Updating Probability Weights Over Time: Dynamic Scenario Analysis
A valuation built on a static set of probabilities is a photograph, not a film. When you first value a stock at $35 across bull, base, and bear cases, you've made implicit bets that 30% of investors are right about the future, 50% are right, and 20% are right. But stock prices move 3-5% weekly based on earnings surprises, management commentary, and competitive developments. These movements contain information—they are telling you that your original probability weights may have been wrong.
The investor who doesn't update scenario weights is essentially ignoring the market's continuous reassessment of likelihood. A company that guided for 5% growth but delivers 2% is not a 2% story anymore; it's a signal that the base case was too optimistic. The investor who rigidly sticks to a "50% base case" despite this evidence is making a valuation error.
Quick definition: Updating probability weights means revising your estimate of how likely each scenario is as new information arrives. If your base case assumed 5% revenue growth and the company just reported 2% growth with pessimistic guidance, you should reduce the probability of the base case from 50% to 35-40%, shifting weight to the bear case. This is how real-world valuation thinking works—fluid, data-driven, always in motion.
Key takeaways
- Scenario probabilities are not fixed; they must evolve as new information arrives
- Earnings misses, management guidance changes, and competitive moves are evidence that forces probability reassessment
- A disciplined framework for updating (Bayesian thinking) prevents both overreaction to noise and underreaction to signals
- Over-updating to short-term volatility is a common mistake; so is under-updating to genuine structural shifts
- Portfolio positions should be revalued and reweighted quarterly at minimum, more frequently during material developments
- Systematic probability updates create a living valuation that tracks reality rather than remaining anchored to initial assumptions
The Static Valuation Problem
Imagine you valued a software company in January using these assumptions:
| Scenario | Probability | Revenue Growth | Operating Margin | Exit Multiple | Fair Value |
|---|---|---|---|---|---|
| Bull Case | 30% | 20% CAGR | 35% | 6x Revenue | $65 |
| Base Case | 50% | 8% CAGR | 20% | 4x Revenue | $40 |
| Bear Case | 20% | 2% CAGR | 10% | 2x Revenue | $15 |
| Weighted Fair Value | 100% | $41.50 |
You buy at $42, thinking you're getting fair value with an option to the upside if the bull case materializes.
In March, the company reports Q1 earnings:
- Revenue growth: 4%, not 8%
- Operating margins: 16%, not 20%
- Guidance: "expecting 3-5% growth for the year"
- Management commentary: "macro headwinds, customer churn at SMB segment, delayed enterprise deals"
This is not noise. This is evidence. The base case assumption of 8% growth is now outdated. The true base case is more like 4-5% growth. But what should your new probability weights be?
A disciplined investor recalculates:
| Scenario | Old Probability | New Probability | Rationale |
|---|---|---|---|
| Bull Case (20% growth) | 30% | 10% | Management guided 3-5%, ruling out 20% trajectory |
| Base Case (8% growth) | 50% | 20% | Original base case is now above guidance; less likely |
| New Base Case (4% growth) | — | 50% | Management guided 3-5%, so 4% is most likely now |
| Bear Case (2% growth) | 20% | 20% | Still possible, but guidance suggests 4% is more likely |
New Fair Value at 4% growth base case: ~$32. Your fair value has fallen from $41.50 to $32, an 23% repricing based on the actual results, not sentiment.
If you don't update, you're anchored to an outdated base case. If you do update, you recognize that the investment case has deteriorated and must either sell at $42 (taking a 23% loss on fair value) or reassess whether the stock is still worth owning at that price.
Why Investors Fail at Probability Updating
Most investors fall into one of two traps:
Trap 1: Over-updating to Noise
Every single announcement triggers a repricing. Earnings beat by $0.01? Buy. Miss by $0.01? Sell. Quarterly probability shifts become extreme and chaotic. This leads to excessive trading, transaction costs, and whipsaw losses.
Over-updaters assume that short-term data is high-quality signal. But quarterly earnings are noisy—guidance changes get missed, one-time items distort results, sector headwinds are temporary. Over-updating mistakes noise for signal and results in constant position rewrites.
Trap 2: Under-updating to Signal
A company's growth slows from 15% to 10% to 6% over three years, yet the investor sticks with a "bull case of 12% growth" because "I believe in the company long-term." This is anchoring—the willingness to ignore evidence because it contradicts your original thesis.
Under-updaters confuse faith with analysis. They hold positions for emotional reasons rather than probabilistic ones. They see deteriorating fundamentals but rationalize them away ("it's a one-time thing", "management knows what they're doing"). Two years later, they own a company that has permanently decelerated but they priced it for continued growth.
The truth is between these extremes: update materially on strong signals, be more cautious about weak signals, and revalue systematically rather than reactively.
A Framework for Disciplined Probability Updates
Use this three-step process when new information arrives:
Step 1: Identify the Evidence
When data arrives, ask: Is this a genuine signal about the scenario that will play out, or is it noise around the true trend?
Examples of strong signals:
- Management guidance changes (especially downward; executives are conservative about revisions)
- Structural competitive changes (a new competitor, a product disruption, regulatory action)
- Persistent margin compression (if it appears for two consecutive quarters, it's likely structural)
- Customer concentration loss (losing major clients signals market share loss)
Examples of weaker signals:
- Single-quarter beats or misses
- Cyclical industry timing (a down quarter in a typically cyclical sector)
- One-time items or accounting charges
- Macro headwinds that are expected to reverse
Step 2: Assess Scenario Likelihood
Given the evidence, which scenario becomes more or less likely?
Example 1: A software company misses revenue by 5% in one quarter due to delayed enterprise deals.
- This suggests the base case (8% growth) was too optimistic.
- It does not suggest the bear case (2% growth) is likely; delayed deals will eventually close.
- New weight: reduce base case from 50% to 40%, increase bear case from 20% to 30%.
Example 2: A hardware manufacturer's gross margins have compressed 200 basis points over two years due to wage inflation and raw material costs.
- This is structural, not cyclical; margin compression is not expected to reverse.
- The base case (20% margins) is now unrealistic.
- New weight: shift 15-20% from base to bear case, which assumes 15% margins.
Example 3: A biotech stock's lead drug gets FDA approval ahead of schedule.
- This is strong evidence for the bull case (drug reaches market faster, captures sales sooner).
- New weight: increase bull case from 25% to 40%, reduce base case from 50% to 40%.
Step 3: Recalculate Fair Value and Reposition
Once probabilities are updated, recalculate fair value. If it's materially different (10%+ change), reposition:
- If new fair value is 15% higher, add to the position or trim if already full-sized.
- If new fair value is 15% lower, trim or exit entirely depending on conviction in the remaining upside.
- If new fair value is similar but with much wider range of outcomes, reduce position size (you're less certain of direction).
Example: Your software stock was fairly valued at $41.50 with 50% probability in the base case. After the guidance miss, fair value drops to $32 with only 20% probability in the old base case. At a current price of $42, the stock is now 31% overvalued relative to the new fair value. This is a signal to sell.
Quantifying Updates: Bayesian Thinking
A more formal way to think about updating is Bayesian:
Posterior Probability = Prior Probability × Likelihood of Evidence / Prior Probability of Evidence
This means: Your updated belief should be your original belief, adjusted by how likely the new evidence is under each scenario.
Example: You originally thought a 50% chance of 8% revenue growth (base case). The company just reported 4% growth.
- How likely is 4% growth under the base case (8% expected)? Fairly likely; quarterly can vary around trend. Say 40% chance.
- How likely is 4% growth under the bear case (2% expected)? Less likely, but possible. Say 20% chance.
- How likely is 4% growth under the bull case (20% expected)? Very unlikely. Say 5% chance.
Bayesian update:
- Base case posterior: 50% × 40% / (50% × 40% + 20% × 20% + 30% × 5%) = 20% / 27.5% = 73% weight on base case in the evidence
- This means base case should drop from 50% to around 35-40% after seeing this evidence.
This is more rigorous than eyeballing it, though in practice, disciplined investors develop intuition for these adjustments without calculating the exact formula.
Common Updating Mistakes
Mistake 1: Confusing Temporary Headwinds with Permanent Deterioration
A cyclical company hits a temporary slowdown. The investor immediately moves probability away from the base case to the bear case. But cyclical downturns are, by definition, temporary. The correct response is to increase bear case probability slightly while maintaining base case, then revert probabilities when the cycle improves.
Example: A banking stock's loan loss provision spikes due to a credit cycle downturn. This does not mean 20% of the loan book will default permanently. Update to reflect the cycle, but don't assume permanent impairment.
Mistake 2: Over-Weighting Recent Data
If a stock beats earnings four quarters in a row, the investor shifts heavily to the bull case, assigning 60-70% probability. But four quarters is a small sample; quarterly data is noisy. The correct response is to increase bull case probability, but only to 40-50%, not 70%.
Use this rule of thumb: A string of results supporting one outcome should increase that scenario's probability, but not above 60%. Keep meaningful probability (at least 20%) on other scenarios, because you will eventually see evidence supporting them.
Mistake 3: Updating on Expected Results
A company guides 5% growth and delivers 5% growth. Investors sometimes update this as "good news," raising the base case probability. But this is the expected outcome—it provides no information. Don't update on expected results; only update when evidence differs from expectations.
Mistake 4: Asymmetric Updating (Bullish Bias)
When a stock beats, investors update aggressively to the bull case. When it misses, they rationalize it away and barely update downward. This is psychological bias, not disciplined analysis. Your probability updates should be symmetric—if you raise bull case probability by 10% on a beat, you should lower it by 10% on a miss of equal magnitude.
Mistake 5: Forgetting to Update Multiple Times
Many investors value a stock once and then check back only after a year. But new information arrives quarterly. A disciplined valuation is recomputed at least quarterly, ideally after each earnings report. This doesn't mean trading constantly; it means being aware of how fair value is changing so you can maintain appropriate position sizing.
Scenario Drift: When Bull Case Becomes Base Case
Over time, if a company consistently delivers results closer to the bull case than the base case, the bull case becomes the new base case. This is normal and expected.
Example: A company originally valued with:
- Bull: 15% growth, 25x multiple, $50 fair value (25% probability)
- Base: 10% growth, 20x multiple, $35 fair value (50% probability)
- Bear: 5% growth, 15x multiple, $20 fair value (25% probability)
After three years of delivering 13-14% growth and commanding a 22x multiple, the market reprices:
- New Bull: 18% growth, 28x multiple, $65 fair value (25% probability)
- New Base: 13% growth, 24x multiple, $48 fair value (50% probability)
- New Bear: 8% growth, 18x multiple, $28 fair value (25% probability)
The old bull case has been replaced by a more optimistic bull case. This is not error; it's the market digesting evidence that the company is stronger than originally thought.
Updating When You're Uncertain
Some information is genuinely hard to interpret. Management guidance changes slightly, but you're not sure if it's structural or temporary. Revenue growth slows, but competitors are slowing too, so is it market-share loss or sector headwinds?
In cases of uncertainty, use this principle: Update partially, then monitor for confirmation.
Example: A company's growth slows from 12% to 9%, but the CEO attributes it to macro headwinds, not competitive loss. You're uncertain whether growth will revert to 12% or settle at 9%.
- Reduce base case (12% growth) from 50% to 35%.
- Increase cautious base case (9% growth) from 20% to 40%.
- Keep bear case at 25%.
- Monitor Q2 results: if growth rebounds to 11%+, revert to higher base case. If it stays at 9%, cement the lower base case.
This approach acknowledges uncertainty while remaining disciplined. You're not over-updating on a single data point, but you're not under-updating either.
Common Mistakes
Mistake 1: Updating on Single Quarter Instead of Trend
One quarter of disappointing results doesn't mean the trend has changed. Wait for 2-3 quarters before materially shifting probabilities. Quarterly data is noisy; trends matter.
Mistake 2: Anchoring to Old Probabilities Too Strongly
You set probabilities 12 months ago. Since then, the company has consistently beaten expectations, management has raised guidance twice, and growth has accelerated. But you're still using the old probabilities because "I don't want to chase performance."
Update them. Evidence forces update.
Mistake 3: Confusing Price Movement with Fundamental Change
Stock price fell 20% this month. Does this mean base case probability has fallen? Not necessarily. Price moves are often driven by sentiment, not fundamentals. Only update probabilities when fundamentals change, not when price changes.
Mistake 4: Being Too Aggressive with Updates
You see one quarter of good results and shift base case probability from 50% to 25%, moving to bull case at 60%. This is whipsaw. Be measured. A single beat suggests slight increase in bull case (from 30% to 35%), not dramatic shift.
Mistake 5: Not Recalculating Fair Value After Updates
You update probabilities but don't recalculate fair value. This means you're not actually using the updated probabilities to inform investment decisions. Always recalculate fair value after probability updates.
Tools for Tracking Scenario Weights
Maintain a simple spreadsheet or table for each stock you own:
| Date | Bull % | Base % | Bear % | Key Driver | Fair Value |
|---|---|---|---|---|---|
| Jan 1 | 30% | 50% | 20% | Initial valuation | $41.50 |
| Apr 1 | 15% | 45% | 40% | Q1 growth miss | $32.00 |
| Jul 1 | 20% | 50% | 30% | Q2 stabilized growth | $35.50 |
| Oct 1 | 25% | 50% | 25% | Reacceleration signal | $38.00 |
This table makes your thinking explicit and lets you spot your own biases. If you find yourself always moving probability to the bull case, you're probably over-optimistic. If you're always moving to the bear case, you're probably too pessimistic.
FAQ
How often should I update scenario probabilities?
Quarterly at minimum (after earnings). More frequently if material news arrives (guidance changes, competitive disruptions, regulatory actions). Don't update on stock price moves alone—price moves are market estimates, not fundamental evidence.
Should I use the same scenario probabilities for all stocks in my portfolio?
No. Each stock has different fundamental drivers and different likelihoods. A mature utility might be 60% base case, 20% bull, 20% bear. A growth biotech might be 10% bull, 30% base, 40% bear, 20% tail risk. Tailor probabilities to each company's fundamentals.
What if I realize my original scenarios were completely wrong?
Rebuild them. If a company's fundamentals have changed so drastically that your original scenarios no longer apply, don't force-fit new probabilities to old scenarios. Write new scenarios that reflect the current reality, then assign probabilities to those.
Example: You originally valued a company assuming it would be acquired, with scenarios for acquisition price. Then the company IPOs and stays independent. The acquisition scenarios are now irrelevant; build new scenarios for standalone operation.
How do I avoid recency bias when updating?
Use time-weighted evidence. One quarter of data is much weaker than four quarters. A company that misses guidance in Q1 but then hits for three quarters should not have dramatically lower base case probability by Q4. Instead, update cautiously on Q1, then revert as subsequent quarters confirm original thesis.
Should I revalue my entire portfolio quarterly?
Yes. Revalue after earnings season. Some positions' fair values won't change (boring businesses continue boring), but others will shift 10-20%+. This revaluation lets you see which positions are becoming more or less attractive, informing trim/add decisions.
What if I update probabilities but the stock price doesn't move?
The stock price may not move immediately, but your expected return has changed. If you updated fair value down 15% but the stock price is unchanged, you're now expecting a 15% loss. This is a signal to sell, even if the market hasn't repriced it yet. You don't need to wait for price to confirm your analysis.
Related concepts
- Building Bull, Base, and Bear Cases
- Tail Risks and Black Swans
- Decision Making with Scenarios
- Sensitivity Analysis and Variable Range
- Monitoring and Portfolio Rebalancing
Summary
Scenario probabilities are living estimates that must evolve as new information arrives. A stock valued at fair value of $41.50 in January might have a fair value of $32 in April if management guidance misses materially. Disciplined investors update quarterly at minimum, carefully distinguishing between noise and signal, and rebalancing positions based on changing fair values.
The key is avoiding both over-updating (treating every quarterly variation as a major repricing) and under-updating (ignoring structural deterioration because you like the company). A simple framework—identify evidence, assess scenario likelihood, recalculate fair value—keeps updating systematic and prevents both emotional overreaction and stubborn anchoring.