Local volatility model Created 2026-10-06 Updated 2026-11-05
A local volatility model specifies a deterministic function of time and current spot, with pricing dynamicsfor a non-dividend-paying stock. Unlike a stochastic volatility model, it introduces no separate random variance state.
A standard completeness result assumes a well-defined model in the usual natural Brownian filtration, a positive bank account, and nonzero traded diffusion exposure for almost every time and path. The Brownian martingale representation theorem then converts an integrable claim-price martingale into stock gains, subject to the corresponding trading integrability. The Brownian representation replication in a local volatility market gives the holdings explicitly. Positive relative volatility alone is insufficient if the stock can be absorbed at zero, where its traded exposure vanishes.
Log-density domination in an exponential family Created 2026-10-07 Updated 2026-10-14
On a bounded parameter set, has an integrable envelope of a function class whenever is bounded and is integrable under the sampling law. A compact subset of the interior natural-parameter domain bounds the cumulant function of an exponential family and provides moments under model sampling; log-base-density integrability is an additional condition.
Martingale deflator Created 2026-09-24 Updated 2026-11-05
A martingale deflator rescales asset values so that their deflated gains have a stated martingale property. For a price with cumulative cash dividends , the correctly deflated gain process iswith a strictly positive adapted deflator . Dividends are deflated at their payment times; multiplying by the undiscounted total is generally different. A local martingale deflator requires the local-martingale property, while true expectation pricing requires the corresponding true-martingale and integrability conditions.
For a positive bank account , put . If is a true mean-one martingale, it defines an equivalent probability measure by its terminal density. Deflated-price local martingales then yield an equivalent local martingale measure. Obtaining an equivalent martingale measure, and ordinary conditional-expectation pricing, additionally requires true martingales for the asset-price ratios. Positivity and a local property alone do not justify that promotion.
Mean first-passage time Created 2026-09-28 Updated 2026-11-05
The Mean first-passage time to a target is , which can be infinite. For a diffusion with backward generator , a finite sufficiently regular mean solves the Poisson equationoutside , with on the absorbing target. Reflecting parts of the boundary require the reflecting condition of the generator, rather than an absorbing value. These conditions express different physical rules and cannot be exchanged.
For Brownian diffusion with generator on , , absorption at both endpoints givesIf zero instead reflects while absorbs, the boundary conditions are , and the solution is . In particular the mean time from the reflecting endpoint is , displaying the squared-distance diffusion scale.
Dynkin's formula explains the equation: before absorption, has constant expected value. For these bounded-interval nonnegative bounded solutions, this identity first bounds by , proving integrability of ; passing to the limit then proves . Boundary equations without a finiteness or minimality condition can have extraneous solutions on unbounded domains.
Let be an integrable random variable, let be a finite -measurable random variable, and suppose is an integrable random variable. Then a known factor can be taken outside the conditional expectation:For a bounded , first verify the defining integral identity when is the indicator function of a measurable set in , extend by linearity to simple functions, and then use bounded approximation and the dominated convergence theorem. For nonnegative , truncating and using the conditional monotone convergence theorem proves the identity with extended nonnegative conditional expectations. In particular,For general under the stated integrability assumptions, the conditional Jensen inequality bounds by that integrable random variable. Apply the bounded-factor result to truncations of , then use the dominated convergence theorem on the products and the L1 contraction of conditional expectation. This proves the identity without assuming itself is an integrable random variable.
The product-integrability condition is essential to the ordinary finite conditional expectation statement. An unbounded factor cannot be moved through an expected value simply because its partner is an integrable random variable.[ref]
Mixed power moments of a beta distribution Created 2026-10-25 Updated 2026-11-05
For with positive parameters, and real , , multiply its density by . The resulting integral is a beta function with shifted parameters, henceThe restrictions on the powers are exactly the endpoint-integrability conditions. This includes ordinary moments, reciprocal moments when they exist, and quantities involving both the occupied fraction and remaining fraction . In particular, gives .[ref]
Suppose is an integrable adapted process, is finite and adapted, and . The displayed correction is a nonnegative supermartingale: its next denominator is measurable at time , and its reciprocal is at most one. If almost surely, the denominators converge to a finite positive limit since . The almost sure supermartingale convergence theorem then gives finite almost sure convergence of . No integrability of the infinite product or of is needed.
Negative part of a real-valued function 2026-10-14
The negative part of a real-valued function is the nonnegative function . It records the magnitude of the negative values, rather than retaining their sign: if , then . Together with the positive part, it gives and . This decomposition defines the Lebesgue integral of a signed function and makes the integrability condition explicit.
Optimal marginal utility as a one-period pricing density Created 2026-10-07 Updated 2026-10-14
For a negative, strictly increasing, concave differentiable utility with finite expected utility at every holding, a gain having both signs makes the objective coercive at both ends. At a finite optimum its derivative is zero. Secant domination for expected utility derivatives gives integrability of ; boundedness of on the remaining compact set gives integrability of the positive marginal utility itself. Its normalization has expectation one and prices the gain at zero.
Past exam of the distribution theory course of the University of Oxford 2023 2 b iii Solution 2026-10-25
To repeat integration by parts across whole periods, choose periodic primitives with zero mean. Set , and recursively defineEvery has zero mean. Part (i) gives almost everywhere, and because vanishes at both endpoints. Its periodic extension is consequently locally an absolutely continuous function, with no jump at the integers. Part (ii) gives the uniform estimateThe weak derivative identity on the whole line is . After scaling by the integer and repeating integration by parts times,The boundary terms vanish because has compact support in . The interval contains exactly periods, soTaking therefore yieldsThis rapid distributional decay of periodic oscillations uses only integrability of and smoothness of the test function.
Past exam of the mathematics course of the University of Cambridge 2002 iii Paper 31 2 Solution Created 2026-10-03 Updated 2026-10-14
For on ,The Itô formula, stopped in a compact annulus, therefore givesThis proves the local martingale property up to a possible hitting of the origin. For the annulus exit time , is bounded between and , so it is a true uniformly integrable martingale.
The exit is finite almost surely. Indeed is a martingale, so stopping at gives . Letting proves . Bounded convergence in the logarithmic martingale now yieldsIf is the probability of reaching radius before radius , continuity givesHitting zero before reaching radius would require reaching every positive radius first. Its probability is bounded by , which tends to zero as . Any finite-time path hitting zero is bounded before that time and hence hits zero before exiting some integer-radius disk. A countable union over those disks proves for every , almost surely. This is planar Brownian motion avoids a fixed point; it also makes a globally defined continuous local martingale.
To decide whether it is a true martingale, compute its expectation directly. Condition on if necessary and use rotation to take and write . Conditional on , the angle of is uniform. The angular average of a logarithmic potential isFor this follows by taking the real part of the convergent power series for : each nonconstant Fourier term integrates to zero. For factor out and apply the same argument to . The equality at one follows by the integrable limiting logarithmic singularity. All interchanges are valid: the shifted Gaussian density is bounded near the origin, where , and its Gaussian tail controls logarithmic growth.
The radius has density . Consequently, for ,Since , is not a martingale, despite integrability at each deterministic time. This is the logarithmic radius of planar Brownian motion example.
The optional hint is consistent with this conclusion. The upper-radius exit is finite and never occurs at zero. After that exit the stopped square is ; before it, a radius in also gives a logarithmic square no larger than , while a radius below contributes precisely the extra indicated term. This proves the hinted inequality without assuming that the unstopped process is a true martingale.
Past exam of the mathematics course of the University of Cambridge 2002 iii Paper 31 3 Solution Created 2026-10-03 Updated 2026-10-14
A simple predictable process can be writtenwhere each is bounded and -measurable. Define its integral to start at zero:Refining the partition leaves this sum unchanged. Every summand is a martingale, by conditioning its future increment on the information at . Bounded coefficients and the bound on make the integral an L2-bounded continuous martingale. Its terminal value is the same sum with substituted at any infinite endpoint; martingale convergence supplies this limit.
We need the conditional square identity, rather than assuming independent increments. The Doob L2 maximal inequality gives . Localize the square identity for at level and bracket stopping times. The stopped identity and this maximal bound give by monotone convergence. Nowis integrable, so localization passes to conditional expectations and is a uniformly integrable martingale. Expanding a squared increment and using provesThe infinite endpoint follows from convergence and integrability of the bracket. The initial value need not be zero: the integral itself is zero-starting.
In the square of the terminal integral, all cross terms vanish. For , the th increment and both coefficients in its cross term are measurable at , whereas the conditional mean of the th increment there is zero. For the diagonal terms, the preceding conditional identity can be multiplied by the bounded . ThereforeThis proves the Itô isometry for these simple integrands. Ordered stopping-time intervals also work, replacing the conditional identity by optional stopping theorem, but deterministic intervals suffice for the approximation requested here.
For that approximation, put on the predictable sigma-algebra. This is a finite measure. That sigma-algebra is generated, apart from the zero-time slice of measure zero, by rectangles with . The indicator of every such rectangle is a bounded simple predictable process.
Here is a density proof. Let be the closed linear span of these indicators in , and let contain the predictable sets whose indicators belong to . The whole space belongs to ; complements stay in ; and disjoint countable unions do also, because their finite partial indicator sums converge in under the finite measure. Thus is a Dynkin system containing the generating rectangle pi-system. The Pi-lambda theorem makes it the entire predictable sigma-algebra. Bounded predictable simple functions therefore belong to . Truncating any function and then approximating its truncated range by finitely many levels shows that every such function belongs to . A finite sum of rectangle indicators can be rewritten on one common deterministic time partition, with the coefficient on each interval measurable at its left endpoint; it is exactly an allowed simple process.
Consequently choose of this form with . This provesIt is density of simple predictable processes for finite measures, with a proof of the actual approximation step.
Finally apply the preceding isometry to , whose bracket is . The elementary integrals satisfyDefine the Itô integral as the limit of these elementary integrals. The same identity shows independence of the approximating sequence, linearity, mean zero, and . Applying the Doob inequality to the difference processes also gives uniform-in-time convergence on , with a continuous martingale version of the integral.
Past exam of the mathematics course of the University of Cambridge 2003 iii Paper 29 1 c Solution Created 2026-10-03 Updated 2026-10-14
Again put . For any fixed real , the random variableis nonnegative and integrable. The sign function of is bounded and -measurable, so conditioning and equality of the probability distributions giveThus almost surely. On this forces ; on it forces ; on it forces . In particular up to a null set. These two events have equal probability, again by equality of the probability distributions, so they agree up to a null set. ConsequentlyTaking establishes the suggested sign assertion, but this alone would not exclude unequal positive values. Apply the same argument simultaneously to the countable set of rational . Outside the union of their null sets, unequal and would have a rational strictly between them, giving opposite signs. This proves the conditional expectation preserving a distribution result:Only integrability was used; no finite second moment or hidden square-integrability assumption is needed.
Past exam of the mathematics course of the University of Cambridge 2003 iii Paper 29 6 a Solution Created 2026-10-03 Updated 2026-10-14
Interpret the comparison at on ; values at an infinite stopping time are not required. The stopping time events make adapted, and gives integrability. On the next value is . On it is unless , and on that latter event switching only decreases the value. Consequently the pointwise inequalityholds even though need not be known at time . Both indicators displayed here are -measurable. Taking conditional expectations and using the two supermartingale inequalities therefore givesThus This proves pasting supermartingales with downward jumps without incorrectly treating the future switching event as predictable.
Past exam of the mathematics course of the University of Cambridge 2003 iii Paper 29 6 b Solution Created 2026-10-03 Updated 2026-10-14
Put . Each successive crossing time , , is a stopping time, since in discrete time the event that a threshold is first met after the previous stopping time can be written as a finite union of events known at the candidate time. Use , with all later crossing times then infinite as well.
For the supermartingale assertion, build a process by finitely many switches. Begin with the constant supermartingale one. At switch to ; the new value at that time is at most one. At switch to the constant ; the old value there is . At switch to , whose value is at most , and at switch to the constant , whose value is at most the previous process. More generally, the switches at and are respectively from cash to and from that stock value to cash . Every switch is downward, including possible overshoots of either threshold.
Each candidate continuation is a supermartingale, being either constant or a fixed positive multiple of . Part (a) therefore proves by induction that the process obtained after the first switches is a supermartingale. It agrees with through . For fixed , its values are bounded in absolute value by a deterministic constant depending on times , which also verifies the required integrability rather than assuming it for an infinite sequence of pastings.
Stopping a discrete supermartingale preserves the property. Indeed its stopped increment is , with an -measurable indicator; integrability at any fixed time follows from a finite sum of the absolute values at earlier times. HenceThis is the multiplicative upcrossing supermartingale construction. The printed case uses , while was only defined from time zero. With the natural pre-start convention , it is the constant process one and the assertion also holds for . Without such an extension, that single displayed stopped process is undefined.
For the probability bound, the actual initial value isbecause exactly when . For an integer , stop after the th completed upcrossing, at . The nonnegative stopped supermartingale satisfiessince on the event in the left side. Letting , and identifying , proves the Dubins upcrossing inequality:Neither finiteness of nor uniform integrability is needed. The bound is for positive integers ; extending it to would generally be false, because its left side would be one while can be less than one.
Past exam of the mathematics course of the University of Cambridge 2003 iii Paper 32 1 b Solution Created 2026-10-03 Updated 2026-10-14
We prove exhaustiveness, including existence of the positive density when neither form of arbitrage occurs. Use the subspace from part (a)(ii), and considerThis is a nonempty convex cone closed under positive scaling: supplies one element and guarantees weighted integrability even when itself has no first moment. Crucially, is relatively open in ; it need not be closed.
To prove openness at a point , put . The vectors , for bounded real measurable , span . Indeed, a vector orthogonal to all of them would satisfy for every such , forcing almost surely. Since , this says . Choose finitely many bounded for which form a basis of . For coefficients sufficiently small,This gives a relative neighbourhood of . It also proves the openness step behind the strictly positive barycentre cone lemma without assuming a closed moment cone.
If , the second arbitrage form was already constructed. If , the separation of a point and an open convex set, applied inside the finite-dimensional space , gives a nonzero with . Positive scaling in the cone forcesWe claim almost surely. Otherwise has positive probability, and the admissible positive weights givecontradicting the separation inequality. Since is nonzero, cannot vanish almost surely; thus it is strictly positive on an event of positive probability. This is the first arbitrage form.
Consequently absence of both forms forces , which supplies the required . Conversely either form excludes every such density by part (a). We have proved exactly one of arbitrage or a strictly positive pricing density holds. If , the same argument reduces to the elementary cases and , so no nondegeneracy assumption has been hidden.
Past exam of the mathematics course of the University of Cambridge 2003 iii Paper 32 1 Solution 2026-10-14
The positive state-price density alternative characterizes one-period arbitrage. Interpret as the vector of current asset prices and as their terminal payments. A fixed vector of holdings produces initial cost and terminal payment . The first alternative is a nonpositive-cost investment with a nonnegative payment and strict improvement either initially or on an event of positive probability. A strictly positive state-price density prices every such payment and excludes both forms of arbitrage, as proved below.
If there is a traded bank account costing one and paying a deterministic , the density satisfies . Hence defines an equivalent martingale measure andConversely this pricing relation under an equivalent martingale measure gives . Without such an account, is a pricing density, not necessarily a probability density or even an integrable random variable by itself; the stated requirement is integrability of . This distinction keeps the abstract alternative valid without an unstated asset paying one.
Past exam of the mathematics course of the University of Cambridge 2003 iii Paper 32 2 Solution Created 2026-10-03 Updated 2026-10-14
Specify the trading conventions first. Let the bank account have value , with and its next gross return known at time . Write for the discount factor. Let be the ex-dividend asset-price vector, and the vector of cash dividends paid per unit at time ; set for non-dividend-paying assets. Under an equivalent martingale measure , each discounted cumulative asset gainis a martingale. Equivalently, . This is the gain-process form of risk-neutral pricing, rather than a martingale assumption on raw ex-dividend prices.
Let be the -measurable holdings after trading and distributions at time . The previous holdings earn the next price change and dividends. Define the strategy's net cash payment and its residual portfolio wealth byA negative payment is an external infusion; a self-financing portfolio with no consumption has . The discounted budget equation isAssume these gains and payments are integrable under , for example by bounded predictable holdings and integrable discounted asset gains. Thenis a true martingale, by conditioning the displayed one-step gains. Merely assuming a local martingale without appropriate integrability would not justify the conditional equality.
Conditioning on proves the general multi-period dividend pricing identityFor a strategy liquidated at time , and all liquidation proceeds are included in . The terminal value disappears, giving exactly the requested dividend-only relation. That liquidation convention is necessary: holding one unit of the bank account with no payments would otherwise have a positive left side and zero right side.
An attainable claim paid at can be replicated by a strategy with no earlier payments and final liquidation payment . Its time- value is thereforeTwo admissible replicating strategies have the same value by this identity, and the value is the same under every equivalent martingale measure for which the replication gains are true martingales. For unattainable claims this expectation is a possible model price, but the replication argument does not establish a unique price. With density process , the equivalent state-price density form is .
Past exam of the mathematics course of the University of Cambridge 2003 iii Paper 32 5 Solution Created 2026-10-03 Updated 2026-10-14
Use the physical dynamics of the Black-Scholes model, , with bank rate and . Write for the market price of risk. If is portfolio wealth, the dollar amount in stock, and the consumption rate, the budget dynamics areAdmissible controls are progressively measurable, locally integrable in the required drift and diffusion norms, and keep . Such a solvency restriction excludes doubling-type strategies. In a complete Brownian market, the investment-consumption problem is to maximizewhere the utility functions are increasing and concave. A terminal-only problem omits the integral. Strict concavity gives uniqueness of optimal consumption and terminal wealth when an optimum exists.
The state-price density isThe Itô formula givesThus is a nonnegative local martingale, hence a supermartingale, yielding the state-price budget constraintConversely, a nonnegative consumption plan and terminal payoff satisfying budget equality and the necessary integrability can be financed in the Brownian complete market. DefineThe Brownian martingale representation theorem for the total future budget supplies a stochastic-integral coefficient; comparing it with defines . This reconstructs the wealth equation with . Localization gives the representation for integrable positive budgets when an assumption is not available.
For smooth strictly concave utilities with the usual marginal-utility range, let and be the inverse marginal utility functions. The Lagrange-multiplier solution iswhere makes the budget an equality. This is a global argument, not merely formal differentiation: the concave supporting-tangent inequality giveswith the analogous terminal inequality. Taking expectations, integrating, and applying the budget bound makes the difference in objectives nonpositive. Replication of the candidate budget proves attainability and optimality whenever these expectations are well-defined and finite. If the marginal-utility ranges or integrability conditions fail, existence must be established separately rather than inferred from the first-order equations.
A detailed example is constant relative risk aversion utility and , with , , , and . PutThe optimal plans from the marginal-utility equations are and . The normal distribution of gives, for ,The budget therefore becomes , determining the multiplier explicitly. Independent Brownian increments in the conditional budget formula give . The diffusion coefficient of is , so matching the wealth diffusion determines the dollar stock holding. Consequently the finite-horizon power-utility investment and consumption optimum isAll candidate moments are finite on a finite horizon, since powers of a lognormal state-price density have finite moments. The budget and tangent argument above therefore verify optimality in this example, including admissibility and positive wealth.
For another verification of the formulas, the Hamilton-Jacobi-Bellman equation for a value function with discount measured from the current time isSubstitute . The two concave maximizations give the displayed controls, and the remaining equation is with , exactly solved by the formula above. ThusFor , interpret the integral as , so . Unlike the infinite-horizon Merton consumption-investment problem, finite horizons do not require . Consumption vanishes in the terminal-only formulation; then the integral term in is absent and the optimal stock fraction remains . Logarithmic utility is handled directly with inverse marginal utility and the same wealth-to-consumption formula at , with .
Past exam of the mathematics course of the University of Cambridge 2004 iii Paper 40 6 iii Solution Created 2026-10-03 Updated 2026-10-14
For observed data and missing data , let be the complete-data model. Starting at , the expectation-maximization algorithm alternatesThe E-step computes the conditional expectation of the complete-data log-likelihood, using the current parameter for the missing-data conditional distribution. The M-step maximizes that expected log-likelihood over a new parameter. Maximizing the likelihood of imputed mean data is equivalent only when the complete-data log-likelihood has the necessary linear dependence on the missing sufficient statistics.
For the monotonicity argument, let . Jensen inequality givesTherefore an M-step increasing cannot decrease the observed-data likelihood. Standard support and integrability conditions are understood. EM need not find a global maximum for a general model; its behavior depends on the objective and starting point.