g********k 发帖数: 15 | 1 我有个朋友,他自己写一个program(一个alpha system,不是 momentum).
来预期一个公司的股票,报完earning后的那天,会涨还是会跌.
过去一年record成功机率是70~80%.
比较重要的是,他认为这个模式可以维持,技巧也越来越纯熟了.
现在他想开始找一些投资者,像hedge funds.
你们觉得他目前条件有ready了吗? 那该如何寻找投资者或hedge funds?
还是需要更多track record或其他条件?
track record放在 stocktwits.com,这个可信度够高吗?
(earnings announcement之前会先纪录在 stocktwits, buy or short股票,
这个网站的纪录无法删除或修改.)
这是这一年的成绩,都放在 stocktwits.com
2013 11月~11月中 对13家,错4 家,平均每家 +3.7%,乘17家,因此总供63%获利.
2014 1月中~2月中 对38家,错10家,平均每家 +4.7%,总供225%获利.
2014的中间两个earning季节,程式修正中,没有作交易.
2014 10月中~11月中 对31家,错10家,平均每家 +4.2%,总供 172%获利.
每一家股票都是报完当天就解除,无法盯著股票上上下下, 因为人手不足,
而且马上又要准备隔天的earnings,有时一天就多达两三百家需要分析.
我不是我朋友本人,他是华人,但来美国的时候年纪小,打中文速度很慢.
背景是藤校本科,西部名校mba,后来到几个大公司做IT. | D**D 发帖数: 1832 | 2 3 months of track record isn't gonna cut it.
You need at least two years and thousands of trades to impress any serious
money.
Not trying to discourage, but a computer model needs even longer track
record to show its effectiveness, even then it can still crash the account
two weeks into operation. Curve fitting is dangerous.
★ 发自iPhone App: ChineseWeb 8.6
【在 g********k 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 我有个朋友,他自己写一个program(一个alpha system,不是 momentum). : 来预期一个公司的股票,报完earning后的那天,会涨还是会跌. : 过去一年record成功机率是70~80%. : 比较重要的是,他认为这个模式可以维持,技巧也越来越纯熟了. : 现在他想开始找一些投资者,像hedge funds. : 你们觉得他目前条件有ready了吗? 那该如何寻找投资者或hedge funds? : 还是需要更多track record或其他条件? : track record放在 stocktwits.com,这个可信度够高吗? : (earnings announcement之前会先纪录在 stocktwits, buy or short股票, : 这个网站的纪录无法删除或修改.)
| B*********e 发帖数: 86 | 3 algorithm trading wont be taken serious unless you have a trading record
winning real money
There are many forex gurus out there posting amazing trading performance,
however, anyone following their "strategy" is/will be bleeding
the problem is there are too many ways to create an overfitting/biased
algorithm than you can image
【在 g********k 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 我有个朋友,他自己写一个program(一个alpha system,不是 momentum). : 来预期一个公司的股票,报完earning后的那天,会涨还是会跌. : 过去一年record成功机率是70~80%. : 比较重要的是,他认为这个模式可以维持,技巧也越来越纯熟了. : 现在他想开始找一些投资者,像hedge funds. : 你们觉得他目前条件有ready了吗? 那该如何寻找投资者或hedge funds? : 还是需要更多track record或其他条件? : track record放在 stocktwits.com,这个可信度够高吗? : (earnings announcement之前会先纪录在 stocktwits, buy or short股票, : 这个网站的纪录无法删除或修改.)
| x****o 发帖数: 29677 | 4 为什么自己不做交易?一个月平均翻一倍,一万块滚两年,你还怕没有投资? | g*****0 发帖数: 2998 | | g********k 发帖数: 15 | 6 thank you for your feedback. his system is not algorithmic trading, it's
event-driven trading, or news-driven trading. the future event/news is the
company's earnings announcement. is the company going to announce good
earnings or miss? and how will the stock react?
【在 B*********e 的大作中提到】![](/moin_static193/solenoid/img/up.png) : algorithm trading wont be taken serious unless you have a trading record : winning real money : There are many forex gurus out there posting amazing trading performance, : however, anyone following their "strategy" is/will be bleeding : the problem is there are too many ways to create an overfitting/biased : algorithm than you can image
| W***n 发帖数: 11530 | 7 skeptical on any program tradings | B*********e 发帖数: 86 | 8 According to CNBC, the percentage of S&P 500 companies that beat their
earnings forecast in the first quarter of 2012 was over 70%....
In 2014 More than 81 percent of the 410 S&P 500 members that have reported
results so far this quarter have topped projections, led by companies from
Caterpillar Inc. and Time Warner Inc. to Exxon Mobil Corp. That’s the
highest rate since the first quarter of 2010
......
This means you can gain 80% winning rate with no strategy
【在 g********k 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 我有个朋友,他自己写一个program(一个alpha system,不是 momentum). : 来预期一个公司的股票,报完earning后的那天,会涨还是会跌. : 过去一年record成功机率是70~80%. : 比较重要的是,他认为这个模式可以维持,技巧也越来越纯熟了. : 现在他想开始找一些投资者,像hedge funds. : 你们觉得他目前条件有ready了吗? 那该如何寻找投资者或hedge funds? : 还是需要更多track record或其他条件? : track record放在 stocktwits.com,这个可信度够高吗? : (earnings announcement之前会先纪录在 stocktwits, buy or short股票, : 这个网站的纪录无法删除或修改.)
| D*****T 发帖数: 126 | 9 要至少一年real money trading record. Paper trading 跟real money 感觉很不一样 | g********k 发帖数: 15 | 10 most companies report "good" earnings, but that doesn't mean its stock will
go up after announcing its earnings. a lot of stock go down after reporting
"good" earnings and vice versa. take your example of time warner inc. twx
reported "good" earnings in 8/6/2014, its stock dropped 12%.
【在 B*********e 的大作中提到】![](/moin_static193/solenoid/img/up.png) : According to CNBC, the percentage of S&P 500 companies that beat their : earnings forecast in the first quarter of 2012 was over 70%.... : In 2014 More than 81 percent of the 410 S&P 500 members that have reported : results so far this quarter have topped projections, led by companies from : Caterpillar Inc. and Time Warner Inc. to Exxon Mobil Corp. That’s the : highest rate since the first quarter of 2010 : ...... : This means you can gain 80% winning rate with no strategy
| | | g********k 发帖数: 15 | 11 有real money trading record,跟 stocktwits.com一样,
只是资金不多.
【在 D*****T 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 要至少一年real money trading record. Paper trading 跟real money 感觉很不一样
| j**s 发帖数: 1518 | 12 资金多不多差距很大。你这种交易方法,就算模型正确,资金多了move the market的
成本就很大。
几万的资金可以一年翻几倍,10亿的资金一年30%都不容易
【在 g********k 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 有real money trading record,跟 stocktwits.com一样, : 只是资金不多.
| j**s 发帖数: 1518 | 13 “2013 11月~11月中 对13家,错4 家,平均每家 +3.7%,乘17家,因此总供63%获利”
这句话逻辑就不对,你自己想想为什么? | j**s 发帖数: 1518 | 14 至少有百万资金真钱连续至少三年的记录才有说服力。去年和今年市场好,02 08年你
试试? | B*********e 发帖数: 86 | 15 so his system can predict stock movement after ER with 80% accuracy?
sounds like a event driven data streaming algorithm. If you can maintain
such performance you actually find something that event driven mutual funds
craves for, since most of their average return this year is below 10%
If hes so confident with his system, he should talk to a professional in
real life, even though I serious doubt the existence of such holy grail
Good luck
will
reporting
【在 g********k 的大作中提到】![](/moin_static193/solenoid/img/up.png) : most companies report "good" earnings, but that doesn't mean its stock will : go up after announcing its earnings. a lot of stock go down after reporting : "good" earnings and vice versa. take your example of time warner inc. twx : reported "good" earnings in 8/6/2014, its stock dropped 12%.
| g********k 发帖数: 15 | 16 谢谢.
同样的文章,我也在quant版帮他提问了.
funds
【在 B*********e 的大作中提到】![](/moin_static193/solenoid/img/up.png) : so his system can predict stock movement after ER with 80% accuracy? : sounds like a event driven data streaming algorithm. If you can maintain : such performance you actually find something that event driven mutual funds : craves for, since most of their average return this year is below 10% : If hes so confident with his system, he should talk to a professional in : real life, even though I serious doubt the existence of such holy grail : Good luck : : will : reporting
| i********y 发帖数: 346 | 17 对研究策略的还是应该有所鼓励。到不一定非要百万资金,如果面向的公司股票流动性
较好,几万资金做测试,连续三年,draw down比较低应该就会有基金公司感兴趣了。
当然最近三年市场都比较好,说服力还不一定够,理论上最好是一个经济周期(7-8年)
才更有说服力。当然试验性操作7-8年如果真是好策略浪费的机会也不少,所以一般3,
5年能稳定盈利的话就会有基金感兴趣了。
【在 j**s 的大作中提到】![](/moin_static193/solenoid/img/up.png) : 至少有百万资金真钱连续至少三年的记录才有说服力。去年和今年市场好,02 08年你 : 试试?
| p******e 发帖数: 17163 | 18 赌本是很难找的。绝大部分的基金开始大约两种途径,一种自己已经是金融业混成名,
然后找原来客户拿钱。另一种就是先从自己亲戚朋友募款。结论就是你得认识有钱人。 |
|