风险与回报

bobcat

For your convenience
Diffusion 发表于 2009-11-21 00:57

Thanks! Part1 is fine, but part2 is broken and cannot be open.

Diffusion

Thanks! Part1 is fine, but part2 is broken and cannot be open.
bobcat 发表于 2009-11-21 02:12

It works for me. Anyway, I just post the link and below.

http://rapidshare.com/files/182791466/0470284889.rar

Diffusion

Probably the most useful info I got from this book is the link to this blog.

http://epchan.blogspot.com/

Diffusion

bobcat,

As mentioned in the book and the blog, illiquid stock will impose high transaction cost especially for momentum strategy. I'm wondering how you handle this problem.

man_student

20# Diffusion


Thanks !

benshou

For your convenience
Diffusion 发表于 2009-11-21 00:57

Thanks for sharing

bobcat

bobcat,

As mentioned in the book and the blog, illiquid stock will impose high transaction cost especially for momentum strategy. I'm wondering how you handle this problem.
Diffusion 发表于 2009-11-21 03:22

Yes, momentum trading is liquidity hungry.
That is why bigger money has much lower return rate,
for the reason that they either have to choose high
liquidity stocks, which have less momentum, or
their trading's time horizon must be much longer.

Very liquid stocks in average do not have enough momentum to trade.
Retail traders have the advantage to trade stocks with relatively low
liquidity for higher return. Of course, for low liquidity stocks,
one should trade fewer share, making certain his trades would not
have too much market impact. This is also consistent with risk
management because low liquidity stock are more volatile.

Of course, big money can follow very long trends, such as turtle traders;
but I don't view long term trend following as momentum play.  
Long term trend following strategy has lower return and much lower
Sharpe Ratio.

bobcat

书的第八章起始句为：

"Quantitative trading gained notoriety in the summer of 2007
when some enormous hedge funds run by some of the most
reputable money managers rung up losses measured in billions
in just a few days (though some had recovered by the end of
the month)."

我想作者说的是8/8/2007后几天。有几个大HFs大亏，有的截至8底亏损仍达30%。
书中说小交易资金却没事。
MIT 的 A Lo 写过专文分析当时的情况。大致说很多HFs用类似的程序，造成突然的流通量崩塌。

我的账户事件发生前处于新高点，8/8/2007 至 8/10/2007 损失 2.8%，为当年至当时的第二大
回退，但至8/22/07， 账户已回到新高。在流通量崩塌时，小钱的机动性是大钱无法企及的。

bobcat

大致读了一下这本书
大多数观点与我的一致。不一致处以后我会解释。
书中有这样一句话：
“I said before that one of
the ways momentum is generated is the slow DIFFUSION of information.”
我前几年在其他论坛上一直特别强调这一点。这可以说是动量交易高回报
低风险的关键所在。

padme

书的第八章起始句为：

"Quantitative trading gained notoriety in the summer of 2007
when some enormous hedge funds run by some of the most
reputable money managers rung up losses measured in billio ...
bobcat 发表于 2009-11-21 14:42

最高drawdown只有2.8%？  太牛了。

bobcat

最高drawdown只有2.8%？  太牛了。
padme 发表于 2009-11-21 16:27

最高的是 4.5%， 2.8% 是第二高的。

bobcat

我每天要做很多交易，低最大回退率是大数定律的结果。
GS 也许每天做几百万个交易，所以一个季度只有一天亏损，回退应极小。

padme

我每天要做很多交易，低最大回退率是大数定律的结果。
GS 也许每天做几百万个交易，所以一个季度只有一天亏损，回退应极小。
bobcat 发表于 2009-11-21 16:45

不完全是。如果完全按照大数定律，每天的盈利也应该很小。而且，盈利或亏损的天数，总还是应该持平的。一个季度只有一天亏损，还是超出了常规的。

bobcat

不完全是。如果完全按照大数定律，每天的盈利也应该很小。而且，盈利或亏损的天数，总还是应该持平的。一个季度只有一天亏损，还是超出了常规的。
padme 发表于 2009-11-21 17:22

不对吧：） 为什么应该持平？ 这些方法都有正的回报期望，每天做大量的交易，按大数定律，每天的回报十分接近回报期望，
所以很难是负的。

padme

不对吧：） 为什么应该持平？ 这些方法都有正的回报期望，每天做大量的交易，按大数定律，每天的回报十分接近回报期望，
所以很难是负的。
bobcat 发表于 2009-11-21 17:29

OK， got you.  这些方法都有正的回报期望，那自然是应该回报也为正。

我同时在看elitetrader里面的一个thread, 讲的是抛硬币系统，这个系统的期望是零。呵呵，不知不觉就把那边的讨论和这里的讨论混在一起了。

padme

其实抛硬币系统那个话题也和风险与回报有关系

Whisky
        10-21-09 04:57 PM

The first conclusion is that a random strategy with costs or without them does not cut it in the long term for the purpose of making money and keeping it.

The second conclusion is that a minor positive edge strategy needs a very large account to allow the "long run" to manifest the edge without going bust in the short term, as the runs of "good luck" and "bad luck" happen in reality. The smaller the edge, the bigger the account must be to avoid ruin.

The third conclusion is that all these losers, that lose more than a random strategy would suggest, are doing something to their accounts that is much worse than random betting, and/or the winners of the game are doing something to their accounts and/or the price that traps all the losers in the wrong side time after time, and therein somewhere lie the two biggest edges of them all, as this is a zero sum game minus commissions: Giving liquidity to losers and/or taking liquidity from winners. Liquidity is a volume function that happens in all timeframes.

The first 2 conclusions are easy to prove mathematically. Maybe someone can do it and offer the proof here.

The third...well...it's not so easy to prove.

Diffusion

Yes, momentum trading is liquidity hungry.
That is why bigger money has much lower return rate,
for the reason that they either have to choose high
liquidity stocks, which have less momentum, o ...
bobcat 发表于 2009-11-21 11:54

Mainstream momentum is measured in a period 12 month. I think liquidity is one major consideration.

Diffusion

大致读了一下这本书
大多数观点与我的一致。不一致处以后我会解释。
书中有这样一句话：
“I said before that one of
the ways momentum is generated is the slow DIFFUSION of information.”
我前几年在其他 ...
bobcat 发表于 2009-11-21 16:25

That's where my ID comes from.

padme

股市的特点之一是它有所谓肥尾现象，从历史数据很难估计肥尾现象带来的风险。我们必须有某种不依赖于历史数据统计的保护。
bobcat 发表于 2009-11-19 20:00

股市的肥尾现象是指什么？是不是说，每次都跟以前不一样，总是在不断创造历史记录？

Diffusion

股市的肥尾现象是指什么？是不是说，每次都跟以前不一样，总是在不断创造历史记录？
padme 发表于 2009-11-21 18:45

"Fat tail"是针对概率分布而言。Normal distribution的tail很小，离mean越远的时间概率越小，超过3 sigma基本上就可以认为不会发生。而股市不是这样，超过3 sigma的时间发生的概率要大得多。