Price, Value and MarkMedium
Three numbers that all look like the worth of a thing. One is a trade that happened, one is somebody's model, and one is what a statement had to print. Confusing them is how a portfolio can be wrong for years without anything looking wrong.
5 min read · 955 words
Three different objects
- A price is an event. Two parties traded, at a size, at a moment. It is the only one of the three that is a fact about the world rather than about a method, and it is historical the instant it exists.
- A value is an argument. Discounted cash flows, a comparable multiple, a replacement cost: a model plus inputs plus assumptions, producing a number that is a conclusion. Valuation is the machinery, and every version of it belongs to whoever chose the assumptions.
- A mark is an obligation. A statement has to print something, so a rule produces a number for every holding on a date, including the holdings nobody traded and the ones nobody would. That is not a weaker price; it is a different kind of thing.
- The three agree most of the time, in liquid markets, which is exactly why the distinction is not learned there. They diverge in the situations that matter, and the divergence is not an error.
Where a mark actually comes from
- An observed trade in the same instrument — the strongest case, and the rarest outside the most liquid markets. Even here, whose trade and at what size is a choice.
- A quote nobody traded on. A dealer's mid, or the average of several. Perfectly reasonable, and a quote is an offer to trade in a size that may be far smaller than the holding.
- An observed price for something similar, adjusted. Matrix pricing: this bond has no trade today, that comparable one does, and a spread relationship carries the number across. The adjustment is a model even when it is a single number.
- A model, on inputs nobody can observe. Which is where the disclosure hierarchy has its third level, and where the amount of a firm's value that outsiders cannot check is stated. How a position hits the books is what that hierarchy is for.
- The point of the ladder is not that lower rungs are dishonest. It is that "the value of the portfolio" is a single figure assembled from four different kinds of evidence, and the composition is disclosed precisely because it matters.
Why the three come apart
| Situation | What happens to the three |
|---|---|
| No trading for weeks | There is no price; the mark carries on being produced, and its stability is a property of the method rather than of the asset. |
| Selling in size | The price achieved is below the mark, because the mark was a quote for a smaller amount — what liquidity costs is that gap. |
| Stress | Prices move faster than models are re-estimated, so a mark can be stale in the exact week its accuracy matters most. |
| A private holding | There is no price at all, ever, until an exit. The reported figure is a valuation, and the smoothness of a series of valuations is not evidence of stability. |
| A forced seller on the other side | The price is a fact and is not an estimate of value — which is why "the market says it is worth this" is a claim about who had to trade. |
The four questions
- Did somebody trade at this number? If yes: in what size, and how long ago. If no, it is a mark or a valuation whatever it is labelled.
- Who produced it, and what do they gain if it is higher? Not an accusation — a structural question, and the reason a fund's valuation, its administration and its custody are meant to sit in three different firms. Madoff, 2008 is what collapsing them removes.
- What would move it? A mark computed from an unobservable input moves when somebody re-estimates the input, which is a decision with a date, not an event in a market.
- What size does it hold for? Almost every number of this kind is quoted for a size, and the size is almost never quoted with it.
The two failures this distinction prevents
- Reading a smooth series as a stable asset. An asset marked quarterly by a model looks calmer than a listed one that does the same thing, and the difference is the observation frequency rather than the risk. Any measure computed from that series — a volatility, a correlation, a ratio of return to risk — inherits the flattery, which is why reading a market number asks how the series was produced before asking what it says.
- Reading a price in a bad week as a value. The opposite error, and just as common: a forced sale sets a price and it is a real one, and it is evidence about the seller's constraints as much as about the asset. A mark that refuses to follow it is not necessarily wrong.
- Both errors have the same repair, and it is one sentence: say which of the three you have before drawing any conclusion from it.
Where the site uses each one
- The calculators produce values, from assumptions the reader sets, and each says so under itself. They are arguments with the inputs made visible, which is the only honest form a valuation takes.
- The case studies mostly turn on marks — what a holding was carried at, by which method, and what happened when the method met a market.
- And no page here publishes a price. A price is an observation about a moment, and a static site cannot hold one that stays true; that is the same reasoning that keeps live market sizes off every page.
Information and education only. This explains how three kinds of figure are produced, in general terms. It is not advice, it is not a valuation of anything, and no number produced by any method described here is one to rely on.
More in Analysis
- MediumIf the Other Side FailsEvery instrument depends on somebody continuing to exist
- MediumThe Other Side of the TradeEvery position has a counterparty with a reason
- EasyWho Gets Paid Along the WayBetween your money and the asset sit six or seven parties, each taking something
- MediumWhat Happens When Everybody Does ItA position that works partly because few people hold it stops working when many do — and the mechanism that ends it…