Options can be priced by the lattice model, the results of which converge to the theoretical option value as the lattice’s number of time steps approaches infinity. The time complexity of a common dynamic programming pricing approach on the lattice is slow (at least ), and a large is required to obtain accurate option values. Although -time combinatorial pricing algorithms have been developed for the classical binomial lattice, significantly oscillating convergence behavior makes them impractical. The flexibility of trinomial lattices can be leveraged to reduce the oscillation, but there are as yet no linear-time algorithms on trinomial lattices. We develop -time combinatorial pricing algorithms for polynomial options that cannot be analytically priced. The commonly traded plain vanilla and power options are degenerated cases of polynomial options. Barrier options that cannot be stably priced by the binomial lattice can be stably priced by our -time algorithm on a trinomial lattice. Numerical experiments demonstrate the efficiency and accuracy of our -time trinomial lattice algorithms.
While ordinary options are wholly exercised at a chosen time to maximize holders' benefits, empirical evidence favors the intertemporal exercise of corporate securities such as convertible bonds and executive/employee stock options. This is because the exercises of these securities can dilute the issuers' equity, inject capital into the issuers' assets, and change the capital structures, which influences the exercise payoffs and the remaining security values. However, the literature simplifies these impacts on exercise policies, leading to biased and even irrational pricing results. We address this via a state-transition forest that partitions the securities into k units and the time span of the forest into n steps. A representative holder can exercise an integral multiple of units at different steps to maximize her benefit, given that the lump sum of the exercise units does not exceed k. It can be further extended to model intertemporal exercises conducted by competitive investors who hold too-small amounts of securities to influence prices. The forest is then composed of k + 1 trees to capture transitions among capital structures due to different lump sums of option exercise units. The intertemporal exercise decision at each forest node is optimally determined by recursively solving the Bellman equation. This carefully designed forest structure reduces the running time complexity from to . We show that ignoring or simplifying intertemporal exercises and corresponding dilutions/injections significantly influences pricing results.
Estimation of pricing kernels has attracted substantial attention following the emergence of the "pricing kernel (PK) puzzle". When the physical density is estimated from historical market returns while the risk-neutral density is derived from option prices with forward-looking information set, one often obtains a non-monotonic PK slope. The Conditional Density Integration (CDI) estimator seeks to address this mismatch by incorporating option-implied forward-looking information to recover theory-consistent kernels. However, we find that CDI does not completely resolve the puzzle; relative to traditional methods, it can still generate non-monotonic kernels that contradict the standard expected-utility theory for risk-averse investors. Moreover, monotonicity depends critically on assumptions of constant market volatility and is highly sensitive to smoothing parameters. Evidence from S&P 500 and Bitcoin options confirms that volatility jumps undermine monotonicity. Our findings highlight fragility and stability challenges in pricing kernel estimation.