读书笔记：Inside the House of Money - Christian Siva-Jothy
2016-05-06 伽玛交易员
首先回答一个读者的问题。
“由于现在期权做市越来越趋进全自动，我想请问下期权交易员在做市中被机器完全取代的可能性大吗？目前尚未被取代的主要原因是什么？一个做市老手相比一个刚入行一年的新手，他的优势与价值又体现在哪方面呢？”
目前的场内期权在世界范围内基本都是机器做市，不同是银行叫法不一样，有的叫做Listed Option Market Making，有的叫做Automated Market Making，有的叫做Electronic Volatility Trading，当然背后算法还是人来设计和维护，计算机目前暂时没有自主学习的能力，这个市场的盈利能力基本和场内期权交易量成正比，和市场参与者的成熟程度成反比，每家银行基本都有这个交易团队。这位读者说得期权做市交易员应该指的是OTC市场的做市。现在集中清算和通过交易所场内期权交易越来越普遍（规避交易对手的信用风险），但是场外期权仍然是人来做，原因有几点：第一，场外市场的基本功能是能交易任何种类的期权，其中包括各种定制化的结构和一些结构化产品，机器暂时不能处理复杂的结构和产品（如何合理定价，如何有效对冲），有一些银行的电子交易平台已经能提供各种简单期权组合的报价和交易，另外有部分银行已经开始尝试针对私人银行客户的结构化产品的电子交易平台报价和交易，也许在将来客户可以不用通过销售的人工报价，而是在电子平台上交易任何的产品结构（和目前股票和外汇现货电子交易类似），只留下少部分交易员来处理一些偏好和人打交道的客户以及对电子交易平台上出现的净风险头寸进行对冲；第二，场外市场能够交易的面额是远远超过场内期权市场某一时间段的bid/ask上的流动性，跟现在股票的block trading类似，很多客户不喜欢自己守在电脑前慢慢执行买卖期权，他们希望通过银行报价，一次性的获得整体的报价，这部分目前是需要人来处理的，但是因为整体报价涉及到承担风险，（和之前人工智能能否进行投资的讨论类似）只要机器不能给短期到中期的市场进行判断，我们还是需要人的智慧来觉得如何报价和如何对冲；第三，在目前的硬件和软件框架下，在任何一个金融职业上，机器始终是一个工具，只不过人的角色是在进化的，很难说机器会完全替代人，比如80年代任何股票和外汇都是通过人在市场上下单执行，现在不需要这样的执行者了，但是我们需要人设计越来越好的算法执行这些交易，所以机器能否完全取代人类交易员可能是一个伪命题。
在海外市场，和很多产品的做市类似，外资投行的场外期权做市本身是不赚钱的，期权交易的来源来自于其他地方，不赚钱的原因无非是整体交易量萎缩但是退出市场的投行不多（大部分银行还是想保持full service investment bank的地位，最多退出一些复杂衍生品交易），交易对手单一，一般都是对冲基金，市场拥挤和同质化的问题非常严重，而且在成熟市场本来利润就很低，在做市这块业务里也就排名靠前的一流投行赚钱，做市老手本身需要很强的风险承受能力，通过一定的市场判断或者自营交易赚钱。另一方面，在一个不完全透明的市场做市，正如之前Jim Leitner所说的，你需要认识尽可能多的市场参与者（比如竞争对手，inter-dealer broker，公司客户，基金），才能知道市场的情绪和目前交易的动态，才给出最具竞争力的报价。
今天我们来回顾一下Inside the House of Money里面对前高盛全球宏观自营交易主管Christian Siva-Jothy的访谈。
Christian Siva-Jothy是苏格兰人，80年代在LSE读计量经济学的硕士时为了免费吃喝（！）参加了花旗在校园的宣讲会，机缘巧合加入了投行做交易。在花旗的前两年他做的是英镑远期的market making，之后两年做英镑利率自营交易（80年代以所罗门兄弟为代表的自营交易经营模式开始入侵华尔街）。Siva-Jothy于1992年加入高盛伦敦办公室，成为高盛的一个明星交易员，他于1998年升任合伙人，成为全球宏观自营交易主管，当时他才32岁。他在2005年离开高盛，成立自己的对冲基金，当时成功募资15亿美元，但是开局不利，开始交易的第一年基金就亏损17%，投资者在锁定期结束后赎回了将近10亿美元。虽然在2008年金融海啸里基金回报超50%，但是在之后的三年连年亏损，基金在2011年关门。
一个有趣的八卦是，他在2011年挂牌出售位于苏格兰西南部海滨的国王岛。该岛1930年代的所有者是末代皇帝溥仪的英语老师Sir Reginald Johnston，当时岛上修建有佛教寺庙，中式庭院里悬挂满洲国国旗。该岛面积将近1平方公里，岛上通水通电，有网络和电视，有直升飞机停机坪和一个四居室的独立屋，距离陆地不过三百米，2014年Siva-Jothy给出的最低的售价为250万英镑，但是2015年他把offer撤下了市场。目前他自己住在岛上。
投行的自营交易台在2008年之前像一个磁铁一般让各路豪杰趋之若骛，2008年DB的自营交易员Christian Bittar拿到的1亿欧元的奖金震惊了整个卖方的交易员圈子。但是在2008年后因为监管的变化，自营交易变成了一个在公开场合不能提及的业务，更因为2012年JPMorgan CIO的巨额损失，自营交易被监管机构和投资圈联合绞杀，各家卖方投资银行的自营交易员纷纷出走成立自己基金，现在只有少数不受美国监管的银行还保留有纯粹意义上的自营交易台。我们在这篇访谈里可以一窥最负盛名的高盛自营交易台究竟有何高明之处。
Siva-Jothy下面提到的高盛的risk appetite直到今天都令人敬仰，他们现在的VAR始终都是各家投行里面最高的，但是盈利也始终是最好的，1999年之前高盛自营员工是拿合伙人的钱来赌，高盛对于员工的信任和鼓励尝试的文化确实独树一帜。在这种文化里面如何选择优秀的交易员，如何控制回撤，如何保证不出现道德风险和流氓交易员，应该是高盛机制的精髓。
Goldman Sachs was very, very different from Citibank. Citibank had a risk appetite, but nothing like Goldman. At Citibank, if you made 15 bucks (US$15 million), you were a hero. At GS, to be noticed your P&L had to be a lot higher.
…
Goldman had a risk appetite that took me a little bit of time to get used to. It was a bit of a free-for-all at GS in the early 1990s, but the opportunities were there. There were no limit structures per se, no value at risk (VAR) system. It was just kind of get on and do it and hope for the best.
...
There was no single mentor—just the openness of the environment was probably the biggest mentor for me. The attitude at GS was “Suck it and see,” “If you want to try a big position, try it,” and “Make mistakes, but learn from them.”That was the culture and the philosophy on the trading side of the firm, which I thought was incredibly powerful.
Siva-Jothy谈到了1994年的一个经典战役，他的1bio long GBPJPY头寸。根据通胀调整的话，1994年的1bio GBP大约相当于现在2bio GBP，2bio GBPJPY用现在的卖方交易台的标准来看是一个非常巨大的头寸，可能大部分交易台最多只允许1/30大小，而他所经历的10%的回撤在外汇里面是一个相当大的波动了，因为头寸过大而不能及时止损从一方面说明他可能对于市场底部的判断过于自信，另一方面可能也说明他没有意识到自己的头寸对于当时市场的流动性来说已经过大了，更何况他当时已经是一个非常有名的交易员了，市场肯定会讨论他的头寸（1998年LTCM事件里面可以说是各家卖方瓜分了LTCM带血的筹码，而2012年伦敦鲸案例里面也是少数几家credit hedge fund瓜分了JPM CIO带血的筹码）。他最后清仓时的心理活动可能很多经历过惨痛损失的交易员都会有共鸣。更令人惊讶的是高盛对于他的信任，经历过1994年的损失之后（当时因为联储意外加快升息步伐，高盛在其它FICC业务也有一些损失），高盛让他领导重组后的欧洲宏观自营业务，这可能是基于他之前长期的优秀表现以及最后止损的坚决？
I’d had a phenomenal year in 1993—I made well over $100 million and I was a bit full of myself—and 1992 had also been a great year. Going into 1994, I’d built a very large sterling/yen position. It was over £1 billion, which was the biggest position I’d ever run. I had put it on the November of the previous year and was up about $30 million on the position. People were starting to say, “Christian is going to do it again,” and it had only moved a few percent. At this point, I had become so confident that besides being long sterling/yen via a mixture of cash and options, I was also selling sterling/yen puts whenever the cross corrected.
Then during the first weekend in February, President Clinton came out and attacked the Japanese on trade policy. He threatened them with tariffs and quotas on cars and car parts.The United States gave the Japanese a real hard time and said,“Look, your currency is going to have to appreciate, or else.”
I remember coming in on the Monday morning and dollar/yen had gapped lower 5 percent with nothing trading. If that wasn’t bad enough, five days later, on Friday, UK inflation numbers were released, and they were simply awful. Sterling went into a free fall. It was classic—the market found me.
The long and short of it is the position dropped about 10 percent in the space of a few weeks. I was selling out of the position as fast as I could but I was selling just to stand still because I was short these puts. It was a disaster. Markets have a great way of taking it out of you.
...
On day eight of this episode, when the biggest move happened and I lost about $40 million in one day, I remember feeling this overwhelming desire to get up and walk out, to pretend it wasn’t happening. It was such a powerful emotion. Instead of walking out though, I took a deep breath and liquidated everything.
The firm was just amazing about it. One of the co-heads of the division happened to be in the London office at the time. I went into his office and he said,“Christian, sit down, what’s this all about?”
I said, “I’ve lost X, I’ve liquidated everything.What do you want me to do?”
He looked at me and said, “If you hadn’t liquidated and come in here, you wouldn’t be working at Goldman Sachs anymore.What I want you to do now is go out and make it back, with lower risk limits.”
Siva-Jothy说最大的教训之一是永远不要short gamma，这个在其它三四个PM的访谈里面也出现了，没有short gamma的portfolio在压力测试里面的确不会出问题，但是在2009~2013年三轮QE产生的implied/realized vol历史极低点里面，很多老牌的对冲基金因为long vol bias而损失惨重。市场的规律是动态变化的，PM必须以灵活的心态对待市场的发展。
Also, you should never be short gamma. I’ve become very anti any kind of short gamma trade. That includes carry trades, which in my mind are essentially synthetic short gamma trades. I’m a great believer that portfolios in the macro space should always be long gamma, not short. Never, ever be short gamma.
另一个经典战役是911发生当天，Siva-Jothy对事件性质认定之快，这和当时市场反应之慢形成了强烈对比。在十五年后的今天，虽然基于自然语言处理的算法交易已经开始在大部分事件上成为最先推动金融资产价格变动的因素，但是就笔者观测，机器未必能对一个事件的中期影响有准确的预测，所以如果一个交易员最擅长的时间窗口不是日内的话，相对于机器他还是有优势的。
On the morning of September 11, I was already long U.S. fixed income as I had a structural view that the U.S. economy was weak. It was a decent-sized position as I had been having a reasonable year. I was on the Goldman trading floor in London and I remember hearing someone say,“A plane’s crashed into the side of the World Trade Center.” The first thing I noticed on the TV was that it was a perfectly clear blue sky day. I’m a helicopter pilot and I’ve been flying for 14 years. I know that when you’ve got a plane that’s going down, you don’t aim for the tallest building to fly into; exactly the opposite, you go into the river or you go for a flat piece of land.
I immediately thought, “Terrorist act.” I figured this was going to whack consumer sentiment, which was the only thing keeping the United States afloat at that point. I bought Eurodollars and calls on Eurodollars after the first plane hit but before the second. Strangely, I think they rallied no more than 13 basis points on the day. Markets can be unbelievably slow to figure out the consequences of big events.
...
Maybe, but as I said, markets were very slow to react. It’s one of my fascinations with markets—that they are unbelievably slow to react to big events that there is no script for. This was a classic example, but one that I don’t really like using, for obvious reasons. But let’s take that event. If the attacks had been in Jakarta or London, the impact on markets would have been much smaller. The United Kingdom had already been through a major terrorist bombing campaign and had grown partially immune to it. The situation in the United States was very different; there had never been a major foreign terrorist attack on U.S. soil so the nuance to the trade was not just economic, it was cultural as well. All these things create the fabric of what we do and add another layer that makes the markets so fascinating and interesting. [Note:This interview was conducted several months prior to the London tube attacks, which had little impact on markets.]
对于利用历史数据预测未来市场的发展（比如技术分析里面的移轴），Siva-Jothy是持怀疑态度的，因为如果只是机械的尝试去拟合，而不是追究深层次的原因，这样的分析是不会具有任何预测力，同时数据里面缺失了人的情绪因素，只要目前投资决策还是人做出的，缺失了心理因素的模型是不完整的。
People spend far too much time saying things like, “This is what happened to the market in 1987 and this looks very similar so I’ll put the same trade on,” or “This is what happened in 1994 when the tightening cycle started; therefore the same thing’s going to happen this time.” I don’t like that approach. Markets are dynamic and people’s reactions are different. It’s much more subtle and nuanced than looking at what happened the last time.
...
I was fortunate enough to work at GS with a prop guy who’s been trading for well over 30 years, Bennett Grau. His first day in the office was the day Bretton Woods collapsed. We’ve got data and charts to look at, but he’s seen and felt the emotion associated with those events, and you can’t get that from the textbooks.
So it’s a mix. You can take a snapshot of any one single event but it excludes a lot of things, so it’s valuable to study history around that event. You’ve got to take everything in the right context, and that’s the difficult bit.
什么样的人会成为一个好的自营交易员以致成为交易主管，Siva-Jothy给出的答案可能会让一些人有些惊讶──谦逊，正直，还有对市场（而不是钱）的热情。记得笔者在2006年开始面试的时候，大部分人给我的答案是：聪明，对钱的渴望，以及勤奋。
Passion and humility are the main qualities I look for but first and foremost—it sounds a bit cheesy—I ask myself, is this a good person? Is it someone I want to sit next to that I trust and want to work with? Integrity is the single most important thing to me in hiring.
...
A lot of this is luck, but humility is probably the most important thing. If you’re not humble, you’re not going to last very long. You also have to enjoy it.Too many people come into this business for the money, and that’s not going to work out over time.
谈到自营交易员的奖金结构，Siva-Jothy说高盛不会给自营交易员一个分成的百分比（这个跟Christian Bittar和DB之间的合同不一样），因为在高盛的盈利模式里，自营交易台在牛市里可以赚的少一点，但在熊市里（当其他业务不赚钱的时候），自营台应该赚大钱来减少高盛作为一个整体的利润的volatility，如果有一个百分比的分成，在心理上会影响交易员努力往尽可能高的pnl奋斗。
In terms of differences, banks, contrary to popular belief, don’t have set payouts because it affects the way traders take risk. When you know exactly what you’re going to get paid, it takes away from one’s ability to push the envelope. The tendency is to think in terms of certainties. The thought process goes like this:“Okay, I’m up $20 million. If I push, I could be up $40 million or I could be flat, but right now I know that X percent of this$20 million is worth Y dollars in my pocket.”
Psychologically, it’s very rational. At GS, they wanted us to push the envelope in dislocated markets so they didn’t give us the certainty of a fixed payout. In fact, their subtle, even subliminal message was, “If you don’t push the envelope in times of stress, we’re going to penalize you.” It was very clever on their part, because it was that pressure that resulted in the outsized returns during periods of large market movements.
最后谈谈笔者自己对于自营交易的理解。在08年之后自营交易成了过街老鼠人人喊打，民众的一般理解是自营交易是少数银行交易员绑架大而不倒的银行，为了最大化自己的奖金而在市场里赌博。这个说法在一些案例里可能是对的，但是客观的理解自营交易业务模式应该从以下几个方面着手：第一，在任何不是单纯execution的交易里，market making和prop trading是很难分开的，因为风险不能立刻完美的平掉，交易员一定要有对市场的看法建立相应的仓位，否则无法赚钱甚至无法报价，当然这里涉及到如何控制一个合理的风险敞口，以及在混业经营的银行里如何保证交易员不会绑架银行的大而不倒的地位来孤注一掷；第二，在高盛这样的没有零售业务的银行里，根据Siva-Jothy的说法，自营交易实际上是一个对于整体盈利的macro hedge，自营交易台的使命就是在牛市里维持生存，在市场出现崩盘的时候需要赚大钱，这也是为什么Siva-Jothy反复说他的交易组合有long vol and long gov bond bias，在上市之前高盛用合伙人的钱来作自营交易没有任何问题，上市之后存在绑架中小股东的问题，这个让自营交易台处境尴尬；第三，投行的自营交易实际上给市场提供了中长期的流动性，这起到了降低了市场出现极端波动的概率，高频交易提供的流动性是超短期的流动性，一旦市场出现巨幅波动时（比如2014年10月的美债，以及2015年8月25日当天美股开盘前后），高频交易往往会被关停，在市场最需要流动性的时候它们跑路了，而对冲基金因为盘子小，能承受的回撤小，他们接纳风险的能力有限，而balanace sheet雄厚、融资能力强劲的投行能够在市场里吸纳一些流动性差的风险（比如长期运作一些relative value arb），在市场出现极端情况时也能出来接盘，这也是为什么08年之后自营交易被大规模绞杀之后，市场出现flash crash越来越频繁，而Volcker Rule实施之后，市场整体的流动性在下降。