The Gateway to Algorithmic and Automated Trading

Max ADV, Minimum Impact

Published in Automated Trader Magazine Issue 24 Q1 2012

US financial stocks had a rough time in 2011, with modest gains up to mid February more than expunged by progressive losses through to midsummer, which were then compounded by a sharp collapse in early August. Kish Desai, head of EMEA Equity Sales, and Ali Mohsin, Algorithmic Trading Product Manager, at ConvergEx Group describe how a trader managed to execute an order approaching half a financial stock's ADV on August 8th - one of the most volatile trading sessions of the year.

The Scenario: It is the afternoon of Friday August 5, 2011. The manager of a UK based hedge fund has been tracking CME for some weeks and is looking for a buying opportunity. The stock has been strong recently relative to the XLF Financial Select Sector ETF, of which it is a component. While XLF has in the past week comprehensively taken out both its two previous lows for the year (which occurred in early June and mid July), CME has to date remained comfortably above its early June lows. Furthermore, the stock has been strong relative to the ETF on a day by day basis. For example, at the close of the current session XLF was down nearly 2% on its previous close, while CME was actually up by about 1.4%.

Nevertheless, while CME looks attractive versus its peers, the manager feels that stock will probably still decline during the coming week in view of poor market sentiment, particularly with regard to financials. The fallout from the US sovereign downgrade continues to weigh on the market as does the European sovereign debt crisis, which continues to deliver unpleasant surprises.

Unless news over the coming weekend is particularly bullish, the manager therefore intends to instruct his financials trader to pick up a large position early the following week.

The Asset: CME Group

The Challenge: To buy 200,000 shares in CME (approximately 50% of ADV).

The Trader: The trader specialises in financial stocks and uses a number of proprietary divergence indicators that she has developed to pick short term buying and selling opportunities.

The Algo: The algorithm takes into account five factors in its operation:

Price: As prices become favourable, the algorithm will execute more of the order.

Timing Risk: As volatility increases, the algorithm will become progressively more aggressive.

Spread: The wider the spread becomes, the more passively the algorithm participates.

Tick: As the stock price is printing, become more passive or aggressive as needed.

Momentum: If there is a strong trend in the stock price, the algorithm becomes more aggressive.

The algorithm constructs a comprehensive, multi-dimensional sensitivity profile that simultaneously measures these factors to determine the optimum strategy. The relative importance of factors can be determined by specifying one of three execution styles - Passive, Neutral, or Aggressive - that will allow the algorithm to run on autopilot.

However, if the trader wants to take over direct control, he/she can use:

Volume Min: This setting allows the trader to control the algorithm's minimum participation rate; the algorithm will not trade any slower than this rate. This is an important setting to use if the market is trading away from a liquidity seeking algorithm and liquidity can be hard to find.

Volume Max: This setting allows the trader to control the maximum speed at which he/she wishes to participate. This is important when the market is moving favourably because without it, most liquidity-seeking algorithms would want to fill the entire order as soon as liquidity is readily available.

MinDarkFill: This controls the minimum fill size a trader will accept from dark venues. The average execution size in dark pools tends to be a little over 200 shares. By increasing the minimum fill size, this lets the trader only interact with larger than average dark orders in the marketplace. Traders use MinDarkFill on sensitive orders where their concern is that multiple small fill s may leak information to the marketplace.

7.30am - (all times are US EST) - At 200,000 shares, the order represents almost 50% of CME's ADV and the trader's instructions are to fill as much of the order as possible. She realises this is a difficult trade, so she wants to use a liquidity-seeking algorithm and chooses this particular algorithm because of the large number of venues it trades in and the simple but powerful controls it offers her.

8.00am - The market is filled with negative news from the weekend and the futures are down. CME closed on Friday at $269.42 and the trader expects the stock to gap down. The prior low is at $261.10 set on June 8, 2011 and she expects sellers to hit the stock if it breaks this low. She will use this information to help her navigate her large buy order. She also knows that the CME is a component of the XLF, so she has a chart of the XLF on to gauge the selling pressure in the market.

She enters an order with a Volume Max of 30% and a MinDarkFill of 500 shares because she doesn't want to be more than 30% of the volume at open. Also, she only wants to interact with larger dark orders.

Figure 1: CME Daily Chart Jan to Aug 2011

Figure 1: CME Daily Chart Jan to Aug 2011

9:30am - CME opens at $262 (down 2.75%). XLF opens at $12.99 (down 3.42%). The algorithm immediately gets to work and executions begin to trickle in (500 shares and larger from dark pools). The trader is watching the CME chart and looking at the XLF for clues as to what the market will do. As of the open, both are rallying, but the trader understands that the market is in price discovery mode until about 10am, so for now she waits while the algorithm trades her order.

Figure 2: CME Intraday Chart Monday August 8th 2011

Figure 2: CME Intraday Chart Monday August 8th 2011

10.00am - CME is trading near the highs while the XLF is trading near its open. The trader reads this as CME is stronger that the broader market. This worries her a little because she knows her order is large and she is betting on sellers giving her the liquidity she needs to get the order done. She wonders if she should increase her volume max from 30% to 45% (closer to the 50% of ADV her stock represents) but decides to leave her order in the way it is and watches to see if the XLF wi ll rally from its open or if it breaks to new lows.

10:15am - XLF breaks the opening day range to the downside and makes new lows on volume. CME breaks down and moves towards its opening price. The trader notices more block fills of 500 shares or more from dark venues and views this as an indication of larger sellers coming into CME. For now, she leaves the algorithm as is.

10:36am - CME tests its low of the day (also the open) and rallies. The trader notes the intraday range of CME is $262 - $265.56. XLF also rallies from its lows. The trader makes the determination that the market is oversold right now and is just taking a breather.

11:18am - CME is moving towards the top of the range. This looks bullish at first but the volume is anaemic. The trader determines that this is just a head fake and expects CME to trade near the lows of its range in the next 30 minutes. She replaces her Volume Max in the algorithm with 20% and puts a top at $265.50, near the high of the range. A few minutes later, CME breaks to new highs and is quickly hit by sellers. The trader is happy with her adjustment, as she limited her buying near the high end of the range.

Figure 3: XLF Daily Chart Jan to Aug 2011

Figure 3: XLF Daily Chart Jan to Aug 2011

11:27am - With only a few minutes left to the European market close, the trader watches as XLF is on its way to make new lows for the day. CME is trading near $264, near the lower end of the range. The trader likes the prices here and decides to switch her Volume Max back to 30%. She leaves her top at $265.50 in case CME tests the high end of the range again.

12.00pm - Our trader has a lunch meeting and won't be back until 1pm. She leaves the algorithm to trade, but makes a few modifications. She drops her MinDarkFill from 500 down to 200. She knows the lunch volume won't be great and wants to participate primarily in dark. Reducing the dark size will allow her to interact with more dark liquidity. She also adds a Volume Min of 15%. Now the algorithm will be trading between 15-30% of volume.

12:57pm - The trader is back on the desk. She evaluates where the algorithm is. CME is trading near the low end of the range and the algorithm has filled 90,000 shares on her 200,000 order. CME made a new low during lunch and is trading near the lows at $262. The trader is a little worried because she only has three hours to complete more than half the order.

She notes that the year low is at $261.10. If CME can break these lows, she expects sellers to hit the stock. XLF continues to make new lows so she's reassured that there is a good chance we could test the year lows. She removes her Volume Min, but keeps her MinDarkFill set to 200. She needs all the liquidity she can get at this point.

1:21pm - CME breaks to new lows of the day and the year. Volume picks up and sellers come into the stock. The trader expects the old range to be resistance now so she brings down her top from $265.50 to $262. As volume picks up the trader has now done a little over half her order. XLF has also made new lows so she sees the sell off accelerating. All good things for her order.

Figure 4: XLF Intraday Chart Monday August 8th 2011

Figure 4: XLF Intraday Chart Monday August 8th 2011

2:15pm - CME and XLF have continued to sell off. The trader is filled on 140,000 shares of her order with one hour 45 minutes left in the day. The trader decides that she doesn't want to put a 30% max threshold anymore and replaces it to 45% because she wants to make sure she can get it done. This could also mean she might complete her order early, but she has happy to get done at the current price of $253 if the liquidity is available.

2:18pm - XLF hits a new low on a volume spike. The trader sees this as possible capitulation in the stock and wonders if the bottom is in for the day. CME also had a volume spike, so she decides to do nothing and watch if both the stocks will rally from here.

3:00pm - The trader makes the determination that the low is in for the day. She has filled 165,000 shares of her order. She decides to remove her Volume Max completely and enters a Volume Min of 35%. She also removes her MinDarkFill. She will take everything at this level.

3:21pm - Both XLF and CME are making higher highs and higher lows. She draws a simple down trend line on her XLF chart and sees that it broke to the upside.

The trader sees this as an indication that her bottom call was right. She is worried that the stock will reclaim the $262 range before the day is complete. Luckily, she is filled on 193,000 shares. She just watches to see if CME and XLF continue to make higher lows.

3:30pm - CME sells off a little and the trader completes her entire order. She is happy that she was able to fill her large order and with how she navigated the stock.

3:48pm - Both CME and XLF break to new lows. The trader is filled on her order and did not see new lows for the day. All she can do now is watch to see how much she left on the table.

4.00pm - CME closes at its low. The trader was off on her low call but she was able to get done near the lows, so she is content.