Maintaining Your Models
A valuation is a snapshot in time. The moment you finish, new information arrives: earnings, guidance, management changes, competitive moves, macroeconomic shifts. Rather than let your model ossify, you need a disciplined process to update it, preserve your reasoning, and know when new facts have broken your thesis.
Quick Definition
Model maintenance is the continuous process of updating your three-statement DCF and sensitivity analysis with new information, documenting what changed and why, tracking the impact on fair value, and making explicit go/no-go decisions when data contradicts your assumptions. This separates investors who adapt to new facts from those who anchor to past views.
Key Takeaways
- Establish trigger points for model updates (earnings releases, quarterly guidance, announced M&A, analyst downgrades)
- Keep a change log; each update creates a new version with documented rationale
- Distinguish between noise (small misses) and signal (directional breaks in your thesis)
- Update only the inputs that changed; keep the model structure and other assumptions stable
- Set explicit thresholds for abandoning a position (valuation down 15%+, thesis broken, better alternatives found)
Building an Update Schedule
Quarterly Update (Mandatory)
After each earnings release:
- Actual revenue, margins, FCF: plug into historical data
- Compare to your prior forecast: Did you underestimate or overestimate?
- Calculate new forecast adjustments
- Update WACC if interest rates or beta shifted materially
- Recalculate fair value
- Record all changes in version log
Time commitment: 1–2 hours per stock per quarter for a mature model.
Ad Hoc Update (Event-Driven)
When a material event occurs, update immediately:
- Major M&A announcement → revise terminal value, leverage assumptions
- CEO departure → revise management quality assessment, update margin assumptions
- Regulatory change → revise TAM, growth rate, or discount rate
- Analyst downgrade or upgrade → evaluate their case; update if grounded
- Competitor disruption → revise margin or growth assumptions downward
- Material stock move (>15% in one day) → examine trigger; update if fundamental change
Example Trigger: Your SaaS company stock drops 20% overnight on a "disappointing" earnings miss. Before updating, ask:
- Did growth slow materially (6% miss vs. 12% forecast) or is this noise?
- Did the company lose a major customer?
- Did margins compress unexpectedly?
- Or did the market simply have unrealistic expectations?
If it's noise (modest beat/miss within normal range), don't update—just document "Noise" in your log. If it's signal (structural change), update immediately.
The Update Workflow
Step 1: Identify What Changed
Create a "Delta" sheet in your model:
| Input | Previous | New | Change | Source | Material? |
|---|---|---|---|---|---|
| Revenue Growth Yr 1 | 12% | 11.5% | -0.5% | Q3 results, growth miss | No (within range) |
| Revenue Growth Yr 2 | 10% | 9.5% | -0.5% | Guidance lowered | No (slowing but expected) |
| EBIT Margin Yr 1 | 22% | 21.8% | -0.2% | Actual result | No (noise) |
| Terminal Margin | 25% | 24% | -1.0% | Intensifying competition | YES (structural change) |
| WACC | 8.2% | 8.5% | +0.3% | Risk-free rate rose 30bps | No (modest) |
Step 2: Update Only Changed Cells
Don't rebuild the entire model. Update specific input cells and let formulas cascade. Your model architecture (formulas, structure) stays the same; only inputs change.
Changed cells from above:
- B10 (Revenue Growth Yr 2): 10% → 9.5%
- B18 (Terminal Margin): 25% → 24%
- B32 (WACC): 8.2% → 8.5%
Let intrinsic value recalculate automatically.
Step 3: Analyze the Impact
Before finalizing, ask:
- How much did fair value change? (+5%? -10%? -30%?)
- Did the gap between your valuation and market price widen or narrow?
- Is the stock now more or less attractive?
Example output:
- Previous fair value: $58/share
- New fair value: $54/share (due to margin compression and WACC rise)
- Market price: $56/share
- Prior conclusion: Stock slightly undervalued
- New conclusion: Stock slightly overvalued
Step 4: Document the Change
Add a row to your change log:
| Version | Date | Updated By | Input Changes | Reason | New Fair Value | Prior Fair Value | Impact | Action |
|---|---|---|---|---|---|---|---|---|
| 1.0 | 2024-10-15 | Analyst | Yr1 Rev: 12%→11.5%, Terminal Margin: 25%→24%, WACC: 8.2%→8.5% | Q3 results + margin pressure + rates | $54 | $58 | -6.9% | HOLD (became fairly valued) |
Step 5: Decide: Hold, Update Position, or Exit
Does your conviction change? Use this framework:
| Fair Value vs. Market Price | Prior Thesis | New Thesis | Action |
|---|---|---|---|
| +20% undervalued | BUY | BUY (stronger) | HOLD / ADD |
| +15% undervalued | BUY | BUY (weaker) | HOLD (monitor) |
| +5% undervalued | BUY | NEUTRAL | REDUCE |
| Fairly valued (±5%) | BUY | HOLD | HOLD or REDUCE |
| -5% overvalued | BUY | SELL | REDUCE / EXIT |
| -15% overvalued | BUY | SELL (strong) | EXIT |
Diagram: Update Decision Tree
Handling Earnings Misses vs. Guidance Changes
Earnings Miss (Company Beat/Missed Quarterly Results)
A company guides to $1.00 EPS; reports $0.95. How much should you update?
- Check beat/miss magnitude: 5% miss is noise; 15% miss is signal
- Look at components: Did growth miss (structural problem) or margins miss (one-time item)?
- Review management commentary: Are they confident? Do they maintain full-year guidance?
Decision logic:
- Small miss (2–5%), maintained guidance → Don't update growth; watch next quarter
- Moderate miss (5–10%), guidance cut → Update growth downward 50% of miss; recheck peers
- Large miss (15%+), guidance slashed → Major update; revisit thesis entirely
Example: Your SaaS model forecasts 25% revenue growth. Company reports Q3: 22% growth (3% miss). Maintains full-year growth guidance (25%).
- Conclusion: Minor miss, likely noise. Don't change full-year forecast.
- Action: Note the quarterly miss; monitor for pattern. If next quarter also misses, update then.
Guidance Change (Management Raises or Cuts Outlook)
Guidance is a direct signal from management. Weight heavily.
Example: Your model forecasts 15% growth. Company reports Q3: 14% growth (not great, but close). But cuts full-year guidance to 12%.
- Conclusion: Management sees headwinds; growth will slow. Trust management.
- Action: Update growth assumption from 15% to 12–13% (between old forecast and management guidance).
Version Control and Rollback
Maintain backups of all versions. If you later realize an update was wrong, you can roll back.
File naming convention:
Ticker_ValuationModel_v1.0_2024-10-15.xlsx
Ticker_ValuationModel_v1.1_2024-10-22.xlsx [Q3 earnings update]
Ticker_ValuationModel_v1.2_2024-11-05.xlsx [Margin compression]
Ticker_ValuationModel_v1.3_2024-11-15.xlsx [WACC adjustment]
Each file contains:
- That version's Assumptions sheet
- That version's change log (up to that date)
- Valuation result
If v1.2 turns out to be wrong (new data overturns the margin assumption), you can revert to v1.1 and apply a different update.
Quarterly Maintenance Checklist
After each earnings release, run through:
- Updated historical financials (revenue, EBIT, FCF from Q results)
- Updated guidance interpretation (management's view of next year)
- Sensitivity analysis: Rerun with new base case inputs
- Backtest logic: Did your forecasts match reality? How far off? Why?
- Valuation updated: Calculate new intrinsic value, compare to old
- Market price noted: Current stock price vs. new fair value
- Change log entry: What changed, why, what impact
- Decision recorded: Hold, add, reduce, or exit?
- Version file saved: Backup with today's date
Knowing When to Kill a Thesis
Set explicit exit rules in advance. Don't wait for emotions to decide.
Example Exit Rules for a Long Position:
- Valuation Rule: If fair value falls below 80% of entry price, review position (may indicate thesis is broken)
- Growth Rule: If revenue growth trends below 50% of forecast for 2 consecutive quarters, update growth assumption and recalculate
- Margin Rule: If operating margin falls below historical low by 300+ bps, mark thesis as "under pressure"; if it falls below competitive peer average, exit
- Competitive Rule: If a new competitor launches that materially reduces TAM or pricing power, reduce position by 50% immediately
- Macro Rule: If interest rates rise above WACC by 200+ bps, remodel with higher discount rate; if new fair value is 20%+ lower, review position
- Time Rule: If you've held 5 years and valuation hasn't moved while market returned 10% annualized, the stock is slower than index; look for better ideas
Example Exit Rules for a Short Position:
- Thesis Broken: If you shorted on "margin compression" but margins expand unexpectedly and stay high for 3 consecutive quarters, cover
- Better Alternatives: If you're down 20% (great short) but found a better short elsewhere, take profit and redeploy
- Valuation Extreme: If the stock trades at 0.5x book (vs. 1.0x average) despite business working, cover (risk/reward too ugly)
FAQ
Q: How often should I update my model? A: Quarterly minimum (around earnings). More frequent if event-driven (guidance change, analyst call, competitive news).
Q: Should I update for minor misses (1–3%)? A: Document but don't update forecast. If small misses accumulate into a trend (6 quarters of 2–3% misses = 12–18% cumulative), then update.
Q: What if earnings surprise is positive but valuation falls (market doesn't care)? A: Valuation didn't change; market perception did. Don't update your model just because the market is irrational. Understand why the market is ignoring the good news (is it pricing in reversal? Sector headwind?). Update only if facts justify.
Q: How much should valuation change from a single quarter's update? A: Rarely more than 5–10% if it's a noise quarter. 15%+ changes usually signal material thesis shift (growth acceleration, margin compression, new competitor). Question the update to make sure it's not overreacting.
Q: Can I update my model too frequently? A: Yes. Constant tweaking chases noise. Establish a calendar (monthly review for noise, quarterly deep update after earnings, annual full remodel). Stick to it.
Q: What if I realize my original assumptions were too aggressive? A: Update. Don't be anchored to past views. A disciplined analyst updates when facts contradict assumptions; an undisciplined one defends past guesses.
Related Concepts
- Assumption Drift: Gradual shift away from original assumptions without explicit decision
- Threshold Rebalancing: Systematic rules for when to act on valuation changes
- Thesis Monitoring: Ongoing tracking of whether original investment case remains intact
- Information Cascade: How new information updates valuations over time
- Belief Updating: Bayesian framework for incorporating new data into prior views
Summary
Model maintenance separates long-term investors from one-time analysts. Markets change. Companies evolve. Assumptions that were reasonable become stale. By updating methodically, documenting religiously, and exiting when facts break your thesis, you adapt without capitulating to noise.
The goal isn't to perfectly predict the future (impossible) but to hold a current, defensible valuation that reflects all available information. As new facts arrive, update. As your fair value converges toward market price, question whether your thesis is intact or whether the market is simply re-rating. As conviction weakens, exit gracefully before it disappears entirely.
A model you update quarterly is 100 times more valuable than one you build once and never touch.
Next Steps
You've built, documented, backtested, and maintained your models. Time to synthesize it all: Summary: From Blank Sheet to Conviction shows you how to thread these tools into a unified process that turns spreadsheet rigor into genuine investment conviction.