Quant with Vahab
Quant Systems Lab · Control Systems for Quantitative Finance

Change of Measure (Girsanov)

Change of measure turns one drift into another by reweighting paths, while preserving Brownian noise.

Explanation

Under Girsanov, we tilt probabilities by a density process so that drift terms change and the noise stays Brownian.

Risk-neutral pricing is a change of measure from the physical measure P to a martingale measure Q.

The Radon–Nikodym density encodes how unlikely a path under one measure looks from the viewpoint of another.


measure changerisk-neutralmartingalegirsanov
Interactive visualisation

Under Girsanov we tilt path probabilities so that drift terms change while the Brownian noise structure is preserved. Here S_t under the physical measure P has drift μ, and under the risk-neutral measure Q it has drift r; the same noise path generates both.

Price paths S_t under P (μ) and Q (r) using the same Brownian incrementst=0.0t=0.2t=0.4t=0.7t=0.9S₀P: drift μ = 6.0%Q: drift r = 2.0%
Radon–Nikodym density Z_t = dQ/dP|_t; weights paths when changing from P to QZ = 1
Numbers
θ = (μ − r) / σ ≈ 0.20
Eₚ[S_T] ≈ 106.18 (drift μ)
E_Q[S_T] ≈ 102.02 (drift r)
Z_T ≈ 0.980 (end-point density)
Interpretation

Changing μ and r tilts the blue and orange paths apart: under P the asset drifts with μ, under Q with r, but both follow the same noise realisation. The density process Z_t reweights this noise path so that expectations under Q can be computed as weighted expectations under P.

When μ is far above r, θ is large and Z_t changes strongly over time: paths that look favourable under one measure become relatively unlikely under the other. Risk-neutral pricing is just one particular change of measure chosen so that discounted prices become martingales.