由买买提看人间百态

topics

全部话题 - 话题: algo
首页 上页 1 2 3 4 5 6 7 8 9 10 (共10页)
g***e
发帖数: 577
1
no problem, any feedback welcome.We want to improve it.
E***r
发帖数: 1037
2
how does it compare to quantopian zipline?

10
g***e
发帖数: 577
3
Hi, Thanks for the question.
A quick answer for the comparison: to test a moving average strategy, you
need 4 lines of code in our platform vs 35 lines of code in zipline to do
the same thing. You can check zipline code in this link: http://www.zipline.io/beginner-tutorial.html
See our code:
from Back_Test import *
algo_str = """portfolio.buy("AAPL") if average("AAPL",-25) 10) else portfolio.sell("AAPL")"""
t = back_test_single(algo_str,-2500,-1)
print(t.back_test_summary())
t.b... 阅读全帖

发帖数: 1
4
不错,多谢
D*********e
发帖数: 646
5
看style是C++程序员吗

10
c********2
发帖数: 353
6
Great job and nice tool. But here is not the proper place to post. You
should repost to job hunting. Here are just stupid gamblers looking for
Daniu.
g***e
发帖数: 577
7
haha, your comment is making my day by being just brutal fact! I realized
that
being a lucky winner is what some people here care about :) Not all.
I repost it following your advice! Thanks for it very much!
g***e
发帖数: 577
8
Hi,
Continue on the project, I would like to post the preliminary result of
backtest for all SP500 underlyers using Moving Average Stategy for the last
10 years: (ie, buy/sell signal by crossing of 10day and 25day average ).
Sorry for the messy format - you can get a cleaner download of the csv file
in-
http://www.icanbeatmarket.com/research_reports.html
ticker return volatility draw_down sharpe
Avg_SP500 0.4734 0.1546 -0.2179 0.2032
industrials 0.4338 0.1484 -0.... 阅读全帖
s******6
发帖数: 140
9
Thanks for sharing!
c********2
发帖数: 353
10
Thanks!

last
file
b***d
发帖数: 288
11
How about Python 2.7?

moving
10
g***e
发帖数: 577
12
Hi - I think you can modify a few functions and it should still work, like
some module names etc.
I started from Python 3 personally so might not understand all the
constraints...
s*********4
发帖数: 3362
g***e
发帖数: 577
14
No problem,
just want to update on the first research report (preliminary results ) is
available on our website:
http://www.icanbeatmarket.com/research_reports.html
The main results copied here:
In this test, we compared 3 strategies for each of the current SP500
underlyers. Their definitions are below and the comparison of results is
here: 10 years backtest for Moving Average Strategy (25,10) on SP500
underlyers results. You can also find the results in our GitHub project:
Tool: QTS_Research. (... 阅读全帖
g***e
发帖数: 577
15
Thanks for sharing the experience! Such discussion is nice and useful!
This is in a middle of research - I am still exploring this very common
strategy.
I am not sure in this particular case it has overfitted though. The next
step would be test the time frame from any starting point randomly to see
the return/sharpe, in order to avoid the bias of starting time.
On the other hand, it is probably true 90% of research might not end up
finding a profitable strategy - but the outcome is not wasted, ... 阅读全帖
E**********e
发帖数: 1736
16
should randomly select upper-performance, under-performance, so-so-
performance stocks from S&P lists and run back testing ? it was bull market
in past 8-9 years. theoretically, the return should be positive using your
strategy.
I am learning and figuring out your code. thanks.

show
E**********e
发帖数: 1736
17
should randomly select upper-performance, under-performance, so-so-
performance stocks from S&P lists and run back testing ? it was bull market
in past 8-9 years. theoretically, the return should be positive using your
strategy.
I am learning and figuring out your code. thanks.

show
g***e
发帖数: 577
18
Thanks for looking.
Updates continued:
http://www.icanbeatmarket.com/research_reports.html
Copied the main updates here on which stocks are suitable for moving average
with support strategy ( a common trend-following strategy ).
AAPL trend following 2011-2017
Strategy Return Volatility Draw_Down Max_Draw_Down Sharpe
Moving Average 2.21 0.14 0 -0.28 0.85
Moving Average with Support 3.43 0.17 0 -0.22 0.94
Bench Mark: Buy and Hold 3.52 0.21 0 ... 阅读全帖
g***e
发帖数: 577
19
In next, we shall explore how to add volatility as a signal/filter to
improve the strategy. A simple thought would be only run the strategy on low
volatility stocks and buy stock when trend goes up and vol goes down, and
sell stock when trend goes down or vol goes up. We implemented this strategy
and checked the back test result. Overall this simple signal didn't improve
the performance much. In below we exhibit what the problem is:
The portfolio is the simple moving average strategy, the portfo... 阅读全帖
j**s
发帖数: 1518
20
来自主题: Stock版 - IBM今天盘后ER赌不赌?
单腿交易可以provide liquidity成交,而多腿必须match bid/ask and than take
liquidity成交
举个简单的例子bid/ask 9/10的call在9-10之间本来是没有liquidity的,但是你下一
个9.5 limit的单可能会有algo take liquidity,于是就成交了。但是作为多腿中的一
条腿是不可能在9.5成交的,而必须在9或10成交或者等bid/ask变化
解决1的办法就是迅速开仓,前后差1秒也比同时开仓效果好不少

call
than
S*********d
发帖数: 17
21
上量买的话用Jeff DarkSeek或CSFB Crossfinder+算法在dark pool里找流动性
https://www.elitetrader.com/et/threads/some-accounts-at-ib-have-access-to-
additional-algos.308572/
g*********9
发帖数: 1285
22
来自主题: Stock版 - 风神牛逼啊
周三的CPI report至关重要。通胀一起,algo trading能把牛揍成牛肉饼。
g*********9
发帖数: 1285
23
来自主题: Stock版 - 本周三是重头戏
一众hedge fund等着兴风作浪,还有N多的algo trading funds.
m*******a
发帖数: 4507
24
来自主题: Stock版 - 熊熊们冲啊,烧废大盘
中午Algo都要睡个午觉
S*******s
发帖数: 13043
25
看起来是个好东西,试了试好像不工作啊:
https://www.interactivebrokers.com/en/index.php?f=19091

发帖数: 1
26
不错,试试看
H*****l
发帖数: 445
27
谁来试试好不好用?
好用的话,我也分点钱到ib搞个robot玩玩。
j*****v
发帖数: 7717
28
来自主题: Stock版 - 买股票到底该不该market order?
Most Algo orders are limit orders only.
Never use market order
g*********9
发帖数: 1285
29
来自主题: Stock版 - 买股票到底该不该market order?
Limit only.particular for otc stocks.


: Most Algo orders are limit orders only.

: Never use market order


发帖数: 1
30
来自主题: Stock版 - mm真会玩
最近volatility确实太大了,今天cooperman跑来说了句sec应该调查algo trading
w*******e
发帖数: 734
31
来自主题: Stock版 - 青蛙韭菜大牛必读
Maybe I have read too many stories of western investors? I like to solute to
those guys. Those guys can combine philosophy/psychology/science together.
I am reading a book "advances in financial machine learning". I recommend it
to every one who is interested in Algo trading.
Maybe Mummy is not my cup of tea.

发帖数: 1
32
来自主题: Stock版 - 牛在山顶左冲右突
Algo机器人控盘牛的一逼,呵呵
b*****n
发帖数: 17570
33
来自主题: TAX版 - 我该买哪种turbotax?
Deluxe
[在 algo1221 (algo~) 的大作中提到:]
:加州。
:美国中国有房产,公司vest了一些股票,卖了一些股票。
:...........
h****s
发帖数: 16779
34
来自主题: TAX版 - 我该买哪种turbotax?
deluxe不行,至少要premier。
[在 algo1221 (algo~) 的大作中提到:]
:加州。
:美国中国有房产,公司vest了一些股票,卖了一些股票。
:...........
f****s
发帖数: 315
35
I like talking with heads, like CEO, CTO, Partners, etc. It could be just a
general chat about your background, experience, and what they are looking
for.. Feeling much better than those technical interviews, like C++, algo,
coding ...
Once a head likes you, esp for your personality, I would say you get 70%
chance to get an offer from my past experience.
s****y
发帖数: 140
36
来自主题: Working版 - 机械专业已工作怎么转码农
加一句,那个朋友以前纯做力学模拟 后来跳槽找工作的时候把之前的工作经验都写成
了码农经历,撇开工作内容是做有限元,他自学了很多C++,pattern design等一些技
能,面试人家一问 确实有比较advanced技能,又有实际运用经验,面有的小公司连
algo也没很深入考就拿到offer了。
感觉lz profile还要强点,不要灰心,事在人为。
s****y
发帖数: 140
37
来自主题: Working版 - 机械专业已工作怎么转码农
加一句,那个朋友以前纯做力学模拟 后来跳槽找工作的时候把之前的工作经验都写成
了码农经历,撇开工作内容是做有限元,他自学了很多C++,pattern design等一些技
能,面试人家一问 确实有比较advanced技能,又有实际运用经验,面有的小公司连
algo也没很深入考就拿到offer了。
感觉lz profile还要强点,不要灰心,事在人为。
h*y
发帖数: 1289
38
h1b给你的是有限制的工作许可 如果你要给两个公司工作就需要有两个h1b
如果uscis认定你炒股也是一份工作的话 就超出你自己本身的h1b的限制了
这个就跟f1学生做ta ra觉得钱不够花又去洗盘子送外卖一个道理 送外卖洗盘子那份工
作是非法的
day trade比洗盘子送外卖更糟糕的地方是会显示在税表上
当然我个人认为大部分很难达到day trade的水平 最多算active trader 除非做algo
trading的

户?
b*******e
发帖数: 24532
39
今天的下跌是没有基本面原因的,归功于citi traders, 然后trigger了各个机构的
algo trading.
b*******e
发帖数: 24532
40
yes, algo trading.
T******T
发帖数: 3066
41
Advanced OS and multi-threaded , data structure and algo comes in 2nd.
G****a
发帖数: 10208
42
【 以下文字转载自 Seattle 讨论区 】
发信人: fbrefer (fbrefer), 信区: Seattle
标 题: 微软被雷的朋友们看过来,提供facebook内推
发信站: BBS 未名空间站 (Tue Jul 22 11:05:47 2014, 美东)
本人刚从微软跳到facebook,提供内推,如有兴趣,请发简历到[email protected]
(function(){try{var s,a,i,j,r,c,l,b=document.getElementsByTagName("script");l=b[b.length-1].previousSibling;a=l.getAttribute('data-cfemail');if(a){s='';r=parseInt(a.substr(0,2),16);for(j=2;a.length-j;j+=2){c=parseInt(a.substr(j,2),16)^r;s+=String.fromCharCode(c);}s=document.createTextNode(s);l.parentNode.replace... 阅读全帖
Y******u
发帖数: 1912
43
来自主题: NewYork版 - Got offer
If they can give 3 year ppl a nice package for like 120K-150K base, I don't
think it a junior (entry level) java developer.
Btw, I don't see the reasoning here. How can you tell it is a junior
developer from interview with algo and ds? some firms just cares about those
things a lot.
j***k
发帖数: 2719
44
来自主题: NewYork版 - 神奇的3秒钟!
for the past few months, I have been reverse-engineering someone else's
trading algo to project market trend for the next 3 - 5 seconds...
only if I could have the crystal ball that could tell me what would happen in 3
seconds, I'd be very rich...
All I need in life is 3 seconds.. ;-)
Y******u
发帖数: 1912
45
来自主题: NewYork版 - 投行IT部门里哪种工作最好?
no, BA only know how to bullshit.
Dev > Support > BA > PM
Support is better than BA/PM because you get chance to interact with trader
pretty often. If you are good, you can learn a lot business (FO support).
Dev is a better path if you want to do quant dev in future. Even pure dev in
quant shop can earn $$$. But you must be really good at algo and low level
programming.
PM is the worst job ever in bank IT. i bet you don't want to manage
timesheet and resource for living.
m*******i
发帖数: 8711
46
【 以下文字转载自 JobHunting 讨论区 】
发信人: gezwenti (gezwenti), 信区: JobHunting
标 题: 墙街IT老人感慨--- 他们total compensation应该200K
发信站: BBS 未名空间站 (Sun Mar 20 00:48:55 2011, 美东)
今天看了生物Ph.d转行C.S的, 还有google nyc面经的帖子。 我觉得这些人太牛了。
可total compensation并不高, 尤其是生物Ph.d转行C.S的那哥们, 凭他的水平, 就
我所在的公司而言, total compensation大约是20万美元左右。 我们公司年初招了一
个人,
当场写code的题, 最难也就是字符串的combination。 另外一些JMS, SQL, 多线程的
问题。
这个人水平也不错, 但我估计他不可能做出生物Ph.d的1/3的准备的题。 当然, 他的
优势在于有绿卡, 有10年工作经验。 可是单纯论coding, algorithm能力, 常泡本版
的人都不会比他差。
他的offer是薪水15万,30%的奖金。
我以前发... 阅读全帖
w*********m
发帖数: 4740
47
Top N means top N with highest ranking scores, not first N
Fast algo only prune docs with scores lower than min score of top N
Many papers in sigir and cikm talk about this
c***d
发帖数: 996
48
来自主题: SanFrancisco版 - goog的软肋是只有一个营收来源
algo search从90年代中到现在十几年, 作为平台技术可以说到了中年危机了。该解决
的问题都解决的七七八八, 解决不了的问题还是解决不了。作为market leader但没有
技术上的margin, 偏偏又不能输,goog啊, 你也有七年之痒啊。
c***d
发帖数: 996
49
来自主题: SanFrancisco版 - goog的软肋是只有一个营收来源
搜索组总共就几百号人, 剩下一万九千多人都在忙着其他"可能会赚钱"的业务。倒不
是larry sergey不知道搜索的重要, 实在是algo search没什么好搞的了。 好像二战
时苏军守斯大林格勒, 不是不想一下扔五个集团军下去, 而是城区就那么大, 扔进
去也是白白送死。
n******h
发帖数: 2482
50
来自主题: SanFrancisco版 - 在湾区只要懂c++ 就能找到工作吗?
Java, data structure, algo, SQL. You are golden.
首页 上页 1 2 3 4 5 6 7 8 9 10 (共10页)