Process vs. Outcome and Expected Value Thinking

Key Principle

Investment success is governed by process quality and expected-value discipline, not by the frequency of correct predictions. In probabilistic domains, any single outcome is dominated by noise. The only lever an investor controls is the decision process. Over sufficient iterations, a superior process produces superior aggregate outcomes — identical to how a casino profits despite losing individual hands.

Expected value integrates both the probability and magnitude of each outcome. The formula is simple — EV = sum of (probability of each outcome x payoff of that outcome) — but its implications are counterintuitive. Chasing high hit rates while ignoring payoff asymmetry systematically destroys returns. This is the Babe Ruth Effect: Ruth led the league in both home runs and strikeouts. The magnitude of his hits overwhelmed the frequency of his misses.

A third force compounds the problem: myopic loss aversion. Loss aversion (~2.5x pain asymmetry) is biological and fixed, but evaluation frequency is a policy choice. When investors check returns too often, they experience near-random fluctuations filtered through loss aversion, converting a positive-EV position into a negative-utility experience. The remedy is structural — lengthen the evaluation period.

The Process-Outcome Matrix (Russo & Schoemaker)

The matrix forces explicit separation of process quality from outcome quality, preventing belief-updating from single data points. Most organizations live in the outcome columns and never examine the rows.

Why This Matters

• Outcome bias is structurally misleading. Abandoning a good process after bad outcomes is common and costly. Maintaining a bad process after good outcomes is equally common but invisible until large losses reveal it. The failure mode is asymmetric: bad-process/good-outcome is celebrated while good-process/bad-outcome is punished. (Ch. 1)
• Loss aversion distorts strategy selection. Losses feel ~2.5x as painful as equivalent gains (Kahneman & Tversky, 1979). This creates a causal chain: loss aversion makes frequent losses painful, so investors prefer high-frequency strategies (many small wins), which are not necessarily high-expected-value strategies. The result is a systematic preference for portfolios that feel good but underperform. (Ch. 3)
• Evaluation frequency amplifies the damage. Checking returns hourly yields ~50.4% positive observations; annually yields ~72.6%. Myopic evaluation converts a positive-EV equity portfolio into a negative-utility experience, pushing investors out of their best positions. The evaluation period — not the planning horizon — determines behavior. A retirement saver checking quarterly is behaviorally a short-term investor. (Ch. 8)
• Industry incentives reinforce the wrong metric. The mutual fund business rewards asset gathering, client retention, and index-hugging — all frequency-maximizing tactics. The investment profession rewards long-term magnitude of returns. Index-hugging is explicitly a frequency-maximizing strategy: it minimizes tracking error (frequent small deviations) at the cost of large outperformance. (Ch. 2)

Good Examples

• Taleb's short position (Ch. 3). Taleb was short S&P futures despite believing the market would most likely rise. The most probable outcome (70% chance of +1%) and the best bet (-30% chance of -10%, EV = -2.3%) pointed in opposite directions. Bullish conviction and short positioning are not contradictory when magnitudes are asymmetric.
• The near-fired portfolio manager (Ch. 3). A manager ranked among the best total performers in a group of ~20 but had the worst percentage of stocks beating the benchmark. The company treasurer nearly fired him. Frequency-based evaluation systematically eliminates the best-performing managers.
• DePodesta blackjack anecdote (Ch. 1). A player hits on 17, draws a 4, and wins. The outcome was good; the decision was terrible. This is the "dumb luck" quadrant of the process-outcome matrix.
• Priced-for-perfection stock (Ch. 1). 75% chance of +1%, 25% chance of -10%. EV = (0.75 x 1%) + (0.25 x -10%) = -1.75%. Great probability but negative expected value.
• Samuelson's colleague (Ch. 8). Refused a single $200-win/$100-loss coin toss (positive EV) because "I would feel the $100 loss more than the $200 gain" — but would accept 100 such bets. Narrow framing causes people to evaluate bets in isolation rather than as part of total wealth, rejecting positive-EV opportunities one at a time.
• Evaluation frequency and utility (Ch. 8). At a 1-hour evaluation period, equity returns are positive only ~50.4% of the time, yielding deeply negative experienced utility. At a 1-year period, positive returns jump to ~72.6% and utility turns positive. Same portfolio, same returns — different experience based solely on how often you look.

Counterpoints

• Process evaluation is subjective and harder to observe. Outcomes are objective and visible; this asymmetry biases all observers — investors, boards, allocators — toward outcome-based judgment, making process discipline organizationally difficult to sustain. The investment industry's incentive and measurement systems actively work against process-based evaluation — organizations punish correct-process/bad-outcome decisions, creating perverse incentives that deter necessary risk-taking. (Ch. 1)
• Time-horizon diversity is necessary. Not all investors can adopt long evaluation periods. If everyone did, the equity-risk premium they exploit would vanish. The system requires participants across all time horizons. The "Dow 36,000" theory is cited as a fallacy of composition: universal long-horizon investing would eliminate the premium it depends on. (Ch. 8)
• Information overload undermines process confidence. More information increases confidence without increasing accuracy (handicapper study: accuracy flat at ~17% from 5 to 40 variables, while confidence rose from ~20% to ~30%+). Analysts with more data construct more convincing narratives, not better forecasts. The well-informed analyst can be more dangerous than the lightly-informed one because higher confidence drives larger position sizes. (Ch. 1)
• Profession-vs-business tension is structural, not individual. Ownership consolidation (only 6 of 50 largest fund organizations privately held) creates pressure to maximize firm earnings through rising expense ratios and fund proliferation. Market returned 12%, average fund <10%, average investor 6.9% (1986-2005). The gap is not fixable by individual process discipline alone. (Ch. 2)

Key Quotes

"Individual decisions can be badly thought through, and yet be successful, or exceedingly well thought through, but be unsuccessful, because the recognized possibility of failure in fact occurs. But over time, more thoughtful decision-making will lead to better overall results." — Robert Rubin (Ch. 1)

"Any time you make a bet with the best of it, where the odds are in your favor, you have earned something on that bet, whether you actually win or lose the bet." — David Sklansky (Ch. 1)

"The frequency of correctness does not matter; it is the magnitude of correctness that matters." (Ch. 3)

"The issue is not which horse in the race is the most likely winner, but which horse or horses are offering odds that exceed their actual chances of victory." — Steven Crist (Ch. 1)

"Loss aversion can be considered a fact of life. In contrast, the frequency of evaluations is a policy choice that presumably could be altered, at least in principle." — Benartzi & Thaler (Ch. 8)

"Long-term investors are willing to pay more for an identical risky asset than short-term investors." (Ch. 8)

"Perhaps the single greatest error in the investment business is a failure to distinguish between the knowledge of a company's fundamentals and the expectations implied by the market price." (Ch. 1)

"A quality investment philosophy is like a good diet: it only works if it is sensible over the long haul and you stick with it." (Introduction)

Rules of Thumb

1. Use the process-outcome matrix. After every significant decision, classify it in the Russo-Schoemaker 2x2 (good/bad process x good/bad outcome). Never update beliefs from a single data point.
2. Replace target prices with scenario trees. Map upside, base, and downside cases with explicit probabilities. This counters anchoring and forces acknowledgment of unfavorable outcomes.
3. Ask "what are the expectations embedded in price?" before asking "is this a good company?" Variant perception — a well-founded view meaningfully different from consensus — is the source of edge, not fundamental knowledge alone.
4. Extend your evaluation period. Loss aversion is fixed; evaluation frequency is a policy choice. Checking less often converts the same portfolio from negative to positive experienced utility.
5. Tolerate low hit rates when magnitude compensates. A portfolio with 25% of stocks winning can outperform if the winners are large enough. Resist the urge to prune losers prematurely to raise the batting average.
6. Apply Rubin's four principles sequentially (Ch. 1). (1) The only certainty is no certainty — widen outcome ranges. (2) Decisions are about weighing probabilities — think in expected value, not directional conviction. (3) Despite uncertainty, act — resist gathering more data for false comfort. (4) Judge decisions by how they were made, not only by results — build feedback loops that reward process.
7. Distinguish risk from uncertainty. Risk has a known distribution; uncertainty does not. Treating uncertainty as risk leads to overconfident position sizing and catastrophic loss. Corporate outcomes are uncertain; roulette is risky.

Diagram

Related References

• [Variant perception and expectations investing — Ch. 1]
• [Profession vs. business dynamics in fund management — Ch. 2]
• [Myopic loss aversion and evaluation frequency — Ch. 8]
• [Complex adaptive systems and diversity collapse — Part 4]