Research On The Pricing Of Snowball Structured Products And Their Multi-underlying Variations

In the current market scenario of low interest rates and ample liquidity,snowball options have emerged as a popular choice among investors due to their high coupons and winning rates.The snowball structure is essentially a path-dependent exotic option with embedded barrier conditions,it enables investors to redeem and terminate early when the underlying asset’s price surpasses a specific level.As a result,the longer one holds onto this certificate,the higher their potential income can grow-much like how snowballs accumulate in size as they roll down a hill,this is how the product got its name.Compared to the mature derivative investment environment overseas,the domestic market is still in its early stages of development,lacking liquidity and price discovery functions,therefore,research on pricing of emerging snowball options has profound significance for the development of structured products business.This thesis primarily centers on the pricing methodology of snowball options.Firstly,conduct a systematic review on relevant literature and basic theories.Secondly,analyze in detail the snowball’s payment situation and sources behind its returns,use splitting and replication strategy to estimate option value under Finite Difference(FD)method.Thirdly,build up a Monte Carlo(MC)pricing model,simulate the movement path of underlying assets,take the average of test results as the option price.Finally,expand the research to an innovative mode-worst of snowball options,optimize the Quasi Monte Carlo(QMC)strategy of random number generation with low-difference sequence.By meticulously analyzing the pros and cons of models with regards to convergence and computational time,this thesis offers a fresh theoretical outlook for the advancement of financial derivatives.Through numerical experiments,this thesis tested the performance of three algorithms under the same parameter environment:First,the FD method has a fast convergence speed and short computation time.When the number of grids reaches 700,results tend to be stable.Regrettably,this approach is encumbered with a complex process and plagued by dimensionality curse.Secondly,the MC model boasts a flexible structure and high applicability,enabling it to simulate the complete life cycle of snowball options through experiments.However,calculations can be time-consuming with slow convergence speeds.It is only when the paths quantity exceeds 250,000 that this model attains stability,frequent jumps are observed under multi-dimension scenarios.Thirdly,the QMC method improves the non-convergence weakness of ordinary MC simulation,increases the convergence rate to O(N-1),and reasonable estimations can be obtained for 2D and 3D worst-of snowballs with a path number magnitude of 216.Finally,this thesis summarizes the work done and the conclusions drawn,while also pointing out the limitations during the research process.On one hand,QMC methods can be combined with Principal Component Analysis or Brownian Bridge path generation methods to further improve simulation efficiency;on the other hand,it is hoped that the pricing model established in this thesis can be applied to innovative snowball structures such as step-down barrier type,non-chasing protection type,escape cabin type,etc.