High-Frequency Strategies 高频交易策略介绍(译文)_order flow trading for fun and profit

High-Frequency Strategies  高频交易策略介绍(译文)

Most high-frequency momentum strategies involve extracting information from the order book, and the basic idea is simple: If the bid size is much bigger than the ask size, expect the price to tick up and vice versa. This idea is backed by academic research. For example, an approximately linear relationship between the imbalance of bid versus ask sizes and short-term price changes in the Nasdaq market was found (Maslov and Mills, 2001). As expected, the effect is stronger for lower volume stocks. The effect is not limited to just the national best bid offer (NBBO) prices: an imbalance of the entire order book also induces price changes for a stock on the Stockholm stock market (Hellström and Simonsen, 2006).

大多数高频动量交易策略所涉及的是从订单数据之内提取信息，其基本的思路很简单：如果买入价上的挂单量远远大于卖出价的挂单量，就预期价格会小幅上升几个点位，反之亦然。

这个想法有学术研究的支持。举例来说：在纳斯达克市场，我们发现了买盘量与卖盘量之间的不平衡性导致了一个近似线性关系（Maslov, Sergei, and Mark Mills. “Price Fluctuations from the Order Book  Perspective: Empirical Facts and a Simple Model.” Physica A 299, no.  1–2）。正如所预期的那样：买卖价差对小盘股的影响更强烈。还有，这个效应并不局限于全国最佳买价系统（简称NBBO）里面的价格：在斯德哥尔摩股票市场上，股票的挂单量的不平衡性也可以引起价格变化（Hellström, Jörgen, and Ola Simonsen. “Does the Open Limit Order Book Reveal Information About Short-run Stock Price Movements?” Umeå Economic Studies, 2006.）。

There are a number of high-frequency momentum strategies based on this phenomenon. Many of those were described in books about market microstructure or high-frequency trading (Arnuk and Saluzzi, 2012; Durbin, 2010; Harris, 2003; and Sinclair, 2010). (In my descriptions that follow, I focus on making an initial long trade, but, of course, there is a symmetrical opportunity on the short side.)

有许多高频动量交易策略都是基于上述现象而生成的，许多细节在市场微观结构，或高频交易所相关的书籍当中有所叙述（

Arnuk, Sal L., and Joseph C. Saluzzi. Broken Markets: How High Frequency Trading and Predatory Practices on Wall Street Are Destroying Investor Confidence and Your Portfolio. Upper Saddle River, NJ: FT Press, 2012

Durbin, Michael. All About High-Frequency Trading. New York: McGraw-Hill, 2010.Durden, Tyler. “

Harris, Larry. Trading and Exchanges. New York: Oxford University Press, 2003

Sinclair, Euan. Option Trading: Pricing and Volatility Strategies and Techniques.Hoboken, NJ: John Wiley & Sons, 2010.）（在我接下来的描述中，我主要专注于开始进行买多交易，但是对卖空而言，也存在一个对称的机会）。

In markets that fill orders on a prorata basis such as the Eurodollar futures trading on CME, the simplest way to benefit from this expectation is just to “join the bid” immediately, so that whenever there is a fill on the bid side, we will get allocated part of that fill. To ensure that the bid and ask prices are more likely to move higher rather than lower after we are filled, we join the bid only when the original bid size is much larger than the ask size. This is called the ratio trade, because we expect the proportion of the original order to be filled is equal to the ratio between our own order size and the aggregate order size at the bid price. Once the buying pressure causes the bid price to move up one or more ticks, then we can sell at a profit, or we can simply place a sell order at the best ask (if the bid-ask spread is larger than the round trip commission per share). If the bid price doesn’t move up or our sell limit order doesn’t get filled, we can probably still sell at the original best bid price because of the large bid size, with the loss of commissions only.

在按比例填写订单的市场中，例如 CME 的欧洲美元期货交易，从上述预期获利的最简单方式是立即“加入买方”，所以，只要有一个订单成交在买方价格，我们就要占这个成交的一部分。为确保买入价和卖出价更有可能在我们成交后走高而不是走低，我们仅在原始买入量远大于卖出量时加入买方报单。这被称为比率交易。

因为我们期望原始订单被执行的比例等于我们自己的订单大小与买入价的总订单大小之间的比率。 一旦买压导致买价上涨一个或多个点位，那么我们可以卖出获利，或者我们可以简单地以最佳卖价下卖单（如果买卖价差大于往返佣金）。如果买入价没有上涨或者我们的限价卖单没有被成交，由于买入挂单量较大，我们可能仍然可以以最初的最佳买入价卖出，只是损失了佣金。

In markets where the bid-ask spread is bigger than two ticks, there is another simple trade to benefit from the expectation of an uptick. Simply place the buy order at the best bid plus one tick. If this is filled, then we place a sell order at the best ask minus one tick and hope that it is filled. But if it is not, we can probably still sell it at the original best bid, with the loss of commissions plus one tick. This is called ticking or quote matching. For this trade to be profitable, we need to make sure that the round trip commission per share is less than the bid-ask spread minus two ticks. This strategy is illustrated in Figure 7.3

在买卖价差超过两个点的市场中，还有另一个简单的交易可以从上涨的预期中受益。

只需简单地以最佳买价加一个点位下买单。如果它被成交，那么我们在最佳的卖价减去一个点位下一个卖单，并希望它被成交。但如果它不能成交，我们可能仍然以最初的、最佳买入价平掉它，相应的损失不过是佣金加上一个点差而已。这称为点数或报价匹配。为了使这笔交易有利可图，我们需要确保每份往返佣金小于买卖价差减去两个点。 该策略如图 7.3 所示。

FIGURE 7.3 Ticking Strategy. The original spread must be greater than two ticks. After the buy order is fi lled at B, we will try to sell it at S for a profi t of at least one tick. But if the sell order cannot be fi lled, then we will sell it at S′ at a loss of one tick.

点数策略。初始的点差必须大于两个点。 在 B 点完成买单后，我们将尝试在 S 点卖出，以获得至少一个点的利润。 但是如果卖单不能成交，那么我们将在 S' 处以 1 个点的损失平掉之前的买单。

(Ticking is not a foolproof strategy, of course. The original best bid before it was front-run may be cancelled if the trader knows that he has been front-run, leaving us with a lower bid price to unload our inventory. Or the whole situation could be set up as a trap for us: the trader who placed the original best bid actually wanted to sell us stocks at a price better than her own bid. So once we bought her stocks plus one tick, she would immediately cancel the bid.)

（当然，点数不是万无一失的策略。 如果原始最佳出价的交易者知道他已经被加塞抢先了，那么可能会取消原始最佳出价，从而让我们只能以更低的买入价来平仓我们的库存。或者整个情境可能本身就是为我们设置的一个陷阱：最初提出最佳买价的交易员实际上想以比他自己的出价更高的价格向我们出售股票。所以一旦我们加一个点买了他的股票，他就会立即取消买单。）---译者注: 虚假挂单 高频幌骗。

Even when there is no preexisting buying pressure or bid-ask size imbalance, we can create the illusion of one (often called momentum ignition). This works for markets with time priority for orders instead of using prorata fills. Let’s assume we start with very similar best bid and ask sizes. We will place a large buy limit order at the best bid to create the impression of buying pressure, and simultaneously place a small sell limit order at the best ask. This would trick traders to buy at the ask price since they anticipate an uptick, filling our small sell order. At this point, we immediately cancel the large buy order. The best bid and ask sizes are now roughly equal again. Many of those traders who bought earlier expecting a large buying pressure may now sell back their holdings at a loss, and we can then buy them at the original best bid. This is called flipping.

即使没有预先存在的买方压力或买卖规模不平衡，我们也可以创造一种错觉（通常称为动量点火  https://edu.csdn.net/course/play/26263/439224?spm=1002.2001.3001.4143 ）。 这适用于订单按照时间优先机制而运行的市场，而不是使用按比例填充的市场。 假设我们从非常相近的最佳买入量和卖出量开始。 我们会在最佳买价下放置一个大的限价买单，以创造买入压力的假象，同时在最佳卖价下放置一个小的卖限价单。 这会诱使交易者以买价购买，因为他们预计价格会上涨，从而使我们的小额卖单得以成交。 此时，我们立即取消大买单。 最佳买入价和卖出价现在再次大致相等。 许多之前预计会有很大的买盘压力而买入的交易员，现在可能会亏本出售他们持有的股票，然后我们又可以按照原来的最佳买价购回相应的股票。 这称为翻“炒卖”。 ---译者注: 虚假挂单 高频幌骗。

There is a danger to creating the illusion of buying pressure—somebody just might call our bluff and actually fill our large buy order. In this case, we might have to sell it at a loss. Conversely, if we suspect a large buy order is due to flippers, then we can sell to the flippers and drive down the bid price. We hope that the flippers will capitulate and sell their new inventory, driving the ask price down as well, so that we can then cover our short position below the original bid price. How do we know that the large buy order is due to flippers in the first place? We may have to record how often a large bid gets canceled instead of getting filled. If you subscribe to the private data feeds from the exchanges such as ITCH from Nasdaq, EDGX Book Feed from Direct Edge, or the PITCH feed from BATS, you will receive the detailed life history of an order including any modifi cations or partial fills (Arnuk and Saluzzi, 2012). Such information may help you detect   flippers as well.

制造购买压力的假象是有危险的——有人可能被我们的虚张声势所诱，实际上成交了我们的大买单。在这种情况下，我们可能就要以亏本的方式平调之前的买单。相反，如果我们怀疑一个大的买单是由于“炒卖者” 所为，那么我们可以卖给这些炒卖者并压低买价。我们希望相关的炒卖人士投降并出售他们的新库存，同时压低买入价，这样，我们就可以在低于初始的买入价点位平调我们的空单头寸。那么，我们怎么知道一开始的大买单是由于炒卖者所为呢？我们可能需要记录一个大的买入价被取消而不是被成交的频率。如果您订阅来自交易所的私有数据源，例如来自 Nasdaq 的 ITCH、来自 Direct Edge （美国第四大电子股票交易市场）的 EDGX Book Feed 或来自 BATS （高频电子交易做市商，其英文全称是Better Alternative Trading System，即更好的多元化交易系统）的 PITCH 源，您将收到订单的详细生命历史，包括任何修改或部分成交（Arnuk, Sal L., and Joseph C. Saluzzi. Broken Markets: How High Frequency Trading and Predatory Practices on Wall Street Are Destroying Investor Confidence and Your Portfolio. Upper Saddle River, NJ: FT Press, 2012.）。这些信息也可以帮助您检测到炒卖者的行为模式。

All these strategies and their defenses, bluffs, and counterbluffs illustrate the general point that high-frequency traders can profit only from slower traders. If only high-frequency traders are left in the market, the net average profit for everyone will be zero. Indeed, because of the prevalence of these types of high-frequency strategies that “front-run” large bid or ask orders, many traditional market makers no longer quote large sizes. This has led to a general decrease of the NBBO sizes across many markets. For example, even in highly liquid stocks such as AAPL, the NBBO sizes are often just a few hundred shares. And even for the most liquid ETFs such as SPY on ARCA, the NBBO sizes are often fewer than 10,000 shares. Only after these small orders are filled will the market maker go back to requote at the same prices to avoid being taken advantage of by the high-frequency traders. (Of course, there are other reasons for avoiding displaying large quotes: market makers do not like to keep large inventories that can result from having their large quotes filled.) Similarly, large institutional orders that were formerly executed as block trades are now broken up into tiny child orders to be scattered around the diff erent market venues and executed throughout the day.

所有这些策略及其防御、诈唬和反诈唬都说明了一个普遍观点，即高频交易者只能从速度较慢的交易者那里获利。如果市场上只剩下高频交易者，那么每个人的净平均利润将为零。事实上，由于这类“提前”大额买入或卖出订单的高频策略盛行，许多传统做市商不再使用大额的报单。这导致许多市场的 NBBO 规模普遍减少。例如，即使在苹果（AAPL）一类的具有高度流动性的股票中，NBBO 的报价规模通常也只有几百股。即使对于流动性最强的 ETF，例如ARCA（一个电子交易所）上的 SPY，在NBBO 中的报单额规模通常也少于 10,000 股。只有在这些小额订单被执行后，做市商才会以相同的价格重新报价，以免被高频交易者所利用。 （当然，避免显示大额报单还有其他原因：做市商不喜欢保留大规模的库存头寸，而他们的大额报价单可能会被成交。）同样，以前作为大宗交易执行的大机构订单现在被分解成小额的子订单，其相关指令以全天候的形式被分散在不同的交易场所。

Stop hunting is another favorite high-frequency momentum strategy. Research in the currencies markets indicated that once support (resistance) levels are breached, prices will go further down (up) for a while (Osler, 2000, 2001). These support and resistance levels can be those reported daily by banks or brokerages, or they can just be round numbers in the proximity of the current price levels. This short-term price momentum occurs because of the large number of stop orders placed at or near the support and resistance levels.

打止损是另一种最受欢迎的高频动量策略。 货币市场的研究表明，一旦突破支撑（阻力）位，价格将进一步下跌（上涨）一段时间（Osler, Carol. “Support for Resistance: Technical Analysis and Intraday Exchange Rates.” Federal Reserve Bank of New York Economic Policy Review 6 (July 2000): 53–65.

Osler, Carol. “Currency Orders and Exchange-Rate Dynamics: An Explana[1]tion for the Predictive Success of Technical Analysis,” 2001. Forthcom[1]ing, Journal of Finance. Available at http://newyorkfed.org/research/staff _reports/sr125.pdf）。 这些持位和阻力位可以是从每天的银行或券商所提交的报告当中获取，也可以只是当前价格水平附近的整数。由于止损的订单大都设置在相应支持位和阻力位的附近，所以，极易诱发短期的价格动量模式。

To understand this further, let’s just look at the support levels, as the situation with resistance levels is symmetrical. Once the price drops enough to breach a support level, those sell stop orders are triggered and thereby drive the prices down further. Given this knowledge, high-frequency traders can, of course, create artificial selling pressure by submitting large sell orders when the price is close enough to a support level, hoping to drive the next tick down. Once the stop orders are triggered and a downward momentum is in force, these high-frequency traders can cover their short positions for a quick profit.

为了进一步理解这一点，让我们看看支撑位，因为阻力位的情况是对称的。 一旦价格下跌到足以突破支撑位，这些卖出止损订单就会被触发，从而进一步压低价格。 有了这些知识，高频交易者当然可以通过在价格足够接近支撑位时提交大量卖单来创造人为的卖压，希望推动下一个价格下跌。 一旦触发止损单并且下行势头生效，这些高频交易者就可以平掉他们的空头头寸以快速获利。

If we have access to the order flow information of a market, then we have a highly valuable information stream that goes beyond the usual bid/ask/last price stream. As Lyons discussed in the context of currencies trading, “order flow” is signed transaction volume (Lyons, 2001). If a trader buys 100 units from a dealer/market maker/order book, the order flow is 100, and it is −100 if the trader sells 100 units instead. What “buying” from an order book means is that a trader buys at the ask price, or, equivalently, the trader submits a market order to buy. Empirical research indicates that order flow information is a good predictor of price movements. This is because market makers can distill important fundamental information from order flow information, and set the bid-ask prices accordingly. For example, if a major hedge fund just learns about a major piece of breaking news, their algorithms will submit large market orders of the same sign in a split second. A market maker monitoring the order flow will deduce, quite correctly, that such large one-directional demands indicate the presence of informed traders, and they will immediately adjust their bid-ask prices to protect themselves. The urgency of using market orders indicates that the information is new and not widely known.

如果我们可以访问市场的订单流（即买方/卖方成交额）信息，那么我们就有了一个非常有价值的信息流，它超出了正常的买入价、卖出价或最终报价的信息价值。正如 Lyons 在其关于货币交易的背景下所讨论的那样，“订单流”是签署的交易量（Lyons，2001）。如果交易者从经销商/做市商/订单簿中购买 100 个单位，则订单流为 100，如果交易者卖出 100 个单位，则为 -100。从订单簿中“买入”的意思是交易者以卖价买入，或者等价地，交易者提交市价单进行买入。相应的实证研究表明，订单流信息是价格变动的良好预测指标。这是因为做市商可以从订单流信息中提取重要的基本信息，并据此设置买入价-卖出价格。例如，如果一家大型对冲基金刚刚获悉一条重大突发新闻，他们的算法将在瞬间提交相同方向的大笔市场订单。监控订单流的做市商会非常正确地推断出如此大的单向需求表明存在知情交易者，他们会立即调整相关的买入价-卖出价以保护自己。使用市价单的紧迫性表明该信息是新的，并不广为人知。

Since most of us are not large market makers or operators of an exchange, how can we access such order flow information? For stocks and futures markets, we can monitor and record every tick (i.e., changes in best bid, ask, and transaction price and size), and thus determine whether a transaction took place at the bid (negative order flow) or at the ask (positive order flow). For the currencies market, this is difficult because most dealers do not report transaction prices. We may have to resort to trading currency futures for this strategy. Once the order flow per transaction is computed, we can easily compute the cumulative or average order flow over some look-back period and use that to predict whether the price will move up or down.

由于我们大多数人都不是交易所的大型做市商或运营商，我们如何获取此类订单流信息？ 对于股票和期货市场，我们可以监控和记录每一次报价（即最佳买入价、卖出价以及交易价格和量的变化），从而确定交易发生在买入价（负订单流）还是卖出价 （正订单流）。 对于货币市场来说，这很困难，因为大多数交易商不报告交易价格。 为了这个策略，我们可能不得不求助于交易货币期货。 一旦计算了每笔交易的订单流，我们就可以轻松计算某个回顾期内的累积或平均订单流，并使用它来预测价格是上涨还是下跌。

• Many high-frequency momentum strategies involve imbalance between bid and ask sizes, an imbalance that is sometimes artifi cially created by the high-frequency traders themselves.

许多高频动量策略涉及买卖规模之间的不平衡，这种不平衡有时是由高频交易者自己人为造成的。

• Stop hunting is a high-frequency trading strategy that relies on triggering stop orders that typically populate round numbers near the current market price.

止损狩猎是一种高频交易策略，它依赖于触发止损单，这些止损单通常会成交在当前市场价格附近的整数。

• Order flow can predict short-term price movement in the same direction.

订单流可以预测同一方向的短期价格变动。

广告：

《高频交易系统开发 C++，低延迟》
https://edu.csdn.net/course/detail/24668

高频交易模型开发：《随机开仓，任意开仓，止盈1~2个点，止损1~20个》
https://edu.csdn.net/course/detail/30949

参考论文:

Maslov, Sergei, and Mark Mills. “Price Fluctuations from the Order Book  Perspective: Empirical Facts and a Simple Model.” Physica A 299, no.  1–2 (2001): 234–246

Hellström, Jörgen, and Ola Simonsen. “Does the Open Limit Order Book Reveal Information About Short-run Stock Price Movements?” Umeå Economic Studies, 2006.

Arnuk, Sal L., and Joseph C. Saluzzi. Broken Markets: How High Frequency Trading and Predatory Practices on Wall Street Are Destroying Investor Confidence and Your Portfolio. Upper Saddle River, NJ: FT Press, 2012

Durbin, Michael. All About High-Frequency Trading. New York: McGraw-Hill, 2010.Durden, Tyler. “

Harris, Larry. Trading and Exchanges. New York: Oxford University Press, 2003

Sinclair, Euan. Option Trading: Pricing and Volatility Strategies and Techniques.Hoboken, NJ: John Wiley & Sons, 2010.

Arnuk, Sal L., and Joseph C. Saluzzi. Broken Markets: How High Frequency Trading and Predatory Practices on Wall Street Are Destroying Investor Confidence and Your Portfolio. Upper Saddle River, NJ: FT Press, 2012.

Osler, Carol. “Support for Resistance: Technical Analysis and Intraday Exchange Rates.” Federal Reserve Bank of New York Economic Policy Review 6

(July 2000): 53–65.

Osler, Carol. “Currency Orders and Exchange-Rate Dynamics: An Explana[1]tion for the Predictive Success of Technical Analysis,” 2001. Forthcom[1]ing, Journal of Finance. Available at http://newyorkfed.org/research/staff _reports/sr125.pdf

Lyons, Richard. The Microstructure Approach to Exchange Rates. Cambridge, MA: MIT Press, 2001.

本文来自《Algorithmic Trading - Winning Strategies and Their Rationale (Wiley Trading), 2013》