Degree of Rejoicing and Regret (DRR) (喜悦与后悔程度)
A stock-level cross-sectional return predictor that measures the intensity of investors' emotional experience of rejoicing (when a stock outperforms the industry median) versus regret (when it underperforms). DRR is the first empirical proxy in asset pricing to incorporate both positive and negative emotions from Loomes & Sugden's (1982, 1987) regret-rejoice choice theory, as operationalized by So & Zhang (2026).
Construction
For stock i at month t:
Volatility-Adjusted Return (VAR): VAR_{i,t} = r_{i,t} / \sqrt{Var_d(r_{i,t}) \cdot N}
- r_{i,t} = monthly return
- Var_d(r_{i,t}) = variance of daily returns over the past month
- N = number of trading days
DRR: within the same industry (3-digit SIC for US, 6-digit GICS for international): DRR_{i,t} = \begin{cases} \frac{VAR_{i,t} - \overline{VAR}_t}{\overline{VAR}^h_t - \overline{VAR}_t}, & VAR_{i,t} \geq \overline{VAR}_t [6pt] \frac{VAR_{i,t} - \overline{VAR}_t}{\overline{VAR}_t - \overline{VAR}^l_t}, & VAR_{i,t} < \overline{VAR}_t \end{cases}
- DRR \in [-1, 1]; positive = rejoicing, negative = regret
- Higher DRR → investors anticipate rejoicing → demand lower expected returns → lower future returns
- Lower DRR → investors anticipate regret → demand higher expected returns → higher future returns
Key empirical properties
- Magnitude: DRR-sorted long-short earns EW −16.45% / VW −9.95% annually (US, 1963–2023)
- Monotonicity: nearly monotonic across deciles — factor-adjusted alphas are positive for low-DRR (regret) deciles, negative for high-DRR (rejoicing) deciles, and zero near DRR=0
- Persistence: DRR formed over up to 10-month windows still predicts returns (diminishing with horizon)
- Factor robustness: survives all standard models (4F, 5F, 7F, Q5F, M4F, BF2F)
Relationship to other predictors
| Predictor | Relationship | Key finding |
|---|---|---|
| Short-Term Reversal (STR) | r=0.68 | DRR subsumes STR; STR becomes insignificant after controlling for DRR |
| Arisoy et al. (2024) REG | r=−0.31 | DRR subsumes REG; REG does NOT subsume DRR |
| Salience Theory Value (ST) | r=0.46 | Weak partial correlation |
| MAX effect | r=0.21 | Minimal overlap |
| Prospect Theory Value (PTVB) | weak | DRR remains significant after PTVB control |
What drives DRR's power
Three distinct channels identified by So & Zhang (2026):
- Joint regret+rejoice: DRR integrates both emotional components; each alone (REG, REJ) is weaker
- Volatility adjustment: firm-level vol adjustment beats raw-return DRR, which is itself subsumed by vol-adjusted DRR
- Relative scaling: industry min-max scaling eliminates industry concentration and captures cross-industry vol heterogeneity
Global evidence
- 44 international markets (1990–2023): EW −0.70%/mo (t=−6.98), VW −0.54%/mo (t=−5.27)
- Stronger in developed than emerging markets
- Stronger in high-Individualism (Hofstede) and high-Uncertainty Avoidance (Hofstede) countries
- Stronger in weak-Investor Protection countries
- Regional variation: NA and Japan strong; APAC (ex-Australia) weak; UK shows anomalous positive EW spread
References
- So, R.H.Y. & Zhang, X. (2026). Rejoicing, regret and stock returns – US and international evidence. Journal of Banking and Finance 190, 107742.
- Loomes, G. & Sugden, R. (1982). Regret theory: an alternative theory of rational choice under uncertainty. Economic Journal 92, 805.
- Arisoy, Y.E., Bali, T.G. & Tang, Y. (2024). Investor regret and stock returns. Management Science 70, 7537–7558.