MEASURED FEELING

I. The problem: a variable everyone prices and no one owns

Every practitioner believes markets have moods; the discipline spent half a century deciding whether that belief was respectable. The difficulty was never whether investors feel — it was whether feeling is measurable before the fact and priced after it. A variable that cannot be observed independently of the prices it is invoked to explain is not an explanation; it is a name for the residual. The history of sentiment research is the history of escaping that circularity: finding observables — survey answers, fund discounts, option premia, word frequencies, search queries, now on-chain acts — that proxy the collective emotional state without being mere restatements of the price.

The stakes of the escape are doctrinal. If sentiment is measurable and predictive, then the strict form of market efficiency fails not at the margins but at the mechanism: prices would embed a component that is neither fundamental news nor rational risk premium, and that component would be forecastable from psychological data. This is precisely what the modern measurement literature claims to have found (Baker & Wurgler 2006; Tetlock 2007; Da, Engelberg & Gao 2015). The instruments came first as curiosities and became, cumulatively, an argument.

This paper has a second purpose beyond review. The emotonomics program (the field paper) advances a stronger thesis than behavioral finance ever did: not that emotion distorts value, but that in a designed economic system emotion can be the primary unit of value — "attention is capital, gestures are currency, impact is profit." If that thesis is taken seriously, the sentiment-measurement toolkit changes role: it stops being a diagnostic of mispricing and becomes the system's national accounts. Section IV asks what instruments such a system requires, and builds them from the received literature rather than from slogans.

Definitions. Throughout, sentiment denotes the aggregate, time-varying disposition of market participants toward risk-bearing in an asset or asset class, insofar as that disposition is not reducible to changes in objective fundamentals. An instrument is any procedure producing a time series intended to proxy sentiment. An instrument is stated if it elicits beliefs (surveys), revealed if it infers disposition from costly behavior (flows, option purchases, on-chain transfers), and expressed if it infers disposition from communicative behavior (news text, social posts, search queries). This trichotomy organizes Section III.

II. The quarrel that made sentiment measurable

II.1 The classical inheritance

The word itself is older than the discipline. Adam Smith's first book was a Theory of Moral Sentiments (1759), and its central mechanism — sympathy, the fellow-feeling by which we internalize the judgments of an imagined impartial spectator — is recognizably a theory of socially transmitted evaluation. The crowd-pathology literature of the nineteenth century supplied the negative pole: Mackay's Extraordinary Popular Delusions and the Madness of Crowds (1841) fixed the genre of the mania narrative, and Kindleberger's Manias, Panics, and Crashes (1978) later gave it an analytic skeleton — displacement, credit expansion, euphoria, distress, revulsion — that reads today as a stage theory of sentiment.

The canonical modern source is Keynes. Chapter 12 of the General Theory (1936) makes two distinct claims that the later literature often merges. The first is motivational: long-term expectation cannot be grounded in calculation alone, so enterprise depends on "animal spirits — a spontaneous urge to action rather than inaction." The second is structural: under separation of ownership and trading, professional investors are driven to anticipate "what average opinion expects the average opinion to be" — the beauty-contest recursion — so that valuation becomes a convention sustained by confidence and vulnerable to it. Note what Keynes did not supply: any instrument. Animal spirits entered the discipline as an unmeasured residual, which is why the subsequent efficiency counterattack could dismiss them as a name for ignorance. The rehabilitation of the concept as a measured macro variable is recent (Akerlof & Shiller 2009).

II.2 The efficiency tradition, steelmanned

The efficient-markets school deserves its strongest statement, because the measurement literature exists to answer it. The argument is not that investors are rational; it is that markets are disciplined. Friedman (1953) supplied the selection mechanism: traders who misprice lose money to those who do not, so destabilizing speculation self-liquidates. Fama (1970) supplied the definitional apparatus — weak, semi-strong, and strong form efficiency — and an empirical program in which prices "fully reflect" available information. On this view sentiment may exist as psychology but cannot persist as pricing: any emotional displacement of price from value is an arbitrage opportunity, and arbitrage is the immune system.

The steelman matters because it makes a testable claim about instruments: any purported sentiment measure should have no forecasting power for returns beyond risk. The school's later concession was partial and disciplined — even Black (1986), in his presidential address "Noise," accepted that noise trading is pervasive and indeed constitutive of liquidity ("noise makes trading in financial markets possible"), while retaining the faith that price wanders within a factor-of-two band around value. The efficiency tradition, in other words, predicted that the sentiment-measurement program would find nothing. That prediction failed in a specific and instructive way.

II.3 The noise-trader and limits-of-arbitrage answer

The theoretical answer arrived as a model, not a manifesto. De Long, Shleifer, Summers, and Waldmann (1990) formalized an economy in which noise traders with stochastic, correlated misperceptions trade against rational arbitrageurs with finite horizons. The result overturns Friedman's selection argument: because noise-trader sentiment is itself a source of risk ("noise trader risk"), arbitrageurs bearing it demand compensation, mispricing persists, and — the deepest cut — noise traders can earn higher expected returns than their rational counterparts by loading on the very risk they create. Sentiment is no longer an error awaiting correction; it is a priced factor.

Shleifer and Vishny (1997) completed the argument institutionally: real-world arbitrage is delegated, capital-constrained, and evaluated on short horizons, so it is weakest precisely when mispricing is widest — performance-based fund flows force liquidation into deepening dislocation. Behavioral micro-foundations were supplied in parallel: prospect theory's loss-averse value function and probability weighting (Kahneman & Tversky 1979) explained why misperception is systematic rather than idiosyncratic, and Barberis, Shleifer, and Vishny (1998) modeled investor sentiment as a regime-switching belief process generating both underreaction and overreaction. Shiller's excess-volatility finding (1981) — price variance far exceeding the variance of subsequent dividend realizations — had already shown there was something to explain; Irrational Exuberance (2000) named the ambient state.

The dialectical outcome is precise: efficiency theory demanded that sentiment be unmeasurable or unpriced; noise-trader theory predicted it would be measurable, persistent, and priced, with its strongest effects in assets that are hard to value and hard to arbitrage. That conditional — sentiment bites hardest where valuation is most subjective and arbitrage most limited (Baker & Wurgler 2006) — is the single most confirmed regularity in the measurement literature, and it should be kept in view throughout what follows, because a young, thinly-traded, reflexively-narrated crypto asset is the limiting case of both conditions.

III. The instrument stack: how feeling is measured

III.1 Stated sentiment: surveys

The oldest instruments simply ask. George Katona's program at Michigan, begun in the 1940s and systematized in his Psychological Economics (1975), established that consumer confidence could be elicited, indexed, and used as a leading indicator; the University of Michigan Surveys of Consumers remain the canonical macro-sentiment series. In markets proper, the American Association of Individual Investors has polled its members weekly since 1987 (bullish / bearish / neutral on six-month equity prospects), and Investors Intelligence has classified newsletter writers as bulls or bears since the 1960s — a series whose enduring use is contrarian: extreme stated bullishness historically precedes weak returns, consistent with the noise-trader model in which surveyed enthusiasm proxies the sentiment factor near its peak.

Survey instruments have a clean epistemology (they measure beliefs directly) and two known pathologies: stated beliefs are cheap — no position backs the answer — and panels are unrepresentative. Their persistence in the stack owes to the first property: they are the only instruments that observe expectation rather than its consequences.

III.2 Revealed sentiment: prices of fear and flows of hope

The closed-end fund discount. Zweig (1973) proposed, and Lee, Shleifer, and Thaler (1991) canonized, the discount of closed-end fund prices to their net asset values as an index of individual-investor sentiment: the same assets, wrapped for retail, price differently as retail mood swings — and the discount comoves across funds and with small-stock returns, exactly as a systematic sentiment factor requires.

Derivatives-implied fear. The CBOE Volatility Index (VIX), constructed from S&P index option premia, measures the market price of near-term expected volatility. Whaley (2000) christened it "the investor fear gauge," and the name captured the epistemics: because equity index puts are the canonical insurance instrument, their implied volatility embeds the price participants will pay to shed downside — fear made numerical, in dollars, continuously. The put–call volume ratio serves the same logic more coarsely. These are the purest revealed-preference sentiment instruments in existence: no one buys portfolio insurance rhetorically.

Flows, issuance, and liquidity. Baker and Stein (2004) read unusual market liquidity itself as a sentiment index (overconfident investors trade more, and their presence lowers the price impact of trades); equity issuance timing — firms selling shares when sentiment is rich — supplies a corporate-side revelation of the same state.

III.3 Expressed sentiment: the computational turn

Antweiler and Frank (2004), studying millions of internet stock-message-board posts, gave the field its founding result and its founding caution: message activity predicts volatility, while return predictability, though statistically present, is economically small — talk is not just noise, but neither is it a money pump. Tetlock (2007) moved to mass media, showing that pessimistic word counts in a daily Wall Street Journal column predict downward pressure on prices followed by reversion — the signature of sentiment rather than information. Loughran and McDonald (2011) supplied the essential instrumentation lesson: general-purpose sentiment dictionaries misclassify finance text wholesale (in 10-K filings, words like "liability" or "cost" are operational, not negative), so domain-specific lexica are a validity requirement, not a refinement.

Two extensions define the modern frontier. Bollen, Mao, and Zeng (2011) claimed that Twitter mood dimensions (notably a "calmness" factor) improve directional forecasts of the Dow — the paper that licensed a decade of social-media sentiment engineering. Da, Engelberg, and Gao (2015) built FEARS — Financial and Economic Attitudes Revealed by Search — from the frequency of household Google queries such as "recession" and "bankruptcy," showing that search-revealed anxiety predicts short-horizon return reversals, volatility, and fund flows. Search data occupy a privileged epistemic position: queries are private expressive acts, free of the performative distortion that infects public posting. Garcia (2013), analyzing a century of New York Times financial text, added the state-dependence result: the return-predictive content of sentiment concentrates in recessions — feeling matters most when times are bad. The transformer era has industrialized the pipeline — FinBERT (Araci 2019) fine-tunes pre-trained language models on financial text, and Lopez-Lira and Tang (2023) report that general-purpose large language models score headline sentiment with economically meaningful predictive content — but the Loughran–McDonald lesson still governs: validity lives in the domain fit, not the model size.

III.4 The composite instruments: Fear & Greed

Practitioners fused the families into dashboards, and the fusions became the most-consulted sentiment instruments in the world.

The CNN Fear & Greed Index compresses seven revealed-preference components into a 0–100 dial — from extreme fear to extreme greed. Its inputs are instructive precisely because no survey appears among them: equity momentum against a moving-average benchmark; the breadth of new 52-week highs versus lows; advancing-versus-declining volume; the put–call ratio; junk-bond demand measured as the spread of high-yield over investment-grade debt; the VIX against its own recent history; and safe-haven demand measured as the relative performance of stocks versus Treasuries. Each is a price or flow — fear and greed inferred entirely from what capital does. Methodologically the index is a normalization-and-average heuristic, not an estimated factor model; its value is communicative. It converts the Keynesian confidence state into a single public number, and in doing so becomes part of the state it measures — a reflexivity we return to in Section VI.

The Crypto Fear & Greed Index transplants the design to Bitcoin-centric markets, blending realized volatility against recent baselines, market momentum and volume, social-media activity rates, Bitcoin's dominance share of total crypto capitalization, and search-trend data into the same 0–100 scale. The transplant quietly concedes the expressive turn: where CNN's equity dial uses only market internals, the crypto dial reserves substantial weight for social and search signals — an acknowledgment that in a retail-dominated, narrative-driven, 24/7 market, the conversation is a market internal. The crypto index's canonical use is contrarian ("extreme fear" as accumulation signal), which is the Investors Intelligence logic reborn on-chain.

The composites teach three design lessons for any new sentiment instrument: (i) diversify across stated/revealed/expressed families, because each fails differently; (ii) normalize each component against its own history, because sentiment is a state relative to baseline, not a level; (iii) publish a single legible number, because an instrument that participants can see becomes a coordination device — for better and for worse.

IV. The emotonomic turn: from distortion to denomination

IV.1 The inversion stated

Everything above shares one premise: there exists a fundamental value, and sentiment is a deviation from it — measurable, priced, exploitable, but parasitic on a value defined elsewhere. The emotonomics program inverts the premise. In an emotonomic system, the emotional transaction is not the noise around the signal; it is the signal. Value is constituted by attention, gestures, and impact — "the wealth of a community is measured by engagement velocity, not idle balances." This is a normative-design claim, not a positive-empirical one, and we flag it as such: it does not assert that existing markets price feeling correctly, but that a system can be built whose unit of account is a recorded act of feeling.

The claim has a respectable lineage it should own explicitly. It radicalizes Keynes's convention theory (if valuation is a confidence convention anyway, engineer the convention deliberately and benevolently); it operationalizes Smith's sympathy (the impartial spectator becomes a public ledger of gestures); and it accepts the noise-trader result at full strength — sentiment is priced and persistent — while refusing the pejorative: in De Long et al.'s economy the noise traders are the pathology, whereas in an emotonomic economy the "noise" — the human warmth in the channel — is the payload.

Definitions (from the emotonomics program, formalized). A gesture is an on-chain transfer carrying emotional intent from sender to receiver, logged with timestamp and optional context (the Proof of Gesture ledger). An attention event is a verified engagement act (view, share, post, interaction) bound to an identity. Impact is the downstream engagement causally attributable to a gesture. The SHAMBA LUV implementation gives these teeth: a fixed 111-quadrillion supply; a 3% reflection fee that redistributes every trade to all holders — a structural reward for holding, LUV grows when you hold LUV; an IncentiveDistributor contract paying denominated rewards for verified attention acts; and standardized gesture denominations (10⁹ LUV — "some LUV," the standard gesture; 10¹¹ — "a lot of LUV"; 10¹² — one trillion, a million millions, the measure of value).

IV.2 Native instruments

An economy denominated in feeling requires sentiment instruments in the national-accounts role. The emotonomics program proposes three, formalized here (all three are proposed instruments; none has yet a validated series):

To these ledger-native measures the live system adds market-side instruments already deployed on the SHAMBA LUV measurement stack: the single-line measure (one trillion LUV priced continuously in USDC), interval percent-change fields duplicating the practitioner standard, and the X multiplier — price and liquidity expressed as multiples of the genesis seed, which was deliberately placed at the round point 10⁻¹⁷ ETH per LUV so that every subsequent multiplier is a direct read. The X multiplier deserves theoretical note: by fixing an arbitrary but public and permanent baseline, it converts price into a self-normalizing sentiment series — design lesson (ii) of Section III.4 implemented at genesis.

IV.3 A LUV Fear & Greed composite (proposed)

Following the composite design lessons, we specify a seven-component LUV sentiment dial, each component normalized against its trailing distribution and averaged to a 0–100 scale from fear (hoarding, silence, exit) to — the emotonomic relabeling matters — LUV (gesture, voice, entry):

1. Momentum — price versus its trailing mean (revealed; CNN component transplanted).
2. 24-hour change — the live percent-change field (revealed).
3. Liquidity multiplier trend — growth of the pool's ETH leg from X (revealed; deepening liquidity is capital's confidence).
4. Gesture velocity — GVI against baseline (ledger-native revealed).
5. Attention rate — verified share/task submissions per period through the IncentiveDistributor rail (expressive, but costly — each act is identity-bound and reviewed, which mitigates the cheap-talk pathology of stated sentiment).
6. Holder growth and retention — net new holders and the fraction not reducing balances (revealed; the hold-LUV signal that the reflection mechanism structurally rewards).
7. Reciprocity health — median eROI drift (ledger-native).

Components 1–3 are implementable today from the existing minute-sampled market mirror; 4–7 await the Phase-3 attention rail. We label the composite proposed: its weights are undetermined, and Section VI explains why they should be estimated against outcomes rather than asserted.

V. Testable propositions

The emotonomic framework earns scientific standing only if it risks falsification. Five propositions, ordered from established-literature replications to native claims:

1. P1 — Sentiment susceptibility. As a hard-to-value, limits-to-arbitrage asset, LUV's price series will show sentiment-factor loadings at the extreme of the Baker–Wurgler conditional: return reversals following expressed-sentiment spikes, strongest at short horizons.
2. P2 — Gesture velocity leads price. If gestures constitute value rather than merely celebrating it, GVI should lead market-side measures rather than lag them; Granger-style precedence tests on the Proof of Gesture ledger against the minute-sampled price series are directly computable.
3. P3 — Reflection retention. Holders' effective yield from the 3% reflection stream should predict retention (non-selling) beyond price momentum — the structural hold-to-earn incentive should be separable from bandwagon holding.
4. P4 — Contrarian dial. Extreme readings of the LUV composite should exhibit the same contrarian asymmetry documented for the crypto Fear & Greed index: fear extremes preceding above-median forward returns more reliably than greed extremes precede below-median ones (Garcia's 2013 state-dependence, transplanted).
5. P5 — Reciprocity stability. Networks with eROI distributions concentrated near unity will exhibit lower participant churn than reward-equivalent networks with skewed reciprocity — the claim that measured mutuality, not payout size, is the retention variable. This is the distinctively emotonomic prediction; the incentive-design literature does not make it.

VI. Objections answered

VII. Conclusion

Sentiment research began as an insult ("animal spirits"), matured into a factor, and industrialized into an instrument stack: surveys that ask, derivatives that price fear in dollars, texts and searches that betray mood at scale, and composites — Fear & Greed foremost — that compress the state of collective feeling into one public number. The emotonomic turn takes the final step the literature prepared but never took: if feeling is measurable, persistent, and priced, a system may be built that denominates value in it deliberately. The instruments proposed here — gesture velocity, resonance depth, reciprocity health, and a seven-component LUV composite anchored by a genesis-fixed baseline — are that step's engineering documents. Their scientific fate rests on Propositions P1–P5, and the program should want it that way: a measure of feeling that fears no measurement of itself.

References

• Akerlof, G. A., & Shiller, R. J. (2009). Animal Spirits: How Human Psychology Drives the Economy, and Why It Matters for Global Capitalism. Princeton University Press.
• Alternative.me (ongoing). "Crypto Fear & Greed Index." Published methodology: volatility, market momentum/volume, social media, Bitcoin dominance, search trends.
• Antweiler, W., & Frank, M. Z. (2004). "Is All That Talk Just Noise? The Information Content of Internet Stock Message Boards." Journal of Finance 59(3): 1259–1294.
• Araci, D. (2019). "FinBERT: Financial Sentiment Analysis with Pre-trained Language Models." arXiv:1908.10063.
• Baker, M., & Stein, J. C. (2004). "Market Liquidity as a Sentiment Indicator." Journal of Financial Markets 7(3): 271–299.
• Baker, M., & Wurgler, J. (2006). "Investor Sentiment and the Cross-Section of Stock Returns." Journal of Finance 61(4): 1645–1680.
• Baker, M., & Wurgler, J. (2007). "Investor Sentiment in the Stock Market." Journal of Economic Perspectives 21(2): 129–152.
• Barberis, N., Shleifer, A., & Vishny, R. (1998). "A Model of Investor Sentiment." Journal of Financial Economics 49(3): 307–343.
• Black, F. (1986). "Noise." Journal of Finance 41(3): 529–543.
• Bollen, J., Mao, H., & Zeng, X. (2011). "Twitter Mood Predicts the Stock Market." Journal of Computational Science 2(1): 1–8.
• CNN Business (ongoing). "Fear & Greed Index." Published methodology: momentum, price strength, breadth, put/call, junk-bond demand, VIX, safe-haven demand.
• Da, Z., Engelberg, J., & Gao, P. (2015). "The Sum of All FEARS: Investor Sentiment and Asset Prices." Review of Financial Studies 28(1): 1–32.
• De Long, J. B., Shleifer, A., Summers, L. H., & Waldmann, R. J. (1990). "Noise Trader Risk in Financial Markets." Journal of Political Economy 98(4): 703–738.
• Fama, E. F. (1970). "Efficient Capital Markets: A Review of Theory and Empirical Work." Journal of Finance 25(2): 383–417.
• Friedman, M. (1953). "The Case for Flexible Exchange Rates." In Essays in Positive Economics. University of Chicago Press.
• Garcia, D. (2013). "Sentiment during Recessions." Journal of Finance 68(3): 1267–1300.
• Goodhart, C. A. E. (1975). "Problems of Monetary Management: The U.K. Experience." In Papers in Monetary Economics, vol. I. Reserve Bank of Australia.
• Kahneman, D., & Tversky, A. (1979). "Prospect Theory: An Analysis of Decision under Risk." Econometrica 47(2): 263–291.
• Katona, G. (1975). Psychological Economics. Elsevier.
• Keynes, J. M. (1936). The General Theory of Employment, Interest and Money. Macmillan. (Ch. 12, "The State of Long-Term Expectation.")
• Kindleberger, C. P. (1978). Manias, Panics, and Crashes: A History of Financial Crises. Basic Books.
• Lee, C. M. C., Shleifer, A., & Thaler, R. H. (1991). "Investor Sentiment and the Closed-End Fund Puzzle." Journal of Finance 46(1): 75–109.
• Lopez-Lira, A., & Tang, Y. (2023). "Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models." arXiv:2304.07619.
• Loughran, T., & McDonald, B. (2011). "When Is a Liability Not a Liability? Textual Analysis, Dictionaries, and 10-Ks." Journal of Finance 66(1): 35–65.
• Mackay, C. (1841). Extraordinary Popular Delusions and the Madness of Crowds. Richard Bentley.
• Shiller, R. J. (1981). "Do Stock Prices Move Too Much to Be Justified by Subsequent Changes in Dividends?" American Economic Review 71(3): 421–436.
• Shiller, R. J. (2000). Irrational Exuberance. Princeton University Press.
• Shleifer, A., & Vishny, R. W. (1997). "The Limits of Arbitrage." Journal of Finance 52(1): 35–55.
• Smith, A. (1759). The Theory of Moral Sentiments. A. Millar.
• Soros, G. (1987). The Alchemy of Finance: Reading the Mind of the Market. Simon & Schuster.
• Strathern, M. (1997). "'Improving Ratings': Audit in the British University System." European Review 5(3): 305–321.
• Tetlock, P. C. (2007). "Giving Content to Investor Sentiment: The Role of Media in the Stock Market." Journal of Finance 62(3): 1139–1168.
• Whaley, R. E. (2000). "The Investor Fear Gauge." Journal of Portfolio Management 26(3): 12–17.
• Zweig, M. E. (1973). "An Investor Expectations Stock Price Predictive Model Using Closed-End Fund Premiums." Journal of Finance 28(1): 67–78.