I've been trying to understand smart order routers used by brokerages which do algorithmic trading. As I understand it, you simply watch the best bids/offers of various ECNs, market makers, etc. When you need to rout an order, you simply send it to the one with the best price. However, with RegNMS, as far as I know, market centers have to rout your order to the 'best' destination any way, does that mean smart order routers are not as valuable any more? Since prices and sizes change so quickly, is it possible that an order simply gets ping-ponged all over the street, while potentially missing executions? Of coarse, I'm making the assumption that ISLD has to rout my order to ARCA, if ARCA has a better price. Is that even true? My description of smart routers is extremely basic. Under my assumptions, the only way a smart order router adds value in today's market is if it does things like find hidden liquidity from dark pools, etc. Correct? I'll appreciate any pointers to papers, documents, articles or insight from experience. Thanks
In the current environment, the primary purpose of "smart routing" seems to be to "internalize" customer orders. "Internalization" means that brokers can trade against their own customer while avoiding the "hassle" of external competition on an open exchange. The net result is higher spreads for everyone. That's why "smart routing" generally costs less. Smart for the brokerage, not so smart for the investing public. FYI, such brokerages have been increasing this practice in recent months, and many would like to spread this practice to other instruments. Some brokerages don't do this, but rather just execute client orders and don't trade for themselves; these are often called "agency" brokers. Agency brokers generally operate in the institutional, professional, or "near-professional" customer space. (But even here I think you have to do your homework because no one enforces a strict definition of the term "agency".)