There’s no doubt that the POINT system was a good one. It was developed over many years and with a wide user base. It provides risk, attribution, optimisation and a robust source of fixed income analytics as well as access to Barclays’ wide range of benchmarks. Nevertheless, we think that although it will be very painful to make the move, and by no means underestimating the amount of work required, there may also be opportunities to Point users from a system change.
Unify analytics
The CloudAttribution returns based fixed income attribution model
We often speak to teams where the back office and front office analytics data is sourced from different places. This change provides the opportunity to re-evaluate both to try to consolidate to one source. Often the performance teams are using a different source because it had been price aligned to their accounting data. All eyes on one source saves resources and gives everyone more confidence in the inputs to all downstream systems.
From a performance perspective the increasing popularity of returns based fixed income models, which are less affected by price alignment, and are gaining in popularity, can help too. Their reduced data requirements and particularly their reduced sensitivity to the analytic inputs means that they are worth investigating in the unified framework.
Broader benchmark availability
Being a benchmark provider means that Barclays is able to provide all of their benchmarks to POINT and some useful manipulation tools. But the flip side is that the benchmarks of other providers, with a few exceptions, are not available and that limits the usefulness for many portfolios. Proxies can be used or the benchmarks could be loaded. Neither is attractive, particularly when some providers explicitly prohibit it.
There are several companies that have universal benchmark sourcing with flexible feeds for most downstream systems, and many systems now have good benchmark manipulation tools. This shouldn’t any longer be a reason to choose one system over another.
Reduce the residual
Reducing the residual on any yield based attribution system is hard but not impossible. On a full yield based system it requires a lot of work and a lot of data. It also requires a lot of replication of work already done elsewhere.
KRD based model: We have mixed a yield based KRD model with a returns based model to get the best of both. We start with the accounting data so that the residual should be zero, but use the key rates to provide more detail for the duration management terms. Using DTS on the spread management side, and a detailed classification structure mimics the results from Barclays POINT.
Returns based model: Working completely in return space substantially reduces the data requirements. The model is also closely aligned with the way portfolios are managed, being an extension of the van Breukelen methodology. It leads to a significant reduction in complexity for a similar level of insight.
Improve portfolio coverage, transparency and availability
More access to benchmarks, more accuracy for less work and a smaller analytics requirement all make it easier to consider putting most funds on a system rather than just a few headline portfolios. At the same time, reporting systems and computing generally have moved a long way since POINT was conceived. Now having results on every desktop in an interactive form, managed by the performance team is the norm. We think that this offers team the perfect excuse to move into the 21st century!
