Odd Lots – Questions and Patterns
This past Monday, the consolidated tape began publishing odd lot executions.The Street’s general consensus was that their inclusion would contribute to 3% – 6% increase in reported volume as a whole. The actual numbers appear to be more than that – but we’ll get to that in a bit.
Back in 2011 we wrote a note titled Odd Lots – Another Weapon in the HFT Arsenal. We featured a paper written by Cornell Professor Maureen O’Hara, titled What’s Not There: The Odd-Lot Bias in TAQ Data. That paper correctly asserted that public market data was missing a substantial subset of trades that contained valuable information – information that is available in the direct data feeds. From her paper:
“In this research we investigated the odd-lot bias in TAQ data. We have demonstrated that missing trades are a large and pervasive problem in TAQ data. That trade sizes are truncated below 100 shares means there is a censored sample problem for all stocks. For some stocks, however, this problem is acute, with as much as 40% or more of trades missing from the data. Moreover, these missing trades are highly informative, meaning that analyses of issues related to market efficiency are also subject to error…Our analysis shows that odd-lot trades are now far from unusual, and market practices such as algorithmic trading and high frequency trading are only increasing their incidence.”
“We believe our results also have important policy and regulatory implications. TAQ data is biased because the consolidated tape is biased: odd-lot trades are not recorded to either data source. When odd-lots were a trivial fraction of market activity, this omission was of little consequence. But new market practices mean that these missing trades are both numerous and informationally important. Moreover, while these trades are invisible on the consolidated tape, they are not invisible to all market participants. Market venues now sell proprietary data that allow purchasers to see all market activity. Our results suggest that odd-lot trades now play a new, and far from irrelevant, role in the market. The SEC should recognize this new role and change the reporting rules regarding odd-lot trades.”
“Because odd-lot trades are more likely to arise from high frequency traders, we argue their exclusion from TAQ and the consolidated tape raises important regulatory issues.”
The SEC listened to Professor O’Hara, given her extensive and impressive credentials, and so for the past two days we have born witness to their inclusion.
What Have We Found?
Generally, we are seeing oddlots as a percentage of volume and as a percentage of trades being even greater than O’Hara found in 2011, and greater than what many other brokerage firms had predicted. See this table, where you can see Tuesday’s trades in over 7,000 stocks broken down by symbol.
Why would Noble Energy (NBL), a $69 dollar stock that trades 2.5 million shares per day see 25% of its trades be odd lots? Isn’t that odd (no pun intended)?
Additionally, we were curious if there are any patterns we might be able to see in odd lot trading. For example, are odd lots likely to take place more on one exchange or venue than another? Also, are there some quantity odd lots that are more frequent than others? For this help we turned to Eric Hunsader at Nanex. See the below chart:
First, note how frequent odd lot trades of less than 10 shares take place, regardless of exchange or TRF; they are the most common odd lots, and are seen not only in uber-high-priced stocks. Secondly, note that odd lots tend to trade on NASDAQ (red), the NASDAQ TRF (teal) (dark pools), and ARCA (purple) most frequently, and in that order. Note that they take place most infrequently on inverted exchanges. Note that the number of odd lots that trade on NASDAQ also is easily double the number of oddlots on ARCA alone.
Eric also put together this nice visual chart that demonstrates the out-sized representation of single-share odd lots.
These charts raise some questions. Is there a direct relationship between a high incidence of odd-lot trades and exchanges that pay rebates? If so, why is that? If these odd lots were simply retail order flow, wouldn’t one expect to see a higher incidence on inverted exchanges, as those exchanges pay rebates to get taken, and surely order routers would be expected to take advantage of those rebates? Additionally, do odd lots result mostly from a resting round lot being hit by a an odd lot? Do odd lots result because resting orders start their lives as resting round lots, and get pinged into being an odd lot?
Could it be that this data infers that odd lots are not used predominantly by retail, but instead by high speed traders with some other purpose? What would that purpose be? Morgan Stanley recently published a note titled the Odd Lot Cascade Effect. Please reach out to your Morgan Stanley representative to get that note. The short note demonstrates that a resting round lot order (100 shares) which gets hit with a partial odd lot order (rendering the remaining resting order itself an odd lot), it loses price protection under Reg NMS. Could high speed traders be using small oddlots to degrade a resting front-of-the-queue round lot to a status where it is not protected by the NBBO, and therefore the orders behind it would benefit from a quasi-queue- jump if you will?
These are interesting questions. We hope the SEC looks at their data with a skeptical mindset; perhaps they can confirm such patterns, or others we have not been able to discern as of yet. The SEC then will potentially see how distortive the video game of rebate arbitrage actually is, where the collection of rebates has become so important to high speed traders, that they would willfully engage in games that deter real liquidity providers from ever displaying a public quote.