Recently, Fidelity, one of the largest money managers in the world, has been sending feelers out to its peers to form a private “dark pool” in which they could all trade anonymously.

What’s this all about?

I’m not sure who made up the name “dark pool.”  But it glamorizes a pretty mundane operation.  A dark pool is a computer-driven trading network where professional investors buy and sell securities with each other in a low-cost anonymous way.

They’re meant to solve two problems that every large investor like Fidelity who’s subject to SEC regulation in the US has:

–in trading securities for their clients,money managers are required to obtain the lowest cost in making any trade as well as the best execution of the order.  Best execution, which I take to mean the most favorable price, given the circumstances at the time of the trade, is a rather vague and contentious concept.  But lowest cost, meaning the lowest commission or bid/ask spread paid to get the trade accomplished, is relatively clear.

It’s also very clear that dealing with a third-party broker isn’t the lowest cost way of doing business.  Dealing directly with another institution through a computer trading network may cut commission/spread costs in half.  As a result, increasing amounts of trade is being done through dark pools so institutions can establish that they’re working to fulfil the lowest-cost legal mandate.

–any money manager wants to keep his trading activity as secret as possible.  After all, no one wants others to be freeriding on investment ideas that a manager has developed after long and expensive research efforts.

This is a particularly pressing issue for managers with large amounts of money under management, since such a manager will often have orders that are so big they can take, say, a month to execute–sometimes longer.  The trickiest part of such trading is keeping the manager’s activity secret for as long as possible.

Again, brokerage house trading operations for third parties are not a great way to go.  They tend to leak like sieves.  Part of this is a function of order size.  The broker may approach potentially interested parties.  As/when the size of the order becomes apparent, the other party may be able to guess the identity of the institution the broker represents.

There’s mre than that, however.  Information is also a valuable resource.  In my experience, most important clients of a broker–and the broker’s own proprietary trading desk–would know the general outlines of a Fidelity order within minutes of its being placed.  The broker might not use the Fidelity name, but the description ” a large institution in Boston” would leave little to the imagination.  The idea is that smaller clients will regard this as valuable information and will compensate the broker with increased trading commissions in return for continuing access.  (My tendency would be to do less business with a broker who acts this way, but apparently I’m in the minority.)

Dark pools, though sometimes illiquid, are one solution to this problem.

It turns out, though, that the dark pools also have their issues.  One that I find interesting is that to obtain liquidity dark pools may allow high frequency traders to participate.  And, it turns out, they have found “big data” ways to figure out which orders are Fidelity’s–and to use this information to trade against them.

Fidelity’s response is to try to form a dark pool that will consist only of institutional investors, without high frequency traders.  Such a setup might have issues of its own.  If there are only, say, four major members it may be that the trading intentions of the others will be obvious to all.  But, in Fidelity’s view at least, that would be better than the current situation.