A High Performance Computing of Internal Rate of Return Using a Centroid Based Newton Raphson Iterative Algorithm

A popular financial metric in estimating the profitability of a project or investment is the internal rate of return. However, the IRR variable cannot be easily isolated from the equation. This is effectively solved by using iterative root-finding algorithms, some of the most frequently used of which are secant, bisection, false position, and Newton-Raphson algorithm. Although the Newton-Raphson method is considered to be the fastest to converge and the most popular method, it still requires an initial guess value from the user, which could result in the algorithm to not converge to the root if the user input is far from the actual root. This issue is addressed by a midpoint-based newton-raphson technique, which sets the midpoint of cash flows as the initial guess input. However, the midpoint technique is static as it does not adjust with unequal cash flows. This study presents a centroid-based newton-raphson algorithm in estimating IRR, which dynamically takes into consideration the values of cash flows. The experimental results show that the proposed algorithm ensures convergence by producing an initial IRR with an accuracy of 91.41%. This indicates that it is 26.75% more accurate in approximating the initial IRR than the midpoint-based newton-raphson algorithm. It also reduced the required iterations of convergence by 35.33% over the midpoint-based newton-raphson algorithm. These findings show that the employment of the centroid-based newton-raphson algorithm in approximating IRR provides a significantly better approach in evaluating investments than the current method.