What the SEF Volumes Show
The first calendar week of SEF trading is behind us. If you are like me, you are wondering where the liquidity was / who attracted the most volumes over this period. I was inspired by Kevin McPartland's blog on SEF market data, so I have gone out and collated the data from the various SEF public reporting websites for the first three days (Wed-Fri).
YOU'D THINK IT WAS EASY
I was surprised to find that gathering this information is no easy task. To start with, some SEFs just dont seem to have this information publicly available. Then there are some that make it public, but only have pricing information, no volumes; do I interpret this as there were bids and offers but no volume? Further, no 2 SEF's report the information in the same way. Some use CSV files with dates on them, others use PDFs, there are even XML files. Many use strange product codes. Some quote notionals but dont tell you for which ccy of the FX Pair, or in what terms (eg 1,000s, etc). Many of these report the data only for the previous day. I found myself reaching out to many of them through helpdesk or email support links to help navigate their data. Many of them have responded quickly.
As usual when interpreting data, it is good to point out the assumptions and possible stumbling blocks: 1) All volume data are in millions of USD. 2) I restricted the world of data to just USD volumes. Main reason for this is it minimizes the error interpreting the data. The SEF's are not very clear how they are expressing the data (eg in ccy amounts or USD equivalent). USD volumes should be a good measurement. 3) If the table shows a zero (0), it means there was data published but there was no volume or less than 500,000. 4) If the table shows a hash (#), it means there was pricing data published, but no volumes. I have contacted these SEF's to ask for volume data. 5) If the table shows a double-question-mark (??), it means there was some problem retrieving the data (mostly on my end!). Primarily this happened when the SEF only publish daily data, and I didnt save the files on the appropriate day. But it also happened because it appears they did not report volume for a particular day yet. 6) While I collated the EQD and CMD volumes, they were insignificant, so I decided to remove. 7) Don't forget that CME and LatAm SEF are not yet approved by the CFTC. You can see the current registration status of SEF's in my most recent blog here. Further, the CME is only intending to support CMD for their initial phase. 8) There is no common format for this data, so I found myself normalizing the data. I may have made a wrong assumption or error in doing so (Refer to section above "You'd Think It Was Easy" for more on this).
INTERPRETING THE DATA
So, what to make of all this data? What jumps out is that Bloomberg seem to have the most volumes. Beyond the standard "everyone has a Bloomberg terminal" comment, I don't have much to add, except that I am slightly surprised. The big IDBs have a good showing. GFI is a solid second in Credit, which does not surprise me given their position in the classic IDB CDS world. Likewise ICAP having 70% of the IRD volume makes perfect sense. The seemingly low numbers on Tradeweb and MarketAxess stumps me a bit. And I do suspect that the other IDBs (TP and Tradition) would have some significant pieces of the pie. Lastly, Reuters / FXAll only having 1% of the FXD volume? I have double-checked my math. A wrong assumption in interpreting USDIDR volumes, for example, means I could be off by a factor of 11,500! I think I have interpreted the data correctly.
If anyone is keeping score at home, has any clarifications to make, any justifications, or simply any comments, please leave them below, as I would like to make any corrections and make this as accurate as possible. TOD SKARECKY