Portfolio trading (PT) has grown significantly in corporate bond markets, and it creates some meaningful complications for transaction cost analysis and market data.
Portfolio trades are executed as a package, often with a single all-in price or a set of prices negotiated simultaneously across dozens or hundreds of bonds. The dealer takes the portfolio risk as a whole, meaning individual bond prices within the package may reflect their cross-subsidisation between liquid and illiquid bonds, a portfolio ‘premium/discount’ rather than standalone fair value, and prices that may clear the dealer’s risk appetite rather than the market’s marginal price.
Collectively these can make it hard to benchmark individual legs against a market price using standard TCA methodologies.
Traditional execution metrics that are used in equity markets such as arrival price and volume weighted average price (VWAP), along with more fixed income measures such as estimated bid-ask spread, typically assume bonds are traded individually or in small groups.
As PT sees the decision to trade and the execution happening nearly simultaneously for all bonds there’s no sequential arrival price per bond and the bid-ask spread ‘paid’ reflects a blended concept, not separable per instrument.
PT is often used precisely because some bonds in the basket are illiquid. A TCA framework that treats each bond independently would flag illiquid legs’ execution quality differently, based uoon being the buyer or seller, even though the portfolio mechanism may have improved achievable prices for those bonds versus trading them individually. Consequently, TCA needs to assess the counterfactual cost of alternative execution across the whole basket, which is methodologically demanding.
A dealer can price PT using their own inventory, hedging costs, and correlation assumptions. Buy-side TCA measures often lack visibility into this, making it difficult to assess whether the all-in price was competitive without running an in-comp PT process to multiple dealers for comparison. The perceived value of trading in-comp vs non-comp differs considerably between buy-side trading desks, and even between buy-side traders.
TRACE reporting and its effect on public bond data
This is where the data distortion becomes particularly important. FINRA requires that each individual bond leg of a portfolio trade be reported to TRACE separately, and critically, since 2022 FINRA added a portfolio trade indicator flag to TRACE reports.
This means each bond appears as an individual trade report in TRACE with a price and size with the PT flag identifying it as part of a packaged trade, making PT legs distinguishable from ordinary secondary market trades.
The risks that PT pricing has on the wider market si that they reflect package-level negotiation, not standalone bond liquidity. Consequently including them in spread or price benchmarks can overstate or distort market pricing, while large PT baskets can inflate reported volume in individual ISINs, making bonds appear more actively traded on a net basis than they are on a standalone basis. Collectively this impacts perceived liquidity and cost.
Dated data
Older academic and commercial TCA models that estimate bid-ask spreads from TRACE, such as the 1980’s ‘roll spread’ estimator and the Corwin-Schultz liquidity proxy from 2012, will be polluted by PT prints, since these prices do not reflect normal dealer quote behaviour.
Historical TRACE data from the rapid PT growth period of 2019 to 2022 will contain an unknown volume of package-trade prints because this was before FINRA introduced the flag. These may distort any historical liquidity or cost analysis. Retrospective studies using this data should consider this with respect to pre-flag historical benchmarks
Better data
Robust TCA for portfolio trades requires that TRACE data be filtered to separate PT and non-PT prints when constructing benchmarks with PT executions evaluated at the basket level, not bond-by-bond.
The use of dealer axes and competitive RFQ responses may be a good indicator or primary benchmark supplemented by prices derived from a consolidated tape, such as TRACE.
Crucially, traders and TCA professionals need to recognising that PT’s value proposition is partly avoiding the market impact that post-trade reporting such as TRACE data would reflect, making pre-trade cost estimation a critical analytical step.
©Markets Media Europe 2025
