a*****e
发帖数: 1717 | 1 who starts first? does your system use market depth data? |
f*******s
发帖数: 3046 | 2 我，不特意做当天交易，没觉得有啥大用。搬张板凳学习先。
【在 a*****e 的大作中提到】
: who starts first? does your system use market depth data?
|
f********n
发帖数: 1560 | 3 月费太贵，没定。
【在 a*****e 的大作中提到】
: who starts first? does your system use market depth data?
|
s******5
发帖数: 112 | |
b*******e
发帖数: 24532 | 5 only useful if you do dt
【在 a*****e 的大作中提到】
: who starts first? does your system use market depth data?
|
w******s
发帖数: 16209 | 6 2020 mm was also interested on those. I think the most direct usage is to
tack the size and compare.
【在 a*****e 的大作中提到】
: who starts first? does your system use market depth data?
|
a*****e
发帖数: 1717 | 7 Oh, It is said it's less and less useful as ppl are
better and better at hiding their intentions or even manipulate them.
1. iceberg, first just split volume, then evolved to split random size
, then random time submission, then now, even cross exchange/route to fill
one order.
【在 w******s 的大作中提到】
: 2020 mm was also interested on those. I think the most direct usage is to
: tack the size and compare.
|
w******s
发帖数: 16209 | 8 that is true. and that is one of the reason I can't play options dt well any
more..
【在 a*****e 的大作中提到】
: Oh, It is said it's less and less useful as ppl are
: better and better at hiding their intentions or even manipulate them.
: 1. iceberg, first just split volume, then evolved to split random size
: , then random time submission, then now, even cross exchange/route to fill
: one order.
|
w******s
发帖数: 16209 | 9 the big size hanging there still mean something . if it is fake, then it is
useful information |
a****g
发帖数: 8131 | 10 观察过几个股，大盘股，high beta
level 2变化太快了
估计只有hft才能好好利用
感觉做短／长线还是ta／fa，对散户而言 |
|
|
a*****e
发帖数: 1717 | 11 regarding ES. let's discuss if level 2 is suggestive.
1. I don't think market maker will have edge here. Market maker
prefer bigger spread, but if it's designed to be so and it's highly
liquid and volatile, it's much much more risk to take than to get
1 tick profit. I am not 100% sure if CME disallow even market maker to quote
between one tick for futures like ES.
any
【在 w******s 的大作中提到】
: that is true. and that is one of the reason I can't play options dt well any
: more..
|
c**t
发帖数: 9197 | 12 NYSE or NASQ, Level II data today is useless. even for DT.
It use to carry large order info.(maybe 10 years ago)
But with the introduction of dark pool, island and algo trading.
It becomes almost random data now.
For example, MMID today is replaced by exchange ID and island ID,
so you can not tell who originally put the order, and most fund have
multiple island to route their order.
In additional, market testing computer continue issuing a lot of order and
cancel them within millisecond later to "discover" the price.
This is a good thing, market is more efficient in this regard.
As for future. Globex does not give MMID, only depth.
But with the introduction of "flipper" trading, it also random.
unless it is @ key level, which you can know from the chart anyway.
overall, I think people should focus on TA, do not try to find secret way in
the highly manipulated, second hand data. |
a*****e
发帖数: 1717 | 13 It's different if the big quote comes from many players or just one players.
I believe all level 2 shows only one aggregated size for each level of
depth, right?
is
【在 w******s 的大作中提到】
: the big size hanging there still mean something . if it is fake, then it is
: useful information
|
a*****e
发帖数: 1717 | 14 IOC is used mainly by market market to do HFT.
I am not sure how long SEC will limit flipper.
flipper is beyond gray area in my opinion. Other tricks
like IOC, penny jumper, tow the iceberg, you can still
argue it's trade by gathered public market information, even
only a small loop would detect them.
Order books are becoming noisier and nosier that's for sure.
The only reason retailers shouldn't dig into it is because
it's a game far beyond their abilities.
【在 c**t 的大作中提到】
: NYSE or NASQ, Level II data today is useless. even for DT.
: It use to carry large order info.(maybe 10 years ago)
: But with the introduction of dark pool, island and algo trading.
: It becomes almost random data now.
: For example, MMID today is replaced by exchange ID and island ID,
: so you can not tell who originally put the order, and most fund have
: multiple island to route their order.
: In additional, market testing computer continue issuing a lot of order and
: cancel them within millisecond later to "discover" the price.
: This is a good thing, market is more efficient in this regard.
|
w******s
发帖数: 16209 | 15 a lot of stocks only has few mm.
【在 a*****e 的大作中提到】
: It's different if the big quote comes from many players or just one players.
: I believe all level 2 shows only one aggregated size for each level of
: depth, right?
:
: is
|
w******s
发帖数: 16209 | 16 there are some sophisticated daytrading tools build upon level II info.
someone on the board ever post a set of tools with animated explain, but I
failed to find that link.
【在 c**t 的大作中提到】
: NYSE or NASQ, Level II data today is useless. even for DT.
: It use to carry large order info.(maybe 10 years ago)
: But with the introduction of dark pool, island and algo trading.
: It becomes almost random data now.
: For example, MMID today is replaced by exchange ID and island ID,
: so you can not tell who originally put the order, and most fund have
: multiple island to route their order.
: In additional, market testing computer continue issuing a lot of order and
: cancel them within millisecond later to "discover" the price.
: This is a good thing, market is more efficient in this regard.
|
w******s
发帖数: 16209 | 17 market depth is different from level II
【在 a*****e 的大作中提到】
: It's different if the big quote comes from many players or just one players.
: I believe all level 2 shows only one aggregated size for each level of
: depth, right?
:
: is
|
f****e
发帖数: 227 | 18 I had this question long time ago ... Is there anyway that we can "export"
the level II data and import it into our own SHIT TONG for real-time
analysis? We can view it, say using ScottradeElitte, but how to "download"
or "streaming" it to your computer?
【在 w******s 的大作中提到】
: there are some sophisticated daytrading tools build upon level II info.
: someone on the board ever post a set of tools with animated explain, but I
: failed to find that link.
|
a*****e
发帖数: 1717 | 19 I knew that phrase is confusing
【在 w******s 的大作中提到】
: market depth is different from level II
|
C*G
发帖数: 7495 | 20 我老文科男，野鸡经验，很久以前闲时用过htmlunit。
hiahia
【在 f****e 的大作中提到】
: I had this question long time ago ... Is there anyway that we can "export"
: the level II data and import it into our own SHIT TONG for real-time
: analysis? We can view it, say using ScottradeElitte, but how to "download"
: or "streaming" it to your computer?
|
|
|
g*****u
发帖数: 14294 | 21 To be able to do "real-time" analysis, you would need real-time data feed.
There are ways to continually ping your discount broker web interface to get
data snapshots. But, I doubt its usefulness would go much beyond a test of
ideas.
【在 f****e 的大作中提到】
: I had this question long time ago ... Is there anyway that we can "export"
: the level II data and import it into our own SHIT TONG for real-time
: analysis? We can view it, say using ScottradeElitte, but how to "download"
: or "streaming" it to your computer?
|
w*******o
发帖数: 6125 | 22 这个听起来有点，"以小散之短，攻MM之长"，
不划算,不划算.
get
of
【在 g*****u 的大作中提到】
: To be able to do "real-time" analysis, you would need real-time data feed.
: There are ways to continually ping your discount broker web interface to get
: data snapshots. But, I doubt its usefulness would go much beyond a test of
: ideas.
|
g*****u
发帖数: 14294 | 23 或许以后小散也可以在NYSE或其它venue租个co-lo。
如果确实有edge, 村长的资金就可以搞个quant fund. 有装备了就不是小散了。
【在 w*******o 的大作中提到】
: 这个听起来有点，"以小散之短，攻MM之长"，
: 不划算,不划算.
:
: get
: of
|
w******s
发帖数: 16209 | 24 if u have good idea about this , we may talk further. I purchased most of
the feed.
【在 g*****u 的大作中提到】
: 或许以后小散也可以在NYSE或其它venue租个co-lo。
: 如果确实有edge, 村长的资金就可以搞个quant fund. 有装备了就不是小散了。
|
f****e
发帖数: 227 | 25 Yes, where can I purchase real-time data feed? The brokerage one is no good
because it does not allow me to manipulate the data but just viewing them on
screen.
I am sure they mean something if, say all S&P 500, are monitored real-time,
or combined into an indicator.
get
of
【在 g*****u 的大作中提到】
: To be able to do "real-time" analysis, you would need real-time data feed.
: There are ways to continually ping your discount broker web interface to get
: data snapshots. But, I doubt its usefulness would go much beyond a test of
: ideas.
|
f****e
发帖数: 227 | 26 We do not have to beat MM to make a living. You gurus on this board have
advised frogs to stand on the MM's side, not the opposite, for profitable
trading.
【在 w*******o 的大作中提到】
: 这个听起来有点，"以小散之短，攻MM之长"，
: 不划算,不划算.
:
: get
: of
|
f********n
发帖数: 1560 | 27 像高盛和citi trading group的IT部门，都有一大批软硬件装备精良训练有素的正规军
。如果有肉，根本抢不过他们，弄不好被他们当肉抢了。
【在 w*******o 的大作中提到】
: 这个听起来有点，"以小散之短，攻MM之长"，
: 不划算,不划算.
:
: get
: of
|
f****e
发帖数: 227 | 28 I don't understand. Who says studing levelII data is to fight AGAINST those
institutions?
【在 f********n 的大作中提到】
: 像高盛和citi trading group的IT部门，都有一大批软硬件装备精良训练有素的正规军
: 。如果有肉，根本抢不过他们，弄不好被他们当肉抢了。
|
f********n
发帖数: 1560 | 29 if you can come up with something for high frequency trading by studying it,
it's very likely they can too. most of the time, the market is too
efficient to let you find anything
【在 f****e 的大作中提到】
: I don't understand. Who says studing levelII data is to fight AGAINST those
: institutions?
|
a*****e
发帖数: 1717 | 30 top names in HFT have no GS or Citi,
getco, knight, citedal, and a lot more.
it,
【在 f********n 的大作中提到】
: if you can come up with something for high frequency trading by studying it,
: it's very likely they can too. most of the time, the market is too
: efficient to let you find anything
|
|
|
h***o
发帖数: 5030 | 31 比如我开一单，买5000股
只能在L2上显示一会，然后就没了，可是我的单的确还挂在那里，只是L2上不见了
我看到好多单也是显示了一会，交易价根本没碰到它的时候就消失了
如果是这么不准的数据，看它有什么意义啊
还是说fidelity给我的数据是垃圾，大家看到的都是正常的数据？
【在 a*****e 的大作中提到】
: who starts first? does your system use market depth data?
|
f********n
发帖数: 1560 | 32 Citi的自动MM交易量占10%+左右，很多时候也拿自己的钱交易，一天几十到几百万利润
不等。
【在 a*****e 的大作中提到】
: top names in HFT have no GS or Citi,
: getco, knight, citedal, and a lot more.
:
: it,
|
a*****e
发帖数: 1717 | 33 of course, all major banks has prop dept, but most of them are spinning off
due to regulation. and citi is never a big/good player in HF market making.
10%? I will be surprised if it's more than 2%.
【在 f********n 的大作中提到】
: Citi的自动MM交易量占10%+左右，很多时候也拿自己的钱交易，一天几十到几百万利润
: 不等。
|
b**u
发帖数: 1206 | 34 在NYSE或NASDAQ租一套rack并不便宜，但能保证数据的realtime和connection.主要是
你有了rack以后的trading algorithm和performance.
【在 w******s 的大作中提到】
: if u have good idea about this , we may talk further. I purchased most of
: the feed.
|
c******7
发帖数: 1081 | 35 我老跟大牛们唱个反调，支持你一下。
对数据的处理拟合模型化，“猜”功还是靠天生直觉性的，我老见过猜功最牛的是安德
森的一个学生，唯象的功夫没人学得来，也许你也有这个潜质。
呵呵呵
those
【在 f****e 的大作中提到】
: I don't understand. Who says studing levelII data is to fight AGAINST those
: institutions?
|
f********n
发帖数: 1560 | 36 http://www.reuters.com/article/2007/07/02/us-citigroup-automate
买了ATD后确实至少有12%。不能算是纯粹的HF，是做broker的同时捎带着做。最近听说
即使他们这么猛的平台，high frequency trading也不怎么挣钱了。
off
.
【在 a*****e 的大作中提到】
: of course, all major banks has prop dept, but most of them are spinning off
: due to regulation. and citi is never a big/good player in HF market making.
: 10%? I will be surprised if it's more than 2%.
|
