In a perfectly liquid market, investors’ optimal allocation decisions refer to maximizing all three dimensions of liquidity, namely immediacy, width and depth. To the extent that investors fail to accommodate size (depth) along with price (width) in their optimal allocation decisions, their overall costs may increase. This paper focuses on the substitution of width and depth by investigating the simultaneous determination of price clustering and size clustering in the credit default swap (CDS) market. We report strong evidence that when traders round prices they tend to quote more refined sizes, and vice versa. The findings highlight a clear trade-off between price clustering and notional amount in the CDS market, and contribute to the emerging literature on size clustering.