Recent arguments accuse high-frequency traders (HFTs) of a specific market distortion scheme. The HFTs, the argument goes, use their soon-to-be-cancelled limit orders to mislead large investors about the shape of the supply and demand curve. This HFT strategy is purported to work as follows: 1) an HFT posts lots of limit orders on both the bid and the ask sides of the trades; 2) once the large trader's market order hits the bid (or lifts the offer), the HFT now knows that the large trader is now selling (or buying); 3) the HFT cancels all other limit orders and starts aggressively trading in the same direction as the large trader -- relying on their ultra-fast infrastructure to essentially front-run the large trader. This short note looks through the nuts and bolts of the proposed strategy to show that it is simply unprofitable, and therefore, is not actively practiced in the markets.
The key assumption behind the feasibility of such a strategy is that just by serving as a counterparty to a market order, an HFT obtains superior information about the intentions of the institutional traders. While an HFT may indeed be 'hit' by orders from a well-researched institutional strategy, such an HFT carries significant risks of being on the opposite side of impending direction of the market. Consider the following example: an HFT places a buy limit order, which is then matched by a market sell order of a well-researched institutional investor. If the institutional investor sells because he expects that the market price is about to drop with high probability, the HFT starts losing money as soon as that order is executed. The higher is the predictive power of the institutional investor, the deeper the loss of the HFT. Not many HFTs would be able to stay solvent using this strategy.
The following issues further complicate matters for the HFT.
1) The HFT has to place limit orders. To reel in large traders, the HFTs presumably need to place large limit orders (otherwise, how would the HFTs separate large traders from small traders)? A large order makes the HFT's loss that much bigger.
2) Suppose the HFT then absorbs the initial loss, and starts to aggressively trade in the direction of the institutional trader (sell in the case of our example). With his first sell order, the HFT closes the previous buy position, incurring the first loss. The HFT then proceeds to place a series of really fast market sell orders, with the intention of "front-running" the future market impact of the institutional trader. In this case, the HFT quickly reduces the price to its fundamental value. The institutional trader then has no reason to make any trades following his first large trade -- he can now just close his initial large short position at a profit -- a profit on a smaller dollar amount than planned, yet of expected magnitude, and most importantly, guaranteed via the actions of the HFT! In fact, in this setup it is the institutional traders who are most likely to manipulate the markets -- regardless of any fundamental values, any large institutional trade would lead to a riskless profit!3) If all HFTs trade the same strategy, however, as is often assumed, the seemingly limitless HFT riches in point 2) above vanish. This is why:
- most limit buy orders are cancelled shortly after the institutional trader's sell order goes through
- the HFTs are unable to find counterparties to front-run institutional traders
- the market price instantaneously drops drastically.
As a result, the institutional trader immediately closes his position at a considerable profit, but the HFT is experiencing a severe loss.
This is not an HFT strategy I (or anyone I know, for that matter) would trade!
Irene Aldridge will be teaching a course on the cutting edge developments in best algorithmic execution, specifically geared towards institutional investors and their brokers to help risk-manage their order flow and minimize trading costs. The course will take place in NYC on May 9 and 10, 2012. For more information and to register, please visit www.hftcourse.com.
Follow Irene Aldridge on Twitter: www.twitter.com/irenealdridge