Minimum Quantity (MQ) is popular among traders as a tool for improving execution quality. As an exchange, we have a unique perspective on the effectiveness of MQ use. In this three-part series, we attempt to quantify how effective MQ use is in optimizing execution.
There are many factors that contribute to execution quality, including adverse selection, information leakage, and ability to source liquidity. The weight that is applied to each of these factors varies based on the goals of the trader and the strategy.
In this blog post, we focus on one reason traders use MQ: to reduce adverse selection. We illustrate why our data suggests MQ may not be an effective tool for this purpose.
Trade Size in Today’s Market
Below, we show the distribution of midpoint trades by trade size overall in the market and on IEX. Please note, we are showing the percentage of trades in each bucket rather than the amount of volume.
We look at the data both by the number of shares in the trade and by stock price, and only include trades in stocks priced over $1.00/share. It’s important to account for stock price when looking at the trade size distribution, because a 100-share trade of SIRI is an exchange of ~$600, but a 100-share trade of TSLA is an exchange of ~$42,000 (70x larger in notional value but the same trade size).
Despite the infinite number of potential trade sizes, the plurality of trades across venues are between 100–199 shares. We find that this is broadly consistent across venues regardless of fee structure.
However, one interesting point to note is that trades from 100–199 shares make up more than 55% of the volume on inverted or flat-fee venues, while they make up only 43% of volume on maker-taker venues and 48% on the TRF.
Midpoint Trade Size Distribution — Overall Market *[See Appendix at the bottom of this page for breakdowns of trade size for the TRF, Maker Taker, and Inverted venues]*:
Midpoint Trade Size Distribution — IEX:
Performance by Trade Size
While it’s helpful to know how midpoint trade sizes differ by venue, it’s more important to understand how sourcing liquidity in different trade sizes impacts your execution quality. To measure the impact of trade size on execution quality, we look at trade-to-mid markouts one second after execution of trades on IEX from non-prop firms using our midpoint order types, Midpoint Peg and Discretionary Peg. (Markouts are a popular execution quality measurement that look at the change in stock price a certain amount of time after a trade occurs. In this example, we look at a stock’s change in midpoint price one second after execution as a percentage of the stock’s spread from the perspective of the liquidity adder.) In the industry, markouts closer to 0% of the spread are typically considered better, since they indicate minimal market movement after the trade.
Below, we look at markouts by trade size and stock price to analyze if preventing smaller trade sizes results in less adverse selection. If MQ helped limit adverse selection, we would expect round lot trades over 100 shares to have better markouts than odd lot trades and for markouts to improve (trend closer to 0%) as trade size got bigger, as MQs are usually set at round lot increments equal to or over 100 shares.
However, as you can see below, markouts only slightly differ between different trade sizes. The only clear pattern here is that trades over 1,000 shares tend to have worse markouts than smaller trades. This implies that using MQs likely does not offer material protection against adverse selection on IEX. The data here reflects that there is a negligible impact on the amount of adverse selection you prevent by opting for larger trade sizes.
Trade-to-Mid Markouts by Trade Size:
Performance by Counterparty Trade Size
Thus far, our data does not suggest MQ use reduces adverse selection, as larger trade sizes do not appear to correlate with less adverse selection. However, to get a more complete look at adverse selection, we must also take the counterparty size into account.
IEX is in a unique position to provide data on the size of counterparties because, as an exchange, we see both sides of every trade. Looking at counterparty order size is important because if we just look at trades from the adder’s perspective, the adder’s order size is always going to be the maximum size possible for that trade. We find that the counterparty’s intention, which is more accurately represented by their order size than the size of the trade that occurred, is crucial when considering adverse selection.
Let us demonstrate using two scenarios. In both scenarios, you are a trader looking to buy 100 shares of a stock.
In scenario one, your counterparty is a 1,000-share order. If you trade your 100-share order against a 1,000-share order, your counterparty is only partially satisfied by the 100 shares you traded with it and will continue to trade, pushing the stock to prices that would have been more favorable than where you traded (i.e., adverse selection).
In the other scenario, you trade against another order that is just 100 shares. Your 100-share order is completely filled and your counterparty’s is too. Here the demand of both parties is satisfied (at least for the time being) and neither party is likely to face adverse selection.
The probability of being adversely selected (measured in markouts) differs materially when trading against someone who has 100 shares to trade, versus trading against someone trading a 1,000-share order. Below is a table of trade-to-mid markouts by counterparty size and stock price on IEX.
Trade-to-Mid Markouts by Counterparty Size:
This suggests using a high MQ may result in more adverse selection because restricting the size of the counterparty you are willing to trade with, by design, increases the probability that your counterparty will be larger than you.
Both methods we use, looking at trade size and counterparty size, show you are more likely to be adversely selected against when you limit yourself to only large trade sizes.
Odd Lots vs. Round Lots
To round this out (no pun intended!) we examine differences between odd lots and round lots. MQ is often used specifically to delineate between odd lot and round lot contras, so this provides another look at the impact of MQ on adverse selection.
First, we break out the percentage of midpoint trades executed in odd versus round lots, both in the overall market and on IEX. IEX and inverted venues have far fewer odd-lot trades than the market in general, and the TRF has far more odd-lot trades than the overall market. It is worth noting that far more odd-lot trades occur in higher-priced securities than lower-priced securities, both in the overall market and on IEX.
Overall Market Breakdown of Midpoint Odd Lot vs. Round Lot:
IEX Breakdown Odd Lot vs. Round Lot:
Again, if MQ helped protect against adverse selection, we would expect round-lot markouts to be much better than odd-lot markouts. However, we find that while odd-lot trade sizes do result in slightly worse markouts, the difference is only ~1% of the spread. This extremely small difference suggests there is hardly any reduction in adverse selection when limiting trading to only round-lots.
Trade-to-Mid Markouts by Trade Size:
The advantage of limiting your trading to round-lots was negligible when looking at markouts by trade size, but when looking at counterparty lot size, the pendulum swings in the other direction. We find markouts are actually better against odd-lot counterparties than round-lot counterparties, which is consistent with our prior findings that larger trades tend to result in higher adverse selection.
We hypothesize that this is because when your counterparty is an odd-lot, you are likely the larger party in the trade. If you remember the scenarios from counterparty trade size in the previous section, if the adder’s demand is equal to or greater than the demand of the contra, the adder is less likely to be adversely selected. This further indicates that using MQs to “weed out” odd-lot counterparties is not a good method to prevent adverse selection. Wouldn’t you rather trade against someone who is smaller than you if you are trying to minimize adverse selection?
Trade-to-Mid Markouts by Counterparty Size:
Increased Algo Use and Less Effective MQs
Traditionally, MQs are used to avoid trading with small (often odd-lot) counterparties, but our data suggests that interacting with these small counterparties results in very little adverse selection and less than what you experience when trading with large counterparties. This dynamic may be explained by the increase in algorithmic trading in recent years. As algorithms employ increasingly creative ways to source liquidity through slicing up large orders into smaller “child” orders, it becomes more difficult to use counterparty size as a proxy for counterparty type.
This means that setting a MQ on your orders in the current market may be more likely to exclude counterparties you are trying to interact with than it would have previously. We will delve further into this topic in parts 2 and 3 of this series.
Minimum Quantities Do Not Prevent Adverse Selection
In conclusion, our analysis on IEX suggests that MQ use may not be a very effective tool to prevent adverse selection, and, in some cases, high MQs can increase adverse selection. Our data demonstrates greater adverse selection in trades over 1,000 shares and no benefit from limiting your trading to round-lots.
Employing trading techniques like MQ can be detrimental to performance if sub-optimally implemented, so traders must consider all aspects of this strategy. This analysis only looks at adverse selection, but MQ can be a useful tool to improve other aspects of execution quality. In the next part of this series, we will investigate information leakage.
TRF Breakdown by Midpoint Trade Size:
Maker-Taker Venues Breakdown by Midpoint Trade Size:
Inverted/Flat Fee Venues Breakdown by Midpoint Trade Size:
TRF Odd Lot vs. Round Lot Breakdown:
Maker-Taker Venues Odd Lot vs. Round Lot Breakdown:
Inverted/Flat Fee Venues Odd Lot vs. Round Lot Breakdown:
 IEX classifications of non-prop trades are on a best efforts basis by Member firms’ trading sessions. This method is used throughout for graphics that measure markouts.