QuarcHome
Model catalogueBack to the model index

Transition Densities and Kolmogorov Pricing Equations

The forward equation evolves the distribution of a Markov state from today to the payoff date. The backward equation starts from the payoff and evolves its value back to today. Under compatible assumptions, both routes price the same contract. A comparison can reveal lost probability mass, an incorrect boundary or an error in the discounting step.

How it works

The forward Kolmogorov equation evolves the probability density from a specified initial state. The backward equation propagates the value of a payoff from its final condition to today. Under the same Markov dynamics, boundary conditions and discounting, integrating against the evolved density and solving backward recover the same price.

Worked example

For Brownian motion starting at zero with annual variance one, the one-year terminal density is standard normal. A cash payoff of one if the terminal state is positive is worth 0.5 with zero rates. A forward density integral, a backward solver and simulation should all agree up to their numerical error.

Validation and limitations

Check probability mass, units and boundary handling before comparing prices. For absorbing states, account for mass at the boundary separately from the continuous density. Validate grid convergence and compare with a payoff whose value is known analytically.

Further reading