Analysis & Remarks

The Reference Price Problem

Why Valuation Frameworks Require a Prior Test of Auction Validity

Dejan Shabacker — Marketbet AB Analysis & Remarks, No. 1 — 1 September 2026

Abstract

Every valuation framework produces an estimate of value and compares it to a market price. The comparison is the entire operational content of the exercise: the estimate acquires meaning only in relation to the price it is set against. This paper argues that the comparison rests on an assumption that is almost never stated and never tested, namely that the observed market price is a valid object of comparison. The assumption fails whenever the auction that produced the price was structurally impaired. Under such conditions a valuation gap is not evidence of mispricing; it is evidence of nothing at all, because one of the two terms in the comparison carries no informational content. The paper sets out the structure of the problem, distinguishes two classes of valuation error that are routinely conflated, and specifies what a prior test of auction validity must deliver in order to resolve the ambiguity. It concludes with the testable predictions that follow, and with the limits of the argument.

1. The unstated premise

Valuation frameworks differ in method but share a common form. A procedure generates a value estimate, and that estimate is placed alongside the prevailing market price. Discounted cash flow analysis, multiples-based comparison, residual income models and consensus aggregation all terminate in the same operation. The output of the procedure is not the estimate itself but the difference between the estimate and the price.

This form carries a premise. For the difference to be interpretable, the market price must be the product of a process capable of representing collective assessment. If the price was set through balanced negotiation between participants who were both present and both willing to transact, the difference between estimate and price is informative: it locates a disagreement between the analyst’s method and the market’s aggregate judgment, and the analyst may reasonably take a position on which is correct.

If the price was not set through such a process, the difference is uninterpretable. It may reflect a genuine disagreement, or it may reflect nothing more than the temporary absence of one side of the market. The two cases are indistinguishable from the valuation output alone, because the valuation procedure never examined the price-forming process. It took the price as given.

The premise is not a technical detail. It is the condition under which the entire exercise means anything.

2. Two classes of error

The literature on valuation error treats the estimate as the locus of failure. Estimates are wrong because inputs are wrong, because the model is misspecified, because analysts are subject to bias, or because the future was not knowable. This is the first class of error, and it is well described.

A second class exists and is rarely separated from the first. Here the estimate may be entirely sound, and the failure lies on the other side of the comparison. The price against which the estimate was tested was generated under conditions in which no genuine negotiation occurred. There was no test of intermediate levels, no assessment by both parties, no convergence toward a level that either side would defend. The number exists and is recorded, but it did not emerge from the process that gives market prices their epistemic standing.

Conflating the two classes has a specific and costly consequence. When an estimate is compared against an invalid price, three outcomes are possible: the position established on the basis of the gap performs, it does not perform, or it performs after a delay long enough to have forced its own liquidation. All three outcomes are attributed to the quality of the estimate, because the estimate is the only object under examination. The framework is revised, the inputs are adjusted, and the actual source of the failure is left untouched — it was never represented in the analysis.

Separating the two classes requires a diagnostic that operates on the price rather than on the estimate. That diagnostic must be prior to the valuation comparison, not subsequent to it, because its function is to determine whether the comparison should be performed at all.

3. What the diagnostic must deliver

A test of auction validity is subject to four requirements.

It must be independent of the valuation. If the test uses the value estimate as an input, it cannot arbitrate between the estimate and the price; it inherits whatever error the estimate contains. The test must operate on the observable properties of the price-forming process alone.

It must be directional in time. A test that identifies impairment only after the fact has diagnostic but not operational value. The requirement is identification of the transition — the point at which a structurally impaired auction resumes functioning — because that transition marks the moment at which the price regains its status as a comparison object.

It must be continuous rather than binary. Auction integrity is not a state that is either present or absent. It admits of degree, and the degree is what determines how much weight a given price observation can bear. A binary classifier discards the information that matters most, namely the magnitude of the impairment.

It must be falsifiable. The test must generate predictions that can fail. A framework that explains all outcomes after the fact has no content. This requirement is addressed in Section 5.

The Market Auction Integrity framework is constructed to satisfy these four requirements. Its central claim is that the structural soundness of the auction is separately observable from the price it produces, and that the observation can be made in a way that does not presuppose any view of value.

4. Consequences for two applications

4.1 Intrinsic value estimation

Origo is defined as an integrity-weighted intrinsic value: an estimate anchored not only in the fundamental and forensic assessment of the company but in a determination of whether the market price against which that assessment is set is fit to serve as a reference. The weighting is not a confidence interval around the estimate. It is a separate statement about the other term in the comparison.

The practical distinction is as follows. A conventional framework that produces an estimate diverging materially from the market price reports a gap and leaves its interpretation to the user. An integrity-weighted framework reports the gap together with a statement of whether the price is currently a valid reference. Where it is not, the correct inference is that the comparison is suspended, not that the gap is large.

This resolves a specific and recurrent failure. Valuation gaps are systematically widest precisely when auction integrity is lowest, because the same conditions that impair negotiation also displace price furthest from any level rational participants would defend. A framework blind to auction validity therefore generates its strongest apparent signals at the exact moments its comparison is least reliable. The correlation runs in the wrong direction, and it is structural rather than incidental.

4.2 Hedging and exposure

Delta measures the sensitivity of an instrument’s value to changes in the price of the underlying. The measure is defined within a functional market. It presumes that a change in the underlying price constitutes information about the underlying, since that is what a price change means when the auction is intact.

Where the auction is impaired, the presumption does not hold. Price movement under impairment is not information about the underlying; it is the observable trace of one side’s absence. Hedging against such movement adjusts exposure in response to a signal that carries no content about the object being hedged. The cost of the adjustment is incurred, the position is altered, and the movement subsequently reverses when the withdrawn side returns.

An exposure framework governed by auction integrity rather than by price motion inverts this behaviour in a defined way. It reduces adjustment where movement is structurally invalid and likely to revert, and increases it where movement reflects genuine negotiation. The distinction is not a refinement of delta. It is a statement about the domain within which delta is defined, and about what should be done outside that domain.

5. Testable predictions

The argument generates predictions that can fail. Four are stated here in the form required for falsification.

P1. Valuation gaps measured against prices classified as low-integrity should exhibit lower forecasting power for subsequent returns than gaps of equal magnitude measured against prices classified as high-integrity. If the two classes show equivalent forecasting power, the distinction carries no information.

P2. The dispersion of realised outcomes following the identification of a transition from impairment to function should be narrower than the dispersion following an arbitrary observation of equal price displacement. If the dispersions are equivalent, the transition point has no operational content.

P3. Exposure adjustments suppressed on the basis of low measured integrity should, in aggregate, avoid a cost that exceeds the opportunity cost of the suppression. If suppression is on average costly, the criterion does not identify what it claims to identify.

P4. Integrity measurements should be substantially independent of realised volatility. If the measurement correlates strongly with volatility, it is a volatility estimator under a different name and adds nothing to the existing set of measures.

Each prediction can be tested on historical data without reference to any valuation framework. Each can fail.

6. Limits of the argument

Three limits are stated explicitly.

The argument establishes that auction validity is a necessary prior condition for interpreting a valuation gap. It does not establish that any particular method of measuring validity is correct. The four requirements in Section 3 constrain the class of admissible methods; they do not identify a unique member of it.

The argument applies to continuous double-auction markets. Instruments that trade by negotiation, by auction at fixed intervals, or through a single dealer are governed by different price-formation mechanics, and the framework as stated does not extend to them without modification.

The argument does not address the case in which impairment persists indefinitely. Where the withdrawn side does not return — because the information asymmetry that caused the withdrawal is permanent, or because the instrument has ceased to attract two-sided interest — there is no transition to identify and no reversion to anticipate. Distinguishing temporary from terminal impairment is a separate problem and is not solved here.

7. Conclusion

Valuation frameworks compare an estimate against a price and treat the difference as their output. The comparison presupposes that the price is a valid object of comparison, and that presupposition is neither stated nor tested in standard practice. Where it fails, the output of the framework is not a weak signal but an empty one, and no refinement of the estimate can recover it.

The correction is not a better estimate. It is a prior determination of whether the comparison should be made. That determination requires a measurement of auction integrity that is independent of the valuation, directional in time, continuous, and falsifiable. Where such a measurement is available, valuation gaps and price movements can be separated into those that carry information and those that do not — a separation that the estimate alone cannot supply, because the estimate was never the term in question.

References

Black, F. and Scholes, M. (1973). The pricing of options and corporate liabilities. Journal of Political Economy, 81(3), 637–654.

Glosten, L. R. and Milgrom, P. R. (1985). Bid, ask and transaction prices in a specialist market with heterogeneously informed traders. Journal of Financial Economics, 14(1), 71–100.

Grossman, S. J. and Stiglitz, J. E. (1980). On the impossibility of informationally efficient markets. American Economic Review, 70(3), 393–408.

Kyle, A. S. (1985). Continuous auctions and insider trading. Econometrica, 53(6), 1315–1335.

Roll, R. (1984). A simple implicit measure of the effective bid-ask spread in an efficient market. Journal of Finance, 39(4), 1127–1139.

Soros, G. (1987). The Alchemy of Finance. New York: Simon & Schuster.

This document is analytical and methodological in nature. It does not constitute investment advice, a recommendation, or an offer to buy or sell any financial instrument. Marketbet AB trades its own capital and may hold positions in the securities and instruments discussed.