Price Targets and Ratings

The client price target begins with a transparent blend of peer evidence and published consensus. Each component must then survive an independent evidence check before it can determine the displayed target or rating.

​

The Economic Question

What price best summarizes comparable-company evidence and published analyst expectations, when the two sources can be informative, correlated, and occasionally wrong in different ways? The answer has two stages. First, calculate each source on a common per-share and currency basis. Second, decide whether both sources have earned the right to determine the client target. The rating is a separate signal-to-noise judgment. A 12.0% valuation gap is consequential for a stable stock and ordinary for a highly volatile one.

​

First-Principles Derivation

Peer center = a robust, quartile-weighted center of the eligible peer-implied values Sell-side center = the average of the published consensus mean and median Raw target = 50.0% × peer center + 50.0% × sell-side center Expected return = client target / current price − 1 One term, two senses, so the pages keep them apart. Here it means the return implied by the published target. On the reverse-DCF and cost-of-capital pages it means the return the market appears to require at the current price. The first is an output of this calculation. The second is a diagnostic about the price. The raw target is always retained as the audit calculation. Peer multiples and sell-side targets are correlated sources, because published targets are themselves often built from peer multiples. Averaging them does not establish independent corroboration or guarantee lower error. Any error reduction depends on the sources’ biases, error variances and error correlation. Equal weighting keeps both contributions explicit and gives the reader a result they can reconstruct in one line. The blend summarizes peer and sell-side market evidence. The DCF provides a separate intrinsic valuation based on operating assumptions and discounted cash flows. Keeping the two measures distinct lets readers compare market expectations with the fundamental case.

​

The Two Component Checks

Failure of one component does not contaminate the other. It changes which evidence can determine the client target, while the failed component and its reason remain in the audit trail.

​

The Reconciliation Decision

Two diagnostic tests identify cases that deserve attribution rather than blind averaging. The first asks whether the raw target and each component sit within a defined band around the current price. The second asks whether one component center is a large multiple of the other. An evidence reviewer may approve the raw blend, select the existing peer center, select the existing sell-side center, or reject publication. The reviewer cannot enter a bespoke target. Review determines which existing evidence supports publication.

​

How the Diagnostic Band Is Used

The band does not alter any valuation. It asks a narrower question: when an answer is far from price, is that distance supported by two independently qualified components, or is one component carrying the entire conclusion?

​

The Volatility-Aware Rating Gradient

A fixed percentage hurdle treats a utility and a pre-revenue biotech as though a given valuation gap means the same thing in both. It does not. Parallax scales the rating hurdle to the stock’s own realized volatility, measured over a defined trailing window. The ordinary hurdle is a fraction of realized volatility and the strong hurdle is a larger multiple of it. Both are bounded by a floor and a cap.

• The floors stop quiet stocks from receiving a Buy or Sell label for a trivial valuation gap.
• The caps stop volatile stocks from requiring an unreachable target before the system can express a view.

Two limitations come with this and are worth stating plainly. Horizons have to match. An expected return is a return over some period, and a volatility measure is a dispersion over some window. Comparing the two only means something when both are expressed on the same horizon, so the workbook footnote prints the window the hurdle was scaled from. A trailing measure is procyclical. Realized volatility clusters. A volatility spike widens the hurdle, which can soften a Sell toward Hold at the moment downside risk is highest, and a long quiet stretch tightens the hurdle in exactly the conditions that precede a repricing. The floors and caps bound that effect. They do not remove it. A reader comparing two ratings across different volatility regimes should check the printed hurdle rather than assume the labels mean the same thing. A target based on one qualified component can be Buy, Hold or Sell. It can never be Strong Buy or Strong Sell. Strong ratings require two-source corroboration in the same direction. When realized volatility is missing, invalid, or supported by too few observations, Parallax falls back to a fixed hurdle convention. The workbook footnote prints the actual hurdles applied to that company, and states whether they were scaled from realized volatility or drawn from the fallback convention.

​

A Worked Target and Rating

An illustrative security trades at 30.0. The peer center is 40.0. The consensus mean is 33.0 and the median is 35.0, producing a sell-side center of 34.0. Both components pass their checks, the centers are close, and the raw target is in-band. Client target = 0.5 × 40.0 + 0.5 × 34.0 = 37.0 The target is 23.3% above the current price. At a moderate level of realized volatility, that gap clears the ordinary hurdle and the rating is a Buy. At a very high level of realized volatility, the same gap sits inside the wider hurdle and the rating is a Hold. The valuation did not change. What changed is the strength of the signal relative to the stock’s normal price variation.

​

An Extreme-Result Repair

Consider a security where the peer center lands at many multiples of the current price while the sell-side center sits close to it. The raw equal-weight blend inherits the extreme, and the two centers conflict by a wide margin. Nothing is moved to the nearest boundary. The extreme peer result and the raw blend both remain available for challenge. The client target uses the in-band component, and the rating is recalculated from it. Because a single component determines the target, a Strong rating is unavailable.

​

Related Evidence

DCF, APV, factor scores, technical indicators and the 52-week range do not enter this price target. DCF remains valuable research. Factor scores describe systematic return characteristics. Technicals inform timing. The trading range provides context. None of them enters the arithmetic. This exclusion is a commercial and methodological choice. The client target must be concise, current and reconstructable from evidence that fits on one page. More elaborate methods stay available for deeper analysis, without making the product target impossible to explain.

​

The One-Page Reconstruction Test

The workbook shows the current price and date; each peer, the selected multiple and the peer-implied share value; outlier treatment; the distributional statistics and the peer center; the consensus mean and median; the raw blend; each component’s evidence status; any exclusion; the client target; realized volatility; the rating hurdles applied; and the final label. Calculations retain full precision. Client figures display one digit after the decimal point.

​

Interpretation Note

​

How to Assess the Result

Recalculate the peer and sell-side centers separately. Confirm that the peer multiple and target denominator use the same basis, that issuer-wide shares and net debt support any enterprise-value translation, and that both sources refer to the same security, currency and date. Then ask whether one component alone causes an extreme. Finally, compare the rating hurdle with realized volatility, and confirm that any Strong label has two-source support.

​

When to Reassess

The peer cohort no longer represents the relevant economics; duplicate listings survive; the target denominator and peer multiple use inconsistent periods; shares or net debt do not reconcile; analyst summaries are stale or unavailable; too few usable issuer observations support the selected metric; price, peer evidence, analyst targets and volatility refer to incompatible dates; or the workbook cannot reproduce the engine result at declared precision.

​

How Parallax Applies This

Parallax publishes the component centers, the raw blend, the evidence checks, the selected client target, the volatility-scaled hurdles and the calculation trail. Raw values are never clipped or overwritten. An excluded component remains visible with its exclusion reason. Human review is reserved for cases where two qualified sources corroborate an extreme or remain irreconcilable, and the reviewer can choose only among existing evidence or reject publication. The investor can reconstruct the answer, challenge the evidence, and disagree with the conclusion.