Actuarium

Formula sheets

Every formula on the syllabus, rendered natively, with the traps examiners set for each. Print one per exam and keep it by your desk.

MAS-I Modern Actuarial Statistics I

CAS

Poisson process with rate λ\lambda: N(t)Poisson(λt)N(t)\sim\text{Poisson}(\lambda t); inter-arrivals Exp(1/λ)\sim\text{Exp}(1/\lambda); thinning gives independent Poisson processes with rates piλp_i\lambda.

MLE asymptotics   θ^˙N ⁣(θ,  I(θ)1)\;\hat\theta\dot\sim N\!\left(\theta,\;I(\theta)^{-1}\right), I(θ)=E ⁣[2/θ2]I(\theta)=-E\!\left[\partial^2\ell/\partial\theta^2\right].

Likelihood ratio test   2(10)˙χΔp2\;2(\ell_1-\ell_0)\dot\sim\chi^2_{\Delta p}.

GLM   g(μi)=xiβ\;g(\mu_i)=\mathbf{x}_i^\top\boldsymbol\beta, deviance D=2ϕ(satmodel)D=2\phi(\ell_{sat}-\ell_{model}); AIC =2p2=2p-2\ell, BIC =plnn2=p\ln n-2\ell.

AR(1)   Xt=ϕXt1+εt\;X_t=\phi X_{t-1}+\varepsilon_t: Var(X)=σε2/(1ϕ2)\operatorname{Var}(X)=\sigma^2_\varepsilon/(1-\phi^2), ρk=ϕk\rho_k=\phi^{k}.

Traps to remember

  • Mixing rates per hour and per day in Poisson-process questions.

  • Reporting I(θ)I(\theta) instead of its inverse as the variance of the MLE.

  • Using degrees of freedom equal to the number of parameters in a nested test instead of the difference.

  • Stationarity conditions: AR(1) needs ϕ<1|\phi|<1; MA(q) is always stationary.

Ask the tutor