Enabling Platforms For Fast Markets (Earlier in February FINRA filed with the SEC to reduce the time allotted to print a transaction that occurred to the tape from the current 30 seconds to 10. In an exchange environment dominated by low latency connections, high frequency trading and the persistent need for speed measured in microseconds, 10 seconds represents an eternity.
According to FINRA, only 94.41% of all trades are reported within 2 seconds today. Moreover, its stunning to consider that today’s rules allow for prints to occur a full 30 seconds after a trade occurs. Even with 99.96% of all trades reported within 10 seconds today, the rules have been slow to react.
Our current environment leaves most market participants with no certainty whether a particular trade reflects the immediate current market or at some point up to 30 seconds prior, and the tape does not distinguish when the trade actually occurred, which creates issues for upcoming programs, such as limit up/limit down. In an exchange environment dominated by trading, the print is an important and often overlooked component to an efficient market structure, and it is long overdue for change.
This important issue should not to be overlooked, and that is exactly why I’ve commented on the issue. Read my comments below in support of FINRA’s plan to move printing to no later than 10 seconds after the transaction.
Comments | Post a Comment
1 Comment to "Trading May Be High Frequency, But Printing Continues to Be Highly Latent":
stevewunsch
08 March 2013
This view misunderstands the issue. The overwhelming majority of that 0.04% of prints that are late are the result of the subsequent printing of all the individual trades that were batched for single price openings and closings of the market. It is fundamentally misleading to begin with to treat these trades the same as individual trades that were decided on and entered separately by individuals, institutions or algorithms. They were not, but rather were entered prior to the opens and closes in which they participated in batch mode at the same price and time. There is no pattern or practice here to report late for manipulative or other purposes. Instead, the "late" prints are the result of stressed systems mostly at the exchange level that are inadequate to process the long queues that result from trying to treat them as individual trades, not batch trades. While it might be possible to build the systems capacity to handle the added volume within 10 seconds, the cost of doing so could be very large, both at the exchanges and at the individual member firms that will have to beef up their own capacity to deal with the beefed up exchange systems, as well as at the SIPs. One member firm estimated that cost at upwards of $100,000 per month for that firm alone. Regardless of cost, the real issue is, what is the point of pretending that batch trades were not batch trades, but individual trades? Even if this could be done at a reasonable system cost, doing so would not improve transparency, it would degrade it. The right way to handle this is to identify the batch trades as such, i.e., "X number of shares from Y number of orders participated at a price of Z at the open or close," and leave it at that.
Comments (49)