23
Oct, 2009
23
Oct, 2009
Stealthy = Inconceivable ; Another Real World HFT example.
I like examples.
My client routed her small stray sell order (about 5,000 shares) to a well-known electronic brokerage firm’s stealthy algorithm, which utilizes dark pool aggregation, and has a very cool macho helicopter-military-like name. These are the facts.
The stock: small cap OTC restaurant stock, ticker left out of the discussion. Average daily volume = 62,000 shares
The bid/ask: 28.66 28.68 200 shares up
The situation :
1) Very cool macho helicopter-military-like named Algo gets the order. It is a “dark” algo.
2) The stock drops from 28.66 – 28.68, to 28.61 – 28.64. No shares trade. The stock dropped only when she entered the order into the dark algo.
3) She then cancelled the order from the algo.
4) The stock rebounded back to 28.67 – 28.69. Again no shares have traded.
5) Sell order now entered using an aggressive router which uses intermarket sweep (ISO order), which simultaneously routes stock to the 67 cents and 66 cent bids in the book.
6) Execution = sell of 4,770 shares at 28.6625
7) Mission accomplished.
The Moral: Stealthy and Dark is not always good. In fact there are studies out there that question the “low cost of dark algorithmic trading”, and demonstrate that they cost more. In the above example, the trader’s first move was to enter the sell order into a reputable dark algo, which had prided itself on being stealthy and discreet. The stock dropped with the trader still executing nada and holding the bag. The trader’s second move was to use a different, more aggressive method, which was programmed to purposely avoid specific dark pools. The result was a great, tight execution, that was achieved after much angst, much needless labor, and much cursing.
When dark algorithms call themselves stealthy, I can’t help but think of the scene in The Princess Bride where Vizzini keeps saying “Inconceiveable”, and Inigo Montoya responds, “You keep using that word. I do not think it means what you think it means.”
A big thank you to my dear princess in northern Florida, for contributing this tale to me in my caffeine-induced state. You know who you are.
Sal, this is interesting, but I have two problems with it:
First, it is always easy to find examples that make a point, even though those examples may be outliers. How typical is this example?
Second, how do you show that this is due to HFT? Traders have had quotes move against them on no/low volume long before anyone heard of HFT. I don’t see any causality here.
What would be interesting is to see some solid quantitative analysis of trade execution quality. When I was at NYFIX we used QSG to do some analysis, and the data that I saw showed that the Millennium dark pool provided better executions for some categories of names, worse for others.
I have yet to see any objective analysis that shows the impact of HFT.
Brennan
(In the interest of disclosure, while I have friends at NYFIX and QSG, I don’t currently have any professional relationship with either firm, don’t own any of their stock, etc..)
Brennan,
I hope you are well. We respect you greatly, and long for the good old days at Instinet. Let’s connect and we can talk. In the meantime, check out the article in Barron’s this weekend that Joe is linking to this morning. It cuts to the meat of how the markets have changed, and why examples like the one I have pointed out are occurring. Some studies are coming; they were initially slow to come out for two reasons: 1) lack of incentive 2) lack of access to data. I would love to see a tag on orders called HFT.
Best,
Sal