The advantage of a dashboard view is that it can quickly highlight interesting information. Looking at the SDRView Res dashboard this morning, I was surprised to see the following:
Showing massive EUR IRSwap volumes on 19-Nov and 21-Nov, which were larger than USD volume on those days! Very strange as this never happens. Remember in SDRView we source trade data from US SDRs at DTCC and BSDR, which represents trades reported by US Entities, so USD IRS Swaps are (almost) always the highest volume product reported.
What is going on? Lets investigate.
How Large are the Spikes?
Lets go back and look at the past 3 months of daily history in EUR IRS.
Which shows that:
- Nov 19 has > $280 billion
- Nov 21 has > $195 billion
- No other day is anywhere near close to these figures
Off SEF and Uncleared
Using the SDRView Swaps Swaptypes view see that the vast majority of this volume is in Off SEF Uncleared Back Starting trades.
We can see that this is an anomaly by looking at the history of Off SEF Uncleared Swaps.
And looking at a table of Off SEF Uncleared trade counts:
We see a very high number of trades on these two days.
Can there really be 2,461 trades of $195 billion notional executed on 21-Nov, as we are led to believe from the execution time stamp of these trades? Unlikely given that SwapClear reported $250 billion of Cleared EUR IRS on 21-Nov and CME reported $7.4 billion. The trade counts and volume is just too high for that to be the case.
If we drill-down to the list of trade details and sort by effective date (start date of the swap), we see the following:
Which shows that:
- The effective dates are almost all in the past
- 2,318 trades out of 2,461 trades have effective dates prior to 21-Nov-14
- Suggesting these are aged (back start ) trades that were executed in the past
- There are always 2 trades with the same effective date and end date
- The notional is either the same or it is very different
- The price of the two trades is either the same or very different
- The prices do not look like prices of trades executed on 21-Nov, but very old prices
- Sometimes one of the prices looks like a current par rate while the other does not
So what does this tell us?
Well your guess is as good as mine.
It looks similar in footprint to a compression activity that replaces old trades with new, covered in Swap Compression on SEFs.
However it is different. (We don’t see any additional fee as we do in compression list trading).
Time is running out for me to publish this blog, so I will leave it there and think some more and update tomorrow.
If you know the reason please let us know.
In the meantime this is a reminder that spikes in data need to treated carefully.
They may not be due to real trading activity; activity that is price forming.