Market Data for Pricing and Risk

A financial position is not valued from one latest price. Its value depends on a market snapshot at a chosen time: prices, interest rates, dividends, volatility, credit assumptions, calendars, and contract conventions must describe the same market state.

The useful lifecycle is:

position terms → market-data snapshot → valuation → P&L and risk → validation or exception

A listed call option shows why this matters. The position has an underlying share, a strike, an expiry, an exercise style, and a contract multiplier. To value it, a system also needs the share price, interest-rate data, expected dividends, and volatility for the relevant expiry and strike. If one of those inputs is missing or belongs to a different time, the result may look precise while describing no consistent market state.

Start with the position and the as-of time

The position record describes what was traded. It may contain the option series, quantity, side, premium, and settlement rules. Market data describes the conditions under which that position is valued.

A valuation run therefore needs an as-of time, the boundary that says which observations belong to the run. A risk report for 16:00 should not silently combine a 16:00 share price, a 15:00 volatility surface, and a curve built from 17:00 quotes. Such a result can contain a timing difference that looks like P&L or risk.

The same rule applies to historical analysis. To reproduce yesterday’s valuation, the system needs yesterday’s position state and the market snapshot used at that time. Replacing one old observation with today’s value may make the calculation easier, but it no longer explains the original result.

The data record should carry more than a number:

Field	Why it matters
Value and quote type	A bid, ask, mid, fixing, settlement price, and model estimate have different meanings
As-of timestamp	Shows which market state the observation describes
Instrument or curve identifier	Prevents a value for one maturity or underlying being used for another
Currency, unit, and quote direction	Distinguishes dollars per euro from euros per dollar, or a price per share from a contract price
Source and quality state	Makes substitutions, stale values, and fallbacks visible
Effective date and convention	Applies fixings, curves, and corporate actions at the correct time

One instrument needs a set of inputs

Different instruments depend on different market inputs. A price feed cannot be designed as one universal table with only symbol and price.

Position	Main market inputs	Typical valuation consequence
Share	Spot price, dividends, corporate-action-adjusted terms	Equity value and dividend-adjusted forward value
Bond	Cash-flow schedule, discount curve, credit spread, price convention	Present value and interest-rate or credit sensitivity
Currency forward	FX spot quote, domestic and foreign curves, settlement rules	Delivery price or value of the existing contract
Interest-rate swap	Curves, fixings, payment schedule, day-count and collateral convention	Discounted value of the two payment legs
Commodity contract	Spot or futures observations, storage, delivery location and grade	Carry value and basis risk (the hedge price and the physical price moving differently)
Option	Underlying, strike, expiry, rates, dividends, volatility and settlement rules	Premium, exercise value and nonlinear risk

The instrument determines the required data. A missing dividend estimate matters to an equity option, but a fixed-rate bond has no dividend input. A missing fixing can leave a swap payment undetermined, while an outdated delivery location can change the meaning of a commodity hedge.

Curves turn cash flows into present value

A future cash flow is not worth its face amount today. A discount factor The discounted cash flow (DCF) analysis, in financial analysis, is a method used to value a security, project, company, or asset, that incorporates the time value of money. Discounted cash flow analysis is widely used in investment finance, real estate development, corporate financial management, actuarial science, and patent valuation. Terminal value often represents a large share of total value and is highly sensitive to growth and discount rate assumptions. Enterprise DCF commonly uses free cash flow to the firm and a continuing value beyond the explicit forecast horizon. Used in industry as early as the 1800s, it was widely discussed in financial economics in the 1960s, and U.S. courts began employing the concept in the 1980s and 1990s.Read on Wikipedia → maps a payment at a future date to its value today. If a position has cash flows at several dates, its present value can be written as:

$$PV = \sum_{i=1}^{n} CF_i D(0,t_i)$$

Here, $PV$ is the present value, $n$ is the number of cash flows, $CF_i$ is the cash flow at date $t_i$, and $D(0,t_i)$ is the discount factor from that date to today. For example, a discount factor of 0.95 turns a payment of 100 into a present value of 95. The inputs also carry important conventions. The curve must match the currency, compounding method, day-count rule (how the time between two dates becomes a fraction of a year), payment calendar, and collateral or funding assumption (the rate assumed for posted collateral or borrowed money) used by the product.

A yield curveA yield curve shows the relationship between yields and time to maturity for a set of comparable debt securities. In practice the term usually refers to curves built from a single issuer or market segment so that credit quality and other features are as similar as possible, for example the U.S. Treasury curve for government bonds.Read on Wikipedia → supplies rates across maturities rather than one rate for every payment. A discount curve supplies discount factors. A forward curve supplies rates or prices for future periods. These objects are related, but they are not interchangeable labels.

For example, an interest-rate swap has a payment schedule. The fixed leg uses its contractual rate and accrual fractions. The floating leg uses fixings, the published reference-rate values that set each period’s rate, or estimates of fixings that are not yet known. The valuation then discounts the resulting cash flows using the applicable discounting convention. The swaps article describes these payment legs in more detail. A wrong maturity point, day-count rule, or curve can change the result even when the notional and headline rate are correct.

The system may build curves from several market instruments. Market quotes become a curve; the curve becomes discount factors or forward rates; those values enter cash-flow valuation.

Options need volatility and contract state

An option quote belongs to a particular series. The system must know the underlying, call or put, strike, expiry, exercise style, multiplier, settlement method, and any adjusted deliverable. The market data must then match that series.

The main inputs for an equity option are:

• the underlying spot price;
• the discount or funding curve;
• expected dividends or a dividend yield;
• the time remaining to expiry;
• volatility for the relevant expiry and strike or moneyness, the relationship between the strike and the current underlying price;
• bid, ask, or another chosen market-price convention.

Volatility is usually not one universal number. A volatility surfaceVolatility smiles are implied volatility patterns that arise in pricing financial options. It is a parameter that needs to be modified for the Black–Scholes formula to fit market prices. Generally, for a given expiration, options whose strike price differs substantially from the underlying asset's forward price tend to have prices that deviate from their expected prices using a constant-volatility model based on the at-the-money.Read on Wikipedia → maps volatility across expiry and strike or moneyness. It is commonly inferred from option prices under a model and interpolation convention. Using one volatility for every option can hide differences between short and long expiries or between strikes.

The surface also needs a timestamp and quality state. If the underlying price is current but the volatility surface is stale, a new option value may reflect the data mismatch rather than a real market move. A surface with gaps or extreme jumps may need an exception or a controlled interpolation, not silent acceptance.

Bid and ask prices also carry information. A mid-price, the average of bid and ask, can be useful for a valuation report, while a liquidation estimate may use the side of the market at which the position would trade. The choice should be explicit. A wide spread can make a small mark-to-market change less meaningful than the cost of exiting the position.

Fixings, credit, and corporate actions change the data state

Some market inputs are published observations rather than continuous prices. An overnight or term reference rate becomes a fixing used by a contract’s rules. The system must know the publication time, the fixing date, the relevant calendar, and whether the fixing is already known for the valuation run.

A swap can use observed fixings for a period that has started and estimated forward rates for periods that have not. Mixing those states can produce a cash-flow error. A missing fixing should be visible and handled according to a documented fallback rule; it should not be replaced by a plausible number without a quality flag.

Credit instruments need their own data. A credit default swap may use a spread, a reference entity, recovery assumptions, a credit-event status, and a settlement method. A bond may use a credit spread or a default model in addition to the risk-free discount curve. The result is sensitive not only to rates but also to the borrower’s credit state and the protection or collateral terms.

A corporate actionA corporate action is an event initiated by a public company that brings or could bring an actual change to the debt securities—equity or debt—issued by the company. Corporate actions are typically agreed upon by a company's board of directors and authorized by the shareholders. For some events, shareholders or bondholders are permitted to vote on the event. Examples of corporate actions include stock splits, dividends, mergers and acquisitions, rights issues, and spin-offs.Read on Wikipedia → is another time-dependent data event. A stock split can change an option’s strike and deliverable. A dividend can affect an equity forward or option. A merger or rights issue can change the underlying or settlement terms. The adjustment must be applied from its effective date to the correct position state.

A static security master is not enough for these cases. The system needs an event record, an effective date, the affected terms, and a rule for positions opened before or after the event. The options article gives an example of an adjusted strike and quantity.

Data quality is part of valuation

A market-data service should reject or flag values that are structurally inconsistent before they reach valuation. Useful checks include:

Check	Failure it can expose
Bid is not above ask	Reversed fields, bad source data, or crossed-market handling that needs review
Units and quote direction match the instrument	A price per share used as a contract price, or EUR/USD inverted as USD/EUR
Timestamp is inside the run’s as-of boundary	Future data leaking into a historical or intraday result
Curve maturities and conventions are valid	Missing dates, duplicate maturity points, or an incorrect day-count convention
Fixing is available at the expected publication time	A future fixing used too early or a missing fixing silently filled
Observation is not stale without a flag	A frozen price or volatility surface treated as current
Corporate action is applied once, from its effective date	A split or dividend applied twice or to the wrong position version

A fallback source can preserve service availability, but it should not hide data quality. The resulting value should carry a flag such as stale, substituted, estimated, or missing. A risk report can then distinguish a market move from a data problem.

Quality checks should also compare related values. A large jump in a curve, a dividend, or a volatility surface may be real, but it deserves a review when it has no matching market event. A check does not need to reject every unusual value; it needs to make the exception visible to the process that owns the data. Curve maturity points should also be unique; an unexplained duplicate maturity point can distort interpolation.

One snapshot should support valuation, P&L, and risk

Valuation uses the snapshot to calculate the current value. P&L compares that value with a compatible previous snapshot and separates market movement, new trades, cash flows, and data changes where possible. Risk applies shocks to the same clean base state.

If valuation uses today’s spot, yesterday’s volatility surface, and a curve from a different cut-off time, the reported P&L may contain a data-timing difference. If a risk run uses a different snapshot from the valuation run, the portfolio may appear to have changed before any scenario is applied.

Portfolio risk then adds relationships between positions and risk factors, the market variables that can change a position’s value. Factor exposures describe which common drivers affect the portfolio. Correlations describe how drivers may move together. These are important inputs to portfolio risk, but they do not replace instrument-level prices, curves, fixings, and contract state.

To keep a result explainable, store the market snapshot so it can be reproduced, keep conventions with the values, and make exceptions visible to the process that owns the data.