Base Rates

Short forms used in this file: [MW] = Mauboussin & Wang, "The Base Rate Book," Credit Suisse (2016); [Kahn] = Kahneman, Thinking, Fast and Slow (Farrar Straus & Giroux, 2011); [Greenblatt] = Greenblatt, You Can Be a Stock Market Genius (Simon & Schuster, 1997); [Chancellor] = Chancellor (ed.), Capital Returns: Investing Through the Capital Cycle (Palgrave Macmillan, 2015).

The outside view. Before believing any specific thesis about a specific company in a specific industry at a specific time, the agent looks at the historical base rate for that category of outcome. Base rates discipline inside-view enthusiasm and prevent reasoning from a sample of one. T2

The concept

The "inside view" vs. "outside view" distinction is from Kahneman: T2

• Inside view: based on the specific details of this situation, with all its idiosyncratic features
• Outside view: based on the reference class of similar situations, with the average outcome

The inside view is what feels like understanding ("I've studied this company carefully and the management is excellent"). The outside view is what produces calibrated predictions ("turnaround attempts by new CEOs in declining industries succeed 15% of the time historically").

When the two diverge, the outside view is usually closer to reality. T2

Useful base rates the deep-value agent keeps in mind

M&A

• 50-70% of large M&A destroys shareholder value T2
• Synergies achieved: typically 40-70% of management's announced targets T2
• Major integration challenges: > 50% of large deals T2
• Goodwill impairments occur on > 30% of large deals within 5 years T2

Turnarounds

• Successful operating turnarounds by new CEOs: 20-30% historically T2
• Multi-year turnarounds in declining industries: 10-15% T2
• Turnarounds in commodity-cycle businesses (not really turnarounds, but cycle plays): much higher base rate T2

Industry transitions

• Incumbents successfully navigating major technology transitions: 20-30% of historical cases T2
• Examples of incumbents that succeeded: IBM (mainframe → services), Microsoft (PC → cloud), JPMorgan (regional → global)
• Examples that failed: Kodak, Polaroid, Blockbuster, BlackBerry, much of pre-internet retail T2

"This time is different"

• Persistent margin levels above industry historical average: very low base rate of sustaining T2
• "New economy" / "platform" / "asset light" narratives: significant base rate of regression to mean T2
• Promises of permanent disruption: usually partially true but often slower than promised

Founder transitions

• Founder-to-professional CEO transitions successful: ~50% AS-cal
• Founder reverting to operational role: usually a negative signal
• Family successions: < 30% success rate by third generation (Buddenbrooks effect) T2

IPOs

• Underperformance of IPOs vs. market on 3-5 year horizon: well-documented in studies T2
• Concentration of returns in a small number of winners: extreme T2
• Median IPO outcome: significant underperformance T2

Hot themes and narratives

• Sector-level returns following peak narrative coverage: significantly below earlier years T2
• Specific examples: nifty-fifty (1972 peak, decade of poor returns) T2; Japan (1989 peak, decades); tech (2000 peak) T2; commodities (2014 peak)

Earnings surprises

• Companies that miss earnings by more than 10%: significant probability of further misses T2
• Companies that beat by more than 10%: weaker pattern of continued beats T2
• "Beat-and-raise" patterns: typically not sustainable beyond 4-6 quarters AS-cal

Cyclical recoveries

• Cyclical industries with capacity above 90% utilization within 2 years tend to experience trough-to-peak earnings expansion AS-cal
• Trough-to-peak cyclical recoveries: typical 3-5 year cycle T2

Spin-offs

• Spin-offs outperform broader market in 12-36 months post-spin: well-documented T2
• Parent (RemainCo) returns: more mixed T2
• Concentration of outperformance in spin-offs that had been treated as orphans T2

Bankruptcies / restructurings

• Equity recovery in bankruptcy: typically very low (often zero) T2
• Pre-bankruptcy equity: high probability of further loss before final outcome
• Post-bankruptcy reorgs: often perform well for first-day equity holders (low base, normalizing operations) T2

How to use base rates

As priors

Before believing a specific case beats the base rate, identify why this case is different. The differences must be structural, not just optimistic narrative.

A claim that "this M&A will create value despite the 50%+ base rate failure" requires specific reasoning:

• Why are synergies more achievable here?
• Why is integration more straightforward?
• Why is the price paid below the average peak-multiple peer transaction?

If the answers are weak, defer to the base rate.

As calibration check

After completing a thesis, compare the implied outcome to the historical base rate. If the inside-view forecast is far from the base rate, ask: is the difference justified by specific structural advantages, or is the optimism unjustified?

As source of skepticism

When the dominant narrative says "X is going to happen," check the base rate of analogous narratives. The narrative may be right; the base rate provides context for how often similar narratives have been right historically.

Building your own base rates

The agent maintains a working database of base rates for specific outcome categories, updated as new evidence appears. Sources:

• The canonical practitioner reference: T2 — Mauboussin & Wang's The Base Rate Book (Credit Suisse 2016) covers sales growth, operating margins, ROIC persistence, M&A, and more
• Academic finance papers (Sloan accruals, Greenblatt spin-offs, Damodaran general)
• Long-running consultancy studies (McKinsey, BCG on M&A)
• Industry-specific historical studies
• Personal calibration log of past predictions and outcomes

Don't rely on memory; build documentation.

Common base-rate mistakes

1. Treating each case as unique

"This is different because…" is the most common base-rate-defeating phrase. Often the speaker is right that the case is unique, but on the wrong dimension — the uniqueness doesn't actually affect the outcome.

2. Selecting reference class to favor the inside view

The reference class is malleable. The same situation can be cast as "tech company" (one base rate), "founder-led startup" (different base rate), or "this specific industry vertical" (third base rate). Choose the reference class before the conclusion, not after.

3. Using narrow reference classes

"This is only the 3rd Y-class spin-off in the past decade." A reference class of 3 is too small to compute base rates from. Expand the class or accept that base rate evidence is weak.

4. Ignoring base rates entirely

The most common mistake. Analysis proceeds inside-view only, with no reality check.

The base-rate-vs-narrative tension

For each thesis, the tension between base rate and specific narrative:

• Narrative + base rate aligned: strong conviction. The story matches what typically happens.
• Narrative + base rate diverge: weaker conviction. Either the narrative is right and the case is genuinely unusual (rare), or the narrative is the standard optimistic case that usually disappoints.

When narrative diverges from base rate, the burden is on the narrative to justify the divergence with specific structural reasons.

What base rates do not tell you

• The timing of any specific outcome
• The path to the outcome
• Whether your specific thesis is in the part of the distribution that wins or loses

Base rates are statistical regularities about populations, not predictions about individuals. The role is to calibrate, not to determine.

Calibration over time

The agent maintains a calibration log:

• Predictions made (with probabilities)
• Outcomes
• Was the probability calibrated?

Over time, this reveals systematic biases:

• Over-confident predictions (events you said were 80% likely happen 60% of the time)
• Pattern of underestimating tails
• Specific topic areas where calibration is poor

These patterns improve future judgment.

Output

Base-rate usage in any thesis includes:

1. Reference class identified explicitly
2. Base rate stated
3. Comparison to inside view
4. Structural reasons (if any) for divergence
5. Implication for thesis confidence

Sources

• Mauboussin & Wang, The Base Rate Book, Credit Suisse (Sept. 2016) — the master practitioner reference for sales growth, operating margin, ROIC, and earnings persistence base rates
• Kahneman, Thinking, Fast and Slow (Farrar Straus & Giroux, 2011) — esp. ch. 23 on inside vs. outside view; also Kahneman & Lovallo, "Delusions of Success," Harvard Business Review (July 2003)
• Ritter, "The Long-Run Performance of Initial Public Offerings," Journal of Finance (1991); Loughran & Ritter, "The New Issues Puzzle," Journal of Finance (1995) — IPO long-run underperformance
• Bessembinder, "Do Stocks Outperform Treasury Bills?" Journal of Financial Economics (2018) — concentration of long-run returns in a small number of stocks
• Greenblatt, You Can Be a Stock Market Genius (Simon & Schuster, 1997) — the practitioner reference on spin-offs and special situations
• McConnell & Ovtchinnikov, "Predictability of Long-Term Spinoff Returns," Journal of Investment Management (2004) — academic spin-off evidence
• Bernard & Thomas, "Post-Earnings-Announcement Drift," Journal of Accounting Research (1989) and "Evidence That Stock Prices Do Not Fully Reflect…," Journal of Accounting and Economics (1990) — PEAD
• Bruner, Applied Mergers and Acquisitions (Wiley, 2004) — esp. ch. 3 on M&A value creation/destruction
• McKinsey & Company M&A practice publications (multi-year series); KPMG goodwill impairment studies
• Bibeault, Corporate Turnaround (McGraw-Hill, 1982); Slatter & Lovett, Corporate Turnaround (Penguin, 1999) — turnaround base rates
• Christensen, The Innovator's Dilemma (Harvard Business School Press, 1997) — incumbent technology-transition outcomes
• Shiller, Irrational Exuberance (Princeton, 2000) — narrative cycles in markets
• Siegel, Stocks for the Long Run (McGraw-Hill, multiple editions) — Nifty Fifty and long-run equity evidence
• Chancellor (ed.), Capital Returns: Investing Through the Capital Cycle (Palgrave Macmillan, 2015) — cyclical recovery base rates
• Gu & Lev, "Intangible Assets: Measurement, Drivers, and Usefulness," Accounting Review (2011) — goodwill impairment evidence
• American Bankruptcy Institute (ABI) statistics; PwC Family Business Survey (recurring) — bankruptcy and family-succession base rates

This file synthesizes named primary sources. Quantitative thresholds tagged <span class="tier-cal" title="...">AS-cal</span> are AlphaSteve's own calibrations and are revisable. Some base rates are stated as ranges because published estimates from different studies span the range.

Linked

• cognitive-bias-checklist — base-rate neglect is a bias
• variant-perception — variant must clear the base rate
• tail-risk-and-fat-tails — base rates of tail events
• 03-mental-models — model 5 (base rates)
• narrative-cycle — narratives often violate base rates