|
A********a
发帖数: 133 | 2 Robert Almgren and Marco Avellaneda at NYU taught courses on stat arb, basic
stuff, they also wrote couple of paper on the topic, check out their
websites, |
|
c******r
发帖数: 300 | 3 恩，我想说的是和econometrics相关了。
其实1,3还好啦，2的话我不怎么做time series，只知道统计里面的cross-sectional
study 和 panel study,所以也没多问我，不过似乎面我的人觉得我应该很懂这些才对
，然后我就说这种东西catch up应该不难吧，原理都一样。5我就懂一点点，就给他讲
了讲capm本身，不过他具体的问题是问我觉得capm是个theory还是个hypothesis, 赫赫
，这里就不知道他到底想问啥了，我觉得是hypothesis, 不过也不是很清楚为啥，他暗
示econ里面有抓们对这个的讨论，我明显没看过 ...... 另外，我们讨论trading
strategies的时候我知道的一点都是stat arb和market making，似乎也不对他们胃口
...... |
|
l*********t
发帖数: 89 | 4 I believe the Graph of Vol w.r.t Strikes should be convex, otherwise there
should be an arb opportunity. |
|
a********e
发帖数: 508 | 5 It's a reasonable guess. But I'm afriad you can't find arb that easy by just
looking at the implied vol curve
Why not calculate the option price directly to see if the function of
option price is convex w.r.t strike |
|
n****e
发帖数: 629 | 6 把local vol surface construct出来
如果construct不出来就说明有arb... |
|
m********0
发帖数: 2717 | 7 statistical-vol arb easy, quite some common combos
convertible-like put/call parity. |
|
m********0
发帖数: 2717 | 8 没读过这本书，也不熟悉他们的故事，但是如果你对trading比较了解的话，
这种现象人物每个market cycle都会出来，
比如gan zanger从一万一一年多炒到四千多万，很多人也觉得他没什么牛的，
只是无数白天鹅里必然出现的黑天鹅罢了。
11万不叫fund，顶多算retail investor，可能注册了什么小公司，后来做大
了有可能有investment。11万刀到30M，在trending market相对容易。当然也
是要点运气的。
从09年初到现在做options翻了几十上百倍的，我也知道几个。不是很奇怪。
至于street smart代表的discretional的trader好呢，还是讲求技术的
systematic好呢？ 这个很难讲。尽管普遍认为后者要逐渐取代前者。
不过你自己做做options trading就知道了，很多什么pricing不懂的traders
单作directional的一样可以做到很好。
有些top衍生物traders连underlying是什么都不知道，还有bond trader连price和
yield的关系都不知道，纯粹就是个orde... 阅读全帖 |
|
j********k
发帖数: 90 | 9 请问有没有什么好书（关于stat arb 等方面）推荐？
不是为了面试，是想自己研究一下trading strategy。
谢谢。 |
|
p****u
发帖数: 2596 | 10 天啊。今天stat arb叠的一他糊涂。。不会是这拨日本资本在liquid position把，，
，晕死。。。。 |
|
K******r
发帖数: 152 | 11 请问一下，那种contract的roll机制有规定的时间么？Mou present的时候，我就觉得
挺好奇，挺明显的market price
impact为什么不能通过random rolling或者early rolling解决呢？他的paper里似乎提
到说，那个contract总是在mature
之前一个礼拜rolling，为啥不搞成random rolling，这样就没啥arb可以弄了，反正也
没人知道你啥时候roll。 |
|
s******e
发帖数: 1751 | 12 b/c so far it is still fairly efficient (including the rolldown cost). i.e.
not easy to arb it.
if there is an easy money in the process, it was already splitted by those
big brokerage houses. |
|
p***z
发帖数: 132 | 13 我来说说我的quant经历吧，在这几年里既做过buy-side,也做sell-side; 既做过Fixed
-income,也玩过Equity; 既做过前台，也练过后台; 既在academic quant journal发过
paper,也自己做过trade; 既据过人，也被据过；既做过intern, 也做过full-time,
所以还是蛮有意思的.
2006年找summer Intern,拿了JPM QR和CS的offer,据了JPM, 去了CS的summer,因为CS的
intern更像training program. 后来的很多经历也确认了CS的summer intern是街上最
好的，虽然我至今不理解CS为啥要为其他银行搞免费培训.
2007年毕业前决定来industry,老板很理解，找到一个nice的老板是很幸福的。2006年
底投了很多简历，只记得有一个星期连续有4天New York的on-site，除了星期二。周一
面完GS的super day,快虚脱了，晚上7点到La Guardia airport已经登机了准备回
Boston,正准备关手机，收到个电话，说是JPM... 阅读全帖 |
|
p***z
发帖数: 132 | 14 I think he already had enough technical knowledge for a buy-side quant with
that background, assume he is doing fairly well at his research area.
We do volatility stat arb and relative value on almost all asset class. |
|
f*******y
发帖数: 988 | 15 小作坊非常不稳定
干高频要survive只能上量
一上量就变成事实上的market making了
高频最稳定的钱就是mm和latency arb，小公司根本伤不起
而且年内传闻order cancel 要交钱了，有木有有木有 |
|
p***z
发帖数: 132 | 16 我来说说我的quant经历吧，在这几年里既做过buy-side,也做sell-side; 既做过Fixed
-income,也玩过Equity; 既做过前台，也练过后台; 既在academic quant journal发过
paper,也自己做过trade; 既据过人，也被据过；既做过intern, 也做过full-time,
所以还是蛮有意思的.
2006年找summer Intern,拿了JPM QR和CS的offer,据了JPM, 去了CS的summer,因为CS的
intern更像training program. 后来的很多经历也确认了CS的summer intern是街上最
好的，虽然我至今不理解CS为啥要为其他银行搞免费培训.
2007年毕业前决定来industry,老板很理解，找到一个nice的老板是很幸福的。2006年
底投了很多简历，只记得有一个星期连续有4天New York的on-site，除了星期二。周一
面完GS的super day,快虚脱了，晚上7点到La Guardia airport已经登机了准备回
Boston,正准备关手机，收到个电话，说是JPM... 阅读全帖 |
|
p***z
发帖数: 132 | 17 I think he already had enough technical knowledge for a buy-side quant with
that background, assume he is doing fairly well at his research area.
We do volatility stat arb and relative value on almost all asset class. |
|
f*******y
发帖数: 988 | 18 小作坊非常不稳定
干高频要survive只能上量
一上量就变成事实上的market making了
高频最稳定的钱就是mm和latency arb，小公司根本伤不起
而且年内传闻order cancel 要交钱了，有木有有木有 |
|
c*******e
发帖数: 150 | 19 你们那里也要用spreadsheet和add-ins呀，偶以为只有 infrastructure 不强或者数据
量不大的shops才用spreadsheet。 打个比方，偶以前的认识是 在Unix/Linux以外的
OS 下面做cash equity 的 stat arb是不可想象的事情，只有 OTC 一类的trades，
spreadsheets 才比较普遍，一天最多交易不超过两次。
Please correct me if this understanding is not right :)
当代quant行走江湖之必备绝技. |
|
L********a
发帖数: 44 | 20 regulators rules themselves are not 100% clear. there are rooms of arb.
particularly the correlation products |
|
l*********t
发帖数: 89 | 21 你这是static replicate，但是如果dynamic replicate岂不是就没有arb了么？ Black
-Sholes不就是建立在dynamic replicate的基础上么？
并且John Hull的课后题，和有的面试书让price payoff 为 max(S^2-K, 0) 的option
岂不是没有现实意义了么？
还不是完全理解，希望进一步说明，多谢！ |
|
e****r
发帖数: 6 | 22 【 以下文字转载自 USTC 讨论区 】
发信人: elinor (大), 信区: USTC
标 题: junior quant position
发信站: BBS 未名空间站 (Thu May 19 21:19:08 2011, 美东)
A junior quant position focusing on Delta1 projects like index/ETF arb. PhD
preferred and less than 2 years experience. Strong programming (not
necessary to be hard core C++ type, but has to be hands on programmer in
Java, Matlab, etc..)
It is a HK based position and immediate relocation is expected.
If interested, please send resume to r****[email protected]. |
|
Y******u
发帖数: 1912 | 23 Not agree. If u r talking about computer program auto trading strategies(
like stats arb or other HF) in buy side, usually ppl avoid using word algo-
trading but use automatic trading or quantitative trading. As far as I know,
ppl usually mean program auto order routing/execution in sell side/market
marker when they say algo trading. It is more about market micro structure
but not strategies.
Fixed income quant can be either buy side or sell side.
client
pricing |
|
c*******e
发帖数: 150 | 24 one form of volatility arb:
sell implied vol, pay for realized vol
through either vanillas or other instruments, say var swaps. |
|
a***r
发帖数: 594 | 25 I ll take a stab.
a CDS can be bought in two ways. one is you pay a quaterly coupon at a
quoted rate. the other is you pay a cash payment upfront. both in exchange
for the protection payout when the reference entity defaulted.
so the upfront for a 3/5 forward CDS is simply 5y CDS upfront - 3y CDS
upfront. thats dictated by no-arb. depending on when the upfront is paid (
today or in 3 years when the protection starts) you may have to do some
discounting.
now he wants the running. we ll estimate t... 阅读全帖 |
|
J******d
发帖数: 506 | 26 是啊。
LZ要是去读个MFE简直可以堪称行为艺术了。按书上说的，LZ应该比绝大多数faculty都
有资历了。LZ可是自己一个人开发过一个high frequency trading system的啊。谁给
你上C++比较合适呢？谁给你上stat arb比较合适呢？ |
|
h*********n
发帖数: 915 | 27 are they the same things, or hft is a sub-cat of stat-arb?
is two sigma a hft fund? |
|
g***u
发帖数: 22 | 28 可以，只是你得到的时实际分布，市场交易的是风险中性分布，如果在卖方你按此交易
你会在一方找不到对手。在买方你可以称为arb，只是你的推测于市场太远时，你会出
局。
另外，你要描述整个yield曲线，光5年期时不够的，今天5年明天就不时5年了。哪个模
性不重要，在当前和时间上自恰重要，前一点是模型交准问题，后面是动态行为问题，
大部分的模型可以容易的做哈前面，却输于后面。 |
|
z***e
发帖数: 5600 | 29 The rate bank charges could be a lot higher than the 10% you assumed.
Many LBOs failed. The successful LBOs picked up real stable and undervalued
companies and arb'd it. However, in the boom years there were many with
cyclical or declining businesses and blew up during the downturn
cash |
|
e***z
发帖数: 7126 | 30 只要同类ETF之间可以high frequency arb 没量的ETF也可以很快做上量去
而且人家ishares 啥的跟fidelity 关系不是一般的，再联系几个其他retail broker.
只要鱼养大了，大机构玩儿小机构太容易了。
这种事情能长久的话，那说明人家真的有过人之处，只是不明显。 |
|
s******r
发帖数: 324 | 31 Their stat arb hedge fund is much better. But still not good as two sigma,
not sure of DE Shaw's performance these years. |
|
l********e
发帖数: 220 | 32 how about worldquant's stat arb fund performance these years? |
|
s******r
发帖数: 324 | 33 2 sigma is not in same business, sigma take 3/30 in their stat arb fund
while Tower 100% is internal money. Their revenue is pretty close to getco
this year as I understand it. |
|
s******r
发帖数: 324 | 34 D.E. Shaw has multiple fund? You mean their core stat arb fund is up 17.5%?
They have CTA, vol, credit,distressed etc. |
|
u****h
发帖数: 2193 | 35 这stat arb和asset management难道mutually exclusive? |
|
s******r
发帖数: 324 | 36 bank's number are all fake anyway because bank's inherent leverage is very
high. But if you want to calculate approximately, for long only, you can
assume you have 1 leverage, for long short depends on the strategy, for
stock picking long short probably assume 150/50, 2 times leverage, for stat
arb, you can assume 8 times leverage. For option book, you can assume 20
times leverage. |
|
c*******e
发帖数: 150 | 37 呵呵, 偶是混迹于quantitative trader中间的fundamental guy，虽然您说greek
letters 啥的偶只懂一点点，不过做到只 take gamma views 不 take vega views
还是不难的。偶们的优势是可以花好几天时间研究一个 fundamental story 而一个
星期只发一招，所以发招的质量还是可以保证的。作为 market 中心的 sell-side
trader，每天有大量的 flow 要处理，还有很多 hedging works，当然处于market
的中心 flow info 是一个优势，但是几分钟做的一个判断 要想和 几天深思熟虑做
的判断 媲美还是会有差别的。
再说了大家也不识什么竞争或者谁 arb 谁的关系，东家挣flow，b/a的钱，偶们挣个
views 的钱，有其它的确有hedging needs 的 II 埋单，谁都开心不是 ^_^
has
freely.
)
book
少是 |
|
s******r
发帖数: 324 | 38 这几年据说 stat arb赚的钱占总的比例高起来了,大概有一半左右(3-4billion a year
). |
|
s******r
发帖数: 324 | 39 但象renaissance这么大的size(monthly expected pnl 500-700MM),一个strategy有个
500MM MDD没啥大不了的,august 2007,他们stat arb drawdown much bigger than
that. |
|
Y******u
发帖数: 1912 | 40 very few quant do real "research" as you think.
at sell-side, most quant are reading paper, implementing the model and
validating result. Coding and spreadsheet are taking most of their time. A
few client-facing quants will come up with some ideas and write reports to
client but if you know the education degree of clients (most are bachelors)
you would know how "quantitative" the report is.
at buy-side, don't get the name wrong. hedge fund is not a rocket science
business. Fundamental is still ... 阅读全帖 |
|
c*******y
发帖数: 1630 | 41 http://www.cboe.com/products/indexopts/vixoptions_spec.aspx
underlying is index not futures on index.
for SPX, you have options on SPX index and options on SPX futures.
for VIX, currently only on VIX index.
nothing wrong with greeks calculation, the issue is due to no synthetic/direct underlying(so no easy arb), greeks is not informative/suggestive in the usual way.
There is no way AAPL 350C is quoted less than 30 bucks when AAPL@400.
but you will see VIX C40 sold at 5 when VIX@50 and VIX C10 at... 阅读全帖 |
|
a*****k
发帖数: 704 | 42 commodity: c'mon.
I immediately found my sentence interesting! Haha
But yes, you just made my day.
direct underlying(so no easy arb), greeks is not informative/suggestive in
the usual way.
15 a lot. (Just an example) |
|
c*******e
发帖数: 150 | 43 Larry Harris 可以用来扫盲/入门教育
如果还在学校里有大量的休闲的时间，Grinold & Khan 的 bible 当然也是打内功基础
的好书
之后看你 trade 什么方面的 asset class/venue 和 trading styles (e.g. stat-arb
, event-driven, trend-following, factor-based alphas, Options AMM, etc) 每一
门武功 (style) 和兵器 (instrument) 都非常地不同，师傅带进门以后修行就考自己
了。实战里面再不断提高，倒也没有什么统一的书本可以参考的 |
|
|
L******g
发帖数: 1371 | 45 【 以下文字转载自 NewYork 讨论区 】
发信人: LinChong (林冲), 信区: NewYork
标 题: 我想找人一起写code
发信站: BBS 未名空间站 (Sat Jul 14 10:24:47 2012, 美东)
大家一起写equity stat Arb，
一个人写risk model
一个人写optimizer,
一个人写regression,
一个人写个AlgoTrader.exe
一年后大家都写好了， 就可以去Mellenium 开个group 了，
需要牺牲每天一个小时，和所有的周末， |
|
q**j
发帖数: 10612 | 46 SPY up 1.84%
but its weighted components up 1.90%
是不是很多short的导致这个现象的？ |
|
z******a
发帖数: 5381 | 47 每天平均有8个component ex div. |
|
q**j
发帖数: 10612 | 48 this should also impact the individual stocks, right? |
|
c**********6
发帖数: 18 | 49 借人气问一个，哪里可以下到SPY components的列表？ |
|
mw
发帖数: 525 | 50 you want to trade such signals as an individual? that should be very hard |
|
