# 997 – Shades of darkness

• A joint work with Albert J. Menkveld and Haoxiang Zhu

• TL;DR: Not all dark pools are created equal. Together with the lit exchange market, two types of dark venues are ranked according to a “pecking order”, along which trading cost and venue immediacy increases. When investors face high urgency to trade, the pecking order theory predicts that people flow out of the dark to lit. Such a dynamic fragmentation appraoch offers a better understanding of the current financial market.

• A less technical blog by Albert: [here]

The word “dark” always has some negative connotations. Interestingly, the financial market nowadays is populated with a large number of “dark pools”. For example, there are more than 50 dark pools in the U.S. equity market and they account for about 30% of total volume.

Such darkness in our market does sounds a bit, if not scary, “evil”, just like the above two darkest evils in the pop culture (depending on your age, one of the two might appear more familiar than the other):

I am sure there are darker ones. Are dark pools really that “dark”?

In fact, these dark pools, as much of the literature has argued, do play an important role. Their existence tailors to a particular need for some investors. To see this, let’s think about the fundamental difference between trading in the dark and in the lit: visibility of the liquidity provision.

##### Lit v.s. dark: a brief comparison

Liquidity in the lit exchange is readily available: Once you see it, you know you can trade on it (let’s forget about the speed argument for now and assume what you see is what you get). But such liquidity that allows you to trade immediately is very expensive. For example, if you want to buy 2,000 shares and the limit orders available are lying on ask prices of $10.01,$10.02, …, $11.00, and 200 shares at each level. Then to execute these 2,000 shares, you pay an increasing price for each additional 200 shares. The total cost function is convexly increasing. The more you trade, the more expensive is t he marginal cost. Dark pools are the opposite. However large your order is, you pay the same price (in a “midpoint crossing” dark pool). That is, if the best ask is$10.01 and the best bid is $9.99, you always pay$10.00 for each unit you demand. The flat cost is much cheaper than the convexly increasing cost function when trading in the lit exchange. But such a low cost comes at a tradeoff: You do not see the liquidity provision in the dark pool (hence the name “dark”), and such opaqueness implies that if you submit an order to the dark pool, its execution is not guranteed. So, although you pay less for execution, less of your order can be executed timely.

Let’s summarize the difference. Between a lit exchange and a dark pool, one trades off the execution price for immediacy. Such a tradeoff is known as cost-immediacy consideration.

The above is surely a simplified narative of the reality. The truth is that not all of the 50 dark pools in the U.S. equity market are the same. They differ in the trading protocols and create a “shades of darkness”, each developing into a specific niche that targets some investors with the specific cost-immediacy tradeoff in mind.

To this extent, it seems that the darkness in our financial market, afterall, is not that bad. It does create market fragmentation, but for a valid reason: Investors are different, so why should the trading venue be exactly the same?

Just like the French have so many different types of cheeses, each trading venue develops and specializes in their own flavor, finding their own customer niche.

##### When investors are anxious: Flowing away from dark to lit

In a new working paper, my coauthors and I are able to sort, exactly according to thecost-immediacy tradeoff explained above, the trading venues shaded with different degrees of darkness. The sorting ranges from the low-cost, low-immediacy midpoint crossing dark pool (DarkMid) to the high-cost, high-immediacy lit market (Lit), and covers an intermediate venue type called non-crossing dark pool (DarkNMid).

The revealing finding is that when the market is under an urgency shock, investors tilt their order flows from dark to the lit. That is, when investors are anxious to trade, they trade far less in the low-cost, low-immediacy venues but switch to the more costly venues for their immediacy. Using a unique dark pool dataset, we not only evidence such model prediction but also empirically illustrate the magnitude of investors' tilting their order flows.

##### Some side thoughts (not in the paper)

The result is worth chewing over. Among others, it suggests that when the market volatility turns high—hence investors become anxious about their positions and feel urgent to trade—the market fragmentation reduces. The outlain appearnce of an extremely fragmented market is perhaps an incomplete, static look at the current market structure. To have a more complete picture, one should not overlook the mobility of order flows in different environments. Our dynamic approach to market fragmentation offers this better view.

Maybe, our market does appear very fragmented in a peaceful time but reunites (toward the lit market) when “the storm” strikes, as if there is an invisible hand that pushes all investors together, when liquidity begets liquidity. Isn’t it beautiful?

13:27:37, Oct 16, 2014
bzy@Fontainebleau