Clarus Financial Technology

Performance of Block Trades on RFQ Platforms

We’re looking at the price dispersion around the time of block trades on different SEFs this week. This is just a limited initial analysis of 10 year trades, allowing us to present a simplified framework. But this framework can be extended to be maturity agnostic by using our DV01 statistics – or we could constrain this and use DV01 measures across a range of maturities – e.g. 7-13 years.

Block Trades Recap

Last week, we found that due to the Block Trade NARL, customers can choose to transact large trades on an RFQ basis and do not have to ask 3 market participants for a price. These trades cannot go through a SEF’s Order Book, so we know that D2D platforms do not report Block Trades. We therefore know that a Block Trade reported to BSDR was transacted on BSEF, a Block Trade reported to DTCC was transacted on Tradeweb and that all other capped trades reported to DTCC were probably executed on a D2D Order Book.

Given that we also know that the RFQ platforms (BBG and Tradeweb SEF’s) have seen significant volume growth this year, and that their reported block trading volumes have gone up, it is probably fair to say that many of their buy-side clients are happiest with this RFQ-to-a-limited-number-of-dealers. Or at least, they see this as the least worse/easiest choice under the current swap market structure.

Measuring Performance

So this opens a new field of analysis for us as neutral market observers. Our first port of call is to try to measure performance of these block trades by platform. For that, let’s try to create a benchmark for price impact.

Price Diffusion

In theory, prices should follow some kind of random walk/stochastic process. So if we look at price changes between consecutive (or systematically non-consecutive trades), and have a large enough population of trades, we might see a bell-curve reminiscent of a normal distribution for these price changes. So let’s start there today and develop these ideas over some coming blogs.

Trade Population

I’ve used a similar data-set to the one underlying “The Scariest Chart”. On-SEF, USD trades and for the purposes of illustration, 10 year tenors designated by the USSW10 Bloomberg ticker that we augment our data with. I’ve taken August 2015 as my time window.

I’ve collated a time-stamp ordered data set, and looked at the price change, in basis points, between trade T and trades T+1, 2, 3, 4, 5 – to identify the subsequent price impact across the next five trades.

I’ve also looked at the price impact to trades T-1, 2, 3, 4 and 5 just in case there are any nefarious tactics such as front-running evident in the data! I’d be surprised if this were the case, but it seemed a simple addition to the analysis – and (in the future) allows us to test the “T plus” data-set to verify if there are any structural differences versus the “T minus” data-set.

Results

Plotting the histograms for these data sets yields a series of bell-shaped curves. I’ve capped the price jumps at +4 and -4 b.p between trades – as these most likely represent end-of-day to start-of-day jumps. Charts below:

Price changes between consecutive swaps create a series of bell-shaped curves.

Some background statistics for this population of trades are:

And now we can drill-down deeply into the statistical properties of the above price distributions:

This background is all well and good, but what does it mean for markets? Essentially, we can use a number of statistical observations to programmatically measure the “score” for the predictive powers of a given trade. We can therefore assess the likelihood of any particular price action across a future time horizon. The uses are broad:

Measuring the 3 SEF Venues

So, having gone about this process, what do we actually see in the data for August 2015 for Tradeweb, Bloomberg and D2D SEFs handling large trades? Below I have reproduced the previous price distributions, but this time with the 3 platform’s individual price-series overlaid. Remember, these are only for trades of block size:

Tradeweb, Bloomberg and D2D Price distributions during August 2015

A couple of notes to first clarify this chart:

But for the purposes of this blog I think that chart illustrates my point pretty well:

Increased Rigour

For the purposes of this blog, we’ve looked at a single month’s worth of data across a single Bloomberg ticker. The analysis can be improved by looking at:

And when combined with your own private data, this stream of analysis becomes infinitely more powerful:

The SEF Score-card

Bringing this back to the SEF state-of-play, it’s interesting what transparency this offers us as swap market participants.

Given your history of trading by dealer and venue, it is important to understand where your best quality of execution has occurred. In particular:

The Clarus data gives you these answers. Just contact us for more information. And remember to subscribe to stay on top of these topics and more.

Stay informed with our FREE newsletter, subscribe here.

Exit mobile version