Blog article
See all stories »

What I did not do

When I got into work this morning, I decided not to return a missed call, not to review a document, not to read a report, not to ask some colleagues for advice, not to fill in my timesheet for last month, not to think about further job training and, most importantly, not to join an internal weekly meeting which was covering a topic not relevant to me. Instead I decided to write this blog first.

If you thought that my introduction was rather long winded, you will not be pleased to hear that your algo audit trails could look exactly like that too, if ESMA’s current proposal for record-keeping makes it into the final Level 2 text of MiFID II. ESMA suggests that audit trails should include all market data messages relevant when “algorithms make a decision to submit or not submit an order”. This unfortunate wording suggests that firms not only have to store market data when the algo decides to act (add, amend, cancel order), but also when the algo makes the decision not to do anything. How this could be relevant is beyond me, as I don’t see why doing nothing could harm the market. But I certainly do know that algos can decide to do nothing many times a second, thus easily increasing the size of your audit trail by a huge factor.

We will have to wait to see what ESMA finally decide on that point. In the meantime, I am not going to get a coffee or a tea, not going to read my emails and not going to schedule that meeting nor worry about the next piece of regulation coming. Instead, I think I might just go and buy some shares in a hard disk producer, just in case!

 

2628

Comments: (1)

Ketharaman Swaminathan
Ketharaman Swaminathan - GTM360 Marketing Solutions - Pune 19 September, 2014, 12:29Be the first to give this comment the thumbs up 0 likes

Arresting post. Algo trading is known to be complex. It was hard to understand everything written about it even by a bestseller author like Michael Lewis in his latest book bestseller Flash Boys. Within algo trading, the "do nothing" decision is arguably even more complex. Kudos for using a novel style to explain it so that everyone can understand it. 

Regulators probably have the same view of an algorithm that didn’t place the order as Sherlock Holmes had of the "Dog that Didn’t Bark In The Night”!

On another note, it’s shares of producers of "magneto-optical drives" that would interest you. As I learned from a benchmark study conducted a few years ago, the size of files generated during these exercises is too large for hard disks (or even to be transferrred electronically). We had to back up the data on to a set of MO drives and hand carry them from the lab in the south of France to our dev center in the western India via a combination of ferry, train and plane. Because of the risk of transit loss / damage during the long winding journey, we bought another set of drives, copied the logs onto them, and kept them as backup in our London office! Moral of story: The MO market size is actually double of what it appears to be at first glance!

Now hiring