn't notice.  Now, if I was
trading in and out of that stock 50 times during the course of a day,
I'd notice. But those are the people who are causing the problems, not
the ones like Warren Buffet who are soberly researching a company and
then buying or selling it. In that case, 0.1% more or less probably
doesn't make much difference to them, either.

[toc] | [prev] | [next] | [standalone]

#2351

On 2014-04-09, Dave Garland wrote:
> On 4/9/2014 12:51 AM, Randy Westlund wrote:
>> On 2014-04-08, Dave Garland wrote:
>>> Put a 0.1% transaction tax on every completed transaction, and the
>>> problem would cure itself. And probably solve the national deficit in
>>> the process.
>>>
>>> Putting a 5s delay between bid and acceptance (or not) would probably
>>> do the trick, too. (I'm guessing that would make it comparable to
>>> manual floor trading.)
>>>
>>> I don't see why either could be considered "unjust".
>> 
>> I don't see anything necessarily unjust about that, but I don't
>> think it would solve the problem -- it would just change the time
>> scale.
>> 
>> In either case, algorithms would have to wait longer between trades,
>> either because they needed the share value to increase more to cover
>> the cost of the tax, or because of the hard limit caused by the
>> delay.  
>
> I posit that it would make algorithm trading far less attractive,
> because it would increase the chances of a bad outcome. With the
> transaction tax, you need to be confident that the price will change
> (in the direction you want) by enough to at least cover the tax. With
> the delay, you need to be confident that the price will continue to
> change in your direction during the delay.
>
> But we'd still have thousands of algorithms interacting in
>> nondeterministic ways and inexplicable share values that don't
>> reflect market conditions, which I think is the real problem.
>
> Indeed. But how do you make algorithm trading unattractive? Right now,
> for smaller investors (everybody except the 1%) it's a shell game, an
> unregulated gambling device. Some friction needs to be introduced into
> the system to damp wild oscillations.
>
>> 
>> Come to think of it... By skimming some off the top, the tax would
>> reduce the total amount of money invested in the market and reduce
>> the returns each person gets from investing in companies, which is
>> not what we want.
>
> If _my_ return was reduced by 0.1%, I wouldn't notice.  Now, if I was
> trading in and out of that stock 50 times during the course of a day,
> I'd notice. But those are the people who are causing the problems, not
> the ones like Warren Buffet who are soberly researching a company and
> then buying or selling it. In that case, 0.1% more or less probably
> doesn't make much difference to them, either.

You make good points.  My fiancee and I discussed it over breakfast
and couldn't come up with a better idea.  It would be nice if we
could test these ideas out before committing to them, though.
Anything implemented here could have drastic and far-reaching
consequences and lawmakers would have trouble agreeing on anything
without something to back it up.

Maybe what we need is a comprehensive simulation of the stock
market, complete with replicas of these algorithms, all interacting
in a sandbox (a dissertation in itself).  Then we could play around
with enacting various solutions and watch what they do.  A nonlinear
stochastic controls person might have fun with that.

[toc] | [prev] | [next] | [standalone]

#2291

On 2014-04-06 20:44, Randy Westlund wrote:
> On 2014-04-06, RS Wood wrote:
>> Pretty spectacular article at the Guardian about Michael Lewis' new book
>> detailing how high frequency trading is rigging the stock market and why
>> investors are willing to plough hundreds of millions of dollars into a
>> cable that would shorten network latency between Chicago and New York by
>> 4ms.
>>
>> http://www.theguardian.com/business/2014/apr/06/michael-lewis-flash-boys-high-frequency-traders
> 
> I've done some reading about this before.  HFT scares me.  We have
> thousands of algorithms running in NYC and Chicago, all trying to
> buy and sell as fast as they can to rise the waves of whatever
> trends they see.  The scary part is how this causes emergent
> behavior of a complicated system that we can never hope to
> understand.  There are flash crashes [1] that happen from time to
> time, and we don't know why.  Maybe someone's algorithm trying to
> create a trend?  Maybe just a race condition inherent in the
> interactions of the algorithms?
> 
> And then we get silly things like property near the NYSE being
> significantly more valuable because it can ping the NYSE in 0.5 ms
> less time than other property in NYC.
> 
> 1: http://en.wikipedia.org/wiki/Flash_Crash

We won't see any regulation of this while law is made not democratically
but by a minority who, if they are not already oligarchs can easily be
bribed by oligarchs. Perhaps the most democratic response is to disrupt
the trading networks and systems by any means available.

[toc] | [prev] | [next] | [standalone]

#2292

On 2014-04-06 20:44, Randy Westlund wrote:
> On 2014-04-06, RS Wood wrote:
>> Pretty spectacular article at the Guardian about Michael Lewis' new book
>> detailing how high frequency trading is rigging the stock market and why
>> investors are willing to plough hundreds of millions of dollars into a
>> cable that would shorten network latency between Chicago and New York by
>> 4ms.
>>
>> http://www.theguardian.com/business/2014/apr/06/michael-lewis-flash-boys-high-frequency-traders
> 
> I've done some reading about this before.  HFT scares me.  We have
> thousands of algorithms running in NYC and Chicago, all trying to
> buy and sell as fast as they can to rise the waves of whatever
> trends they see.  The scary part is how this causes emergent
> behavior of a complicated system that we can never hope to
> understand.  There are flash crashes [1] that happen from time to
> time, and we don't know why.  Maybe someone's algorithm trying to
> create a trend?  Maybe just a race condition inherent in the
> interactions of the algorithms?
> 
> And then we get silly things like property near the NYSE being
> significantly more valuable because it can ping the NYSE in 0.5 ms
> less time than other property in NYC.
> 
> 1: http://en.wikipedia.org/wiki/Flash_Crash

We won't see any regulation of this while law is made not democratically
but by a minority who, if they are not already oligarchs can easily be
bribed by oligarchs. Perhaps the most democratic response is to disrupt
the trading networks and systems by any means available.

[toc] | [prev] | [next] | [standalone]

#2303

On 2014-04-06, Hils <hils@saynotospam.net> wrote:
> We won't see any regulation of this while law is made not
> democratically but by a minority who, if they are not already
> oligarchs can easily be bribed by oligarchs. Perhaps the most
> democratic response is to disrupt the trading networks and systems by
> any means available.

Hils is probably right.  The regulators and indeed most politicians are
now billionaires, and partly due to gains they're making on the stock
market.  Why kill the golden goose?  We live in bad times, and oligarchy
only begins to describe it.  Plutocracy is more like it.  Maybe
wankerocracy - government by the wankers.

I don't think DOSing the trading system is the right answer though.  The
root ill is far deeper, and OT for this newsgroup. :)

[toc] | [prev] | [standalone]

Back to top | Article view | comp.misc

csiph-web