Benchmarking
Calculating Credit Valuation Adjustment (CVAs)
Credit valuation adjustment (CVA) is a regulatory necessity and a significant computational bottleneck for investment banks. It is a derivatives pricing adjustment used by investment banks and other financial organizations to adjust the price of a transaction to account for the credit risk of that capital during the lifetime of the transaction. Basel III requirements demand high-precision risk-weighted asset (RWA) calculations, which are typically computed using Monte Carlo methods and CVA is typically computed by solving the Hull-White model for interest rates using Monte Carlo methods. This example highlights the dramatic improvements in efficiency achieved by moving CVA computations from a traditional computing platform, to run on Signaloid’s compute platform. The example computes the unilateral CVA of a two-trade interest rate swap portfolio, a process as critical for compliance as it is taxing on infrastructure.
Figure 1. Monte Carlo UQ implementation
Running on Signaloid’s Cloud Compute Engine (SCCE) removes the "sampling tax" of traditional finance modeling. By performing a direct computation on the probability distribution rather than iteratively simulating paths, SCCE allows applications running over it to achieve the required output distribution in a single pass without iterative sampling.
Figure 2. Signaloid UQ implementation
A CVA kernel, running on a single-threaded Signaloid C0Pro-M+ core, achieves runtimes 697× faster than an optimized C-language Monte-Carlo-based implementation (i.e., the industry standard and status quo) running on an Amazon EC2 R7iz instance. With 95% confidence, a 1.1M-iteration Monte Carlo implementation matches the accuracy of the Signaloid UxHw-based version; if requiring higher confidence in the accuracy of the Monte Carlo compared to UxHw, the speedups of UxHw are even greater. This acceleration makes real-time risk modeling feasible, enabling banks to monitor risk instantaneously rather than waiting for overnight batch runs.
Key Performance Indicator
Signaloid Platform Solution
Competing Solution
Signaloid Benefit
Speed for the same uncertainty quantification accuracy.
Run existing non-Monte-Carlo code and use either the Signaloid Compute Engine's automated ingestion of distribution information, or use the Signaloid UxHw API to set program variables as probability distributions.
Run existing Monte Carlo code, or, starting from non-Monte-Carlo code, modify code to implement Monte Carlo sampling, iteration, and aggregation of the results from the Monte Carlo iterations of the computation.
697x faster execution time than 1.1M-iteration Monte Carlo, while achieving same fidelity of full distribution result.
Plot of output distribution when running on a Signaloid C0Pro-M+ core that provides the 697× speedup.
Plot of the output of an 1.1M-iteration Monte Carlo for this use case. Signaloid C0Pro-M+ core is 697× faster than this Monte Carlo, while achieving the same or better Wasserstein distance to the ground-truth (20M-iteration) Monte Carlo.
Plot of ground truth (20M-iteration) Monte Carlo.
Benchmarking Methodology
Monte Carlo simulations work by statistical sampling and therefore each multi-iteration Monte Carlo run will result in a slightly different output distribution. By contrast, Signaloid's platform is deterministic and each run produces the same distribution for a given Signaloid C0 core type.
The performance improvements are calculated by comparing Signaloid's platform with a Monte Carlo simulation of a similar quality of distribution. First, we run a large Monte Carlo simulation (about 50M iterations) on an AWS r7iz high-performance AWS instance: We use this distribution result as a baseline or ground truth reference of distribution quality. Then we calculate the performance of Signaloid's technology, and compare it with the performance of a Monte Carlo iteration count where the output distribution's Wasserstein distance (to the output distribution of the ground truth reference) is as accurate as the Signaloid-core-executed algorithm's output distribution, with 95% confidence level.
Performance data based on Fall 2025 release of Signaloid's technology.