Evidence and Conviction

Investors often talk about conviction.

They say:

I have high conviction in this company.

But where should conviction come from?

Not from:

• enthusiasm,
• familiarity,
• a rising stock price,
• management charisma,
• social-media popularity,
• or the amount of money already invested.

Conviction should come from evidence.

A disciplined investor should be able to answer:

What evidence supports this thesis, how strong is that evidence, and what evidence would cause my conviction to change?

What Is Conviction?

Conviction is the degree of confidence an investor has in an investment thesis.

It is not certainty.

The future remains uncertain regardless of how much research has been completed.

A useful definition is:

Conviction is confidence proportional to the strength, consistency, relevance, and durability of the evidence supporting the thesis.

That definition matters because conviction should be earned.

Conviction Is Not Emotion

An investor can feel extremely confident and still be wrong.

Emotional confidence may come from:

• recent profits,
• repeated exposure to the same narrative,
• agreement from other investors,
• or attachment to the company.

None of these necessarily improve the evidence.

Confidence and Evidence Can Diverge

Imagine two investors.

Investor A

Has spent hundreds of hours discussing a company online and feels extremely confident.

Investor B

Has studied:

• ten years of financial statements,
• customer retention,
• competitor economics,
• management capital allocation,
• debt,
• and valuation.

Investor B may have stronger analytical grounds for conviction even if Investor A feels more certain.

Evidence Should Lead Conviction

The correct relationship is:

Evidence → Interpretation → Thesis → Conviction

A dangerous relationship is:

Conviction → Search for Supporting Evidence

The second approach encourages confirmation bias.

Evidence Is Not All Equal

Different evidence deserves different weight.

Consider these statements:

Management says demand is extremely strong.

and:

Customer orders have increased for twelve consecutive quarters while retention remains above historical averages.

Both are evidence.

They are not equally strong.

Direct Evidence

Direct evidence is closely connected to the economic claim being tested.

Examples include:

• actual revenue,
• customer retention,
• cash flow,
• debt,
• margins,
• returns on capital,
• share count,
• and pricing behavior.

Direct evidence often deserves substantial weight.

Indirect Evidence

Indirect evidence may provide useful clues without proving the conclusion.

Examples might include:

• web traffic,
• app rankings,
• job postings,
• customer surveys,
• or supplier commentary.

These can strengthen research when interpreted carefully.

Narrative Evidence

Narrative evidence includes:

• management statements,
• analyst opinions,
• media commentary,
• and investor narratives.

Narratives can identify hypotheses.

They should not automatically be treated as proof.

Management Claims

Management usually knows more about the company than outside investors.

Its communication can therefore be valuable.

But management also has incentives.

Investors should compare claims with outcomes.

If management repeatedly promises:

• margin improvement,
• disciplined acquisitions,
• or debt reduction,

the evidence should eventually appear in results.

Track Record Converts Claims Into Evidence

A management team that consistently:

• sets realistic expectations,
• executes,
• acknowledges mistakes,
• and allocates capital rationally

earns greater credibility.

Trust should emerge from demonstrated behavior.

Independent Evidence

Evidence becomes stronger when multiple independent sources support the same conclusion.

Suppose management says customer loyalty is exceptional.

That claim becomes more credible if:

• retention data are strong,
• customers describe high switching costs,
• competitors struggle to displace the product,
• and pricing increases do not cause major churn.

The evidence converges.

Converging Evidence

Convergence occurs when different observations point toward the same economic conclusion.

For example:

Moat Thesis

Evidence might include:

• high retention,
• stable market share,
• pricing power,
• high ROIC,
• and competitors failing to achieve similar economics.

No single metric proves the moat.

Together they create a stronger case.

Contradictory Evidence

Good analysis must also consider evidence that points in the opposite direction.

Suppose:

• retention remains strong,
• but customer acquisition cost rises,
• competitors gain share,
• and pricing becomes more difficult.

The moat conclusion becomes less certain.

Conviction should reflect the conflict.

Evidence Is Often Mixed

Real businesses rarely produce perfectly consistent evidence.

A company may have:

• strong growth,
• weakening margins,
• improving cash flow,
• rising competition,
• and falling debt

at the same time.

The investor must weigh the evidence rather than search for a perfectly clean story.

Evidence Quality

Evidence quality depends on several characteristics.

Useful questions include:

• Is the source reliable?
• Is the evidence direct or indirect?
• Is it current?
• Is it independently verifiable?
• Is it economically relevant?
• Is it representative?
• Does it cover enough time?

High-quality evidence deserves greater weight.

Evidence Relevance

A fact can be accurate but irrelevant.

Suppose an investor wants to determine whether a company has pricing power.

The number of followers on its social-media account may be real.

But it may provide little evidence about customers' willingness to accept higher prices.

Evidence should match the question.

Evidence Recency

Older evidence can remain useful.

But the investor should ask whether conditions have changed.

A company may have possessed strong pricing power ten years ago.

That does not prove it still does.

Current evidence matters.

Historical Depth

At the same time, recent evidence alone can be misleading.

One excellent year may result from:

• favorable economic conditions,
• temporary shortages,
• or unusually strong demand.

Long historical evidence helps reveal durability.

Evidence Across Cycles

Some of the strongest business evidence appears during difficult periods.

A company that preserves:

• customers,
• margins,
• liquidity,
• and competitive position

during recession provides evidence of resilience.

Good times often hide weaknesses.

Adversity Tests the Thesis

Suppose two companies appear equally strong during expansion.

Then recession arrives.

Company A:

• remains profitable,
• gains share,
• and continues investing.

Company B:

• loses customers,
• breaches debt covenants,
• and issues shares.

The downturn reveals information that prosperity concealed.

Sample Size

Evidence based on a small sample should usually receive less confidence.

Suppose a new product has:

100 customers

and retention is excellent.

That is encouraging.

It may not yet justify the same conviction as retention measured across:

100,000 customers

over many years.

Base Rates

Investors should consider what normally happens in similar situations.

Suppose a company is attempting a difficult turnaround.

The thesis should not rely only on:

Management believes the turnaround will succeed.

The investor should also consider:

• historical turnaround success,
• industry structure,
• financial resources,
• and comparable situations.

Base rates provide context.

Company-Specific Evidence Still Matters

Base rates should not replace company analysis.

A company may possess unusual advantages.

The correct question is:

What evidence justifies believing this company may differ from the base rate?

That requires specific evidence.

Quantitative Evidence

Quantitative evidence includes measurable data such as:

• revenue,
• margins,
• ROIC,
• retention,
• debt,
• free cash flow,
• market share,
• and share count.

Numbers can improve precision.

They still require interpretation.

Qualitative Evidence

Qualitative evidence includes:

• customer behavior,
• culture,
• management integrity,
• product quality,
• competitive dynamics,
• and strategic positioning.

Important investment questions cannot always be reduced to one number.

Quantitative and Qualitative Evidence Should Connect

Suppose qualitative research suggests strong customer loyalty.

Quantitative evidence should ideally show something consistent, such as:

• high retention,
• recurring revenue,
• pricing power,
• or stable market share.

When qualitative and quantitative evidence reinforce each other, conviction can become stronger.

Numbers Can Mislead

A precise number is not automatically high-quality evidence.

For example:

Total addressable market = $137.4 billion

may look scientific.

But the estimate may depend on questionable assumptions.

Precision and reliability are different.

Adjusted Metrics

Companies often present adjusted measures.

These may help explain underlying economics.

They can also exclude recurring costs.

Investors should compare adjusted measures with:

• GAAP results,
• cash flow,
• dilution,
• and long-term economic reality.

Evidence and Accounting

Financial statements are evidence.

But accounting numbers should be understood rather than accepted mechanically.

For example:

• earnings may rise while cash flow weakens,
• free cash flow may rise because investment was postponed,
• or EPS may rise because shares were repurchased.

Interpretation matters.

Evidence and Causation

Correlation does not automatically prove causation.

Suppose revenue rises after advertising spending increases.

That does not prove advertising caused all the growth.

Other factors may include:

• market expansion,
• pricing,
• product improvement,
• or competitor weakness.

The thesis should avoid claiming more than the evidence supports.

Evidence and Alternative Explanations

For important conclusions, ask:

What else could explain this evidence?

Suppose margins improve.

Possible explanations include:

• genuine scale advantage,
• temporary input-cost decline,
• reduced investment,
• accounting changes,
• or favorable product mix.

Alternative explanations should be tested.

Evidence Hierarchy for a Thesis

A practical evidence hierarchy might be:

Level 1 — Demonstrated Economic Evidence

Observed business outcomes directly related to the thesis.

Level 2 — Supporting Operational Evidence

Customer, product, competitive, or industry evidence supporting those economics.

Level 3 — Management and External Interpretation

Guidance, forecasts, analyst opinions, and narratives.

The lower levels can be useful.

They should not substitute for demonstrated economics when stronger evidence is available.

Evidence Confidence

An investor can classify evidence confidence as:

• High
• Moderate
• Low
• Unavailable

The purpose is not to create artificial precision.

It is to make uncertainty explicit.

High Confidence

High confidence may be appropriate when evidence is:

• direct,
• consistent,
• independently supported,
• historically durable,
• and economically relevant.

Moderate Confidence

Moderate confidence may apply when evidence is:

• generally supportive,
• but incomplete,
• relatively recent,
• or somewhat conflicting.

Low Confidence

Low confidence may apply when the conclusion depends heavily on:

• forecasts,
• limited history,
• indirect evidence,
• or uncertain assumptions.

Unavailable Evidence

Sometimes the necessary evidence does not exist.

That should be stated.

The absence of evidence should not be replaced by confidence.

Conviction Should Be Component-Based

Instead of saying:

I have 90% conviction in this stock

an investor can ask:

• How confident am I in Business Quality?
• How confident am I in the Moat?
• How confident am I in Management?
• How confident am I in Growth?
• How confident am I in Financial Strength?
• How confident am I in Valuation?

This produces a more useful picture.

Different Components Can Have Different Confidence

An investor may conclude:

• Business Quality: High confidence
• Financial Strength: High confidence
• Moat: Moderate confidence
• Management: Moderate confidence
• Growth: Low confidence
• Valuation: Moderate confidence

That is more informative than one overall conviction number.

Weakest-Link Analysis

Sometimes the most uncertain component deserves the most research.

Suppose almost everything about the thesis is well supported except the growth runway.

The next research question should probably focus on growth.

Research should reduce important uncertainty.

Conviction Should Not Exceed Evidence

A useful discipline is:

Never allow conviction to become stronger than the evidence supporting it.

This sounds obvious.

In practice, investors frequently become more confident because:

• the stock rises,
• others agree,
• or the story becomes popular.

Those are not substitutes for evidence.

Price Movement Should Not Create Conviction

One of the easiest mistakes in investing is allowing price movement to influence confidence in the thesis.

Suppose you buy a stock at:

$70

and it rises to:

$100

You may feel smarter.

But what new business evidence appeared?

If nothing changed except price, the thesis did not become stronger.

The valuation may actually have become less attractive.

Falling Prices Should Not Automatically Destroy Conviction

The reverse is also true.

Suppose the stock falls from:

$70 to $45

while:

• customer retention remains strong,
• free cash flow grows,
• debt falls,
• the moat remains intact,
• and intrinsic value is unchanged.

The lower stock price does not necessarily weaken the thesis.

It may improve the margin of safety.

Separate Price Evidence From Business Evidence

Price is important because it affects valuation.

But it should be placed in the correct analytical category.

A price change can tell us:

The market's valuation changed.

It does not necessarily tell us:

The company's economics changed.

Keeping these separate improves discipline.

The Market Can Contain Information

This does not mean price movement should always be ignored.

A sudden major decline may indicate that other investors have learned something important.

The correct response is not:

The market must be wrong.

It is:

What changed, and is there new evidence I have not considered?

Investigate the cause.

Conviction Should Be Updated, Not Defended

An investment thesis is a working hypothesis.

New evidence should be allowed to change it.

If evidence improves, conviction may rise.

If evidence deteriorates, conviction should fall.

If evidence becomes contradictory, uncertainty should increase.

The goal is not to preserve the original conviction level.

The goal is to remain aligned with reality.

Bayesian Thinking

A useful mental model is Bayesian updating.

You begin with a belief based on existing evidence.

New evidence arrives.

You update the belief.

You do not restart the analysis from zero.

You also do not ignore the new evidence.

A Simple Example

Suppose your initial thesis is that a company has strong switching costs.

Evidence includes:

• high retention,
• stable pricing,
• and long customer relationships.

Conviction is reasonably high.

Then a competitor launches a new product.

One quarter later, retention remains unchanged.

That slightly supports the thesis.

Several quarters later, retention begins falling and customers report easier migration.

Now the evidence is materially different.

Conviction should decline.

Evidence Accumulates

One data point rarely deserves enormous weight.

But repeated evidence can accumulate.

Suppose retention changes:

• 95%
• 94%
• 91%
• 88%
• 85%

while competitor share rises.

The pattern becomes more important than any individual observation.

Evidence Can Reverse Earlier Conclusions

Good investors permit new evidence to overturn previous beliefs.

Perhaps the original moat thesis was reasonable.

Later evidence shows the moat is weakening.

Changing your mind is not analytical failure.

Refusing to change when the evidence changes is.

Conviction and Thesis Status

Conviction can be connected to thesis status.

For example:

Strengthening

Evidence increasingly supports important thesis components.

Intact

Evidence remains broadly consistent with the thesis.

Watch

Contradictory evidence has appeared.

Weakening

Important assumptions have lost support.

Broken

Critical assumptions are no longer credible.

Conviction should generally move with these states.

Conviction Should Not Move Too Quickly

The opposite problem also exists.

Investors can overreact to every new piece of information.

One weak quarter should not necessarily destroy a thesis supported by ten years of evidence.

The weight assigned to new information should reflect:

• relevance,
• reliability,
• magnitude,
• and durability.

Durable Evidence Deserves More Weight

Suppose a company has maintained high ROIC for:

15 years

One weak year occurs during recession.

That year matters.

But it should be interpreted within the longer record.

A durable pattern should not be discarded casually.

New Structural Evidence Can Deserve More Weight

Long history is not always dominant.

Suppose a new technology permanently changes the industry.

Fifteen years of historical economics may become less relevant.

Investors must determine whether new evidence represents:

• temporary variation,
• or structural change.

Evidence Half-Life

Some evidence remains useful for a long time.

Other evidence becomes stale quickly.

For example:

Long Half-Life

• capital-allocation philosophy,
• industry structure,
• long-term customer relationships.

Shorter Half-Life

• inventory levels,
• quarterly orders,
• temporary commodity costs,
• near-term demand.

The monitoring frequency should match the evidence.

Confidence in the Business vs. Confidence in the Forecast

An investor may have high confidence that:

This is an excellent business

while having low confidence about:

Next year's revenue growth.

These are different judgments.

Long-term business quality can sometimes be easier to assess than short-term forecasting.

Confidence in Value

Valuation confidence may also differ.

A mature, stable company may support a relatively narrow value range.

An early-stage company with uncertain future economics may require a much wider range.

Conviction in the company should not automatically become conviction in one precise valuation.

Evidence and Valuation Range

Stronger evidence can sometimes justify a narrower valuation range.

Weaker evidence should usually produce:

• wider scenarios,
• wider valuation ranges,
• or greater margin-of-safety requirements.

Uncertainty should appear somewhere in the analysis.

Margin of Safety and Evidence

Suppose two companies both appear worth:

$100

Company A has highly predictable cash flows and strong evidence.

Company B depends on uncertain future adoption.

Paying:

$90

for both may not offer equal protection.

Company B may require a much larger discount because the value estimate is less reliable.

Evidence Risk

Evidence risk is the possibility that the information supporting the thesis is incomplete, unreliable, stale, or misinterpreted.

This deserves explicit attention.

An investor may understand the available evidence perfectly and still have too little evidence to justify strong conviction.

Missing Evidence

Suppose you cannot determine:

• customer retention,
• unit economics,
• debt maturity,
• or acquisition performance.

Do not silently assume favorable answers.

Mark the evidence as unavailable.

Missing evidence should reduce confidence when the missing information is important.

Unknown Is Not Neutral

Investors sometimes treat missing information as though it has no effect.

But if the unknown variable is central to the thesis, uncertainty itself matters.

For example:

We do not know whether the company's largest customer will renew a contract representing 40% of revenue.

That is not a minor omission.

Evidence Contradiction Register

A useful research practice is to maintain a list of evidence that contradicts the thesis.

For example:

Supporting

• retention remains high,
• ROIC remains strong,
• debt is declining.

Contradicting

• customer acquisition cost is rising,
• competitor share is improving,
• pricing increases have slowed.

This prevents negative evidence from disappearing inside a positive narrative.

Weight Evidence by Importance

Not every contradictory fact deserves the same weight.

A small increase in administrative expense may matter little.

A major decline in retention may matter enormously.

Evidence should be weighted according to its relationship with the thesis.

Primary Evidence Drivers

Identify the few pieces of evidence that carry the most thesis weight.

For example:

1. Customer retention
2. Incremental ROIC
3. Pricing power
4. Net debt
5. Reinvestment runway

If these remain strong, the thesis may remain intact despite secondary noise.

Conviction and Position Size

Investors sometimes connect conviction with position size.

But conviction alone should not determine concentration.

Position decisions should also consider:

• permanent-loss risk,
• valuation,
• uncertainty,
• correlation,
• liquidity,
• and personal financial circumstances.

High conviction does not eliminate risk.

High Conviction Can Be Dangerous

The phrase:

My highest-conviction idea

can create psychological danger.

The investor may become:

• less skeptical,
• more concentrated,
• more willing to use leverage,
• or more resistant to contradictory evidence.

The stronger the conviction, the more important disciplined falsification becomes.

Humility and Conviction Can Coexist

An investor can say:

The evidence strongly supports this thesis

while also saying:

I may still be wrong.

These statements are compatible.

Intellectual humility does not require having no convictions.

It requires recognizing uncertainty.

Circle of Competence

Evidence is only useful if the investor can interpret it.

A highly technical business may publish extensive information.

If the investor cannot understand the economics well enough to judge it, confidence should remain limited.

Knowing the boundary of understanding is part of evidence discipline.

Complexity Can Reduce Confidence

More information does not always create more understanding.

A business with:

• dozens of opaque segments,
• complicated financing,
• aggressive adjustments,
• and difficult accounting

may provide thousands of data points while remaining difficult to value.

Complexity itself can widen uncertainty.

Simplicity Can Strengthen Evidence

A company with:

• understandable economics,
• transparent reporting,
• stable customers,
• and a long history

may support stronger conviction because the evidence is easier to interpret.

This does not make simple businesses automatically good investments.

It improves analytical visibility.

Conviction Should Be Revisable

A healthy conviction statement might be:

Current evidence supports a high-quality business with a durable moat, but growth-runway confidence remains moderate and valuation confidence is low.

That statement leaves room for revision.

It is better than:

I know this stock is going higher.

Record Why Conviction Changed

When conviction changes, record the reason.

For example:

Moat confidence moved from Moderate to High because retention remained above 95% through a recession while pricing increased and competitor share did not improve.

This creates an evidence trail.

Avoid Conviction Drift

Conviction drift occurs when confidence gradually changes without an identifiable evidentiary reason.

The investor may simply become:

• more familiar,
• more emotionally attached,
• or influenced by market sentiment.

Periodic written reviews can reveal this.

Evidence Ledger

An evidence ledger can record:

• evidence item,
• source,
• date,
• thesis component,
• direction,
• importance,
• and confidence.

Direction might be:

• Supporting
• Contradicting
• Neutral
• Uncertain

This makes evidence accumulation visible.

Evidence Aging

Evidence should also have a date.

Some evidence may need refreshing.

For example:

Customer retention: High confidence — last verified three years ago

may no longer deserve High confidence.

Evidence can age.

Evidence Provenance

The investor should know where evidence came from.

Examples include:

• audited financial statement,
• regulatory filing,
• company disclosure,
• customer research,
• industry source,
• or analyst estimate.

Provenance helps evaluate reliability.

Separate Observation From Interpretation

An evidence ledger should distinguish:

Observation

Gross margin fell from 62% to 58%.

Interpretation

Competitive pressure may be increasing.

The observation is factual.

The interpretation is a hypothesis.

Keeping them separate reduces overstatement.

Separate Interpretation From Conclusion

The next step may be:

Conclusion

Moat confidence moved from High to Moderate.

Now the reasoning chain is visible:

Observation → Interpretation → Conclusion

This is stronger than jumping directly from data to conviction.

Common Mistakes

Treating conviction as personality

Confidence should come from evidence, not temperament.

Becoming more confident because the stock rises

Price movement is not proof of business quality.

Losing conviction solely because the stock falls

Investigate whether business evidence changed.

Giving all evidence equal weight

Relevance and reliability differ.

Ignoring contradictory evidence

A thesis should be tested, not protected.

Treating management guidance as fact

Claims should be compared with outcomes.

Confusing precision with evidence quality

A precise estimate can still be unreliable.

Allowing conviction to exceed understanding

Complex or opaque businesses deserve appropriate uncertainty.

Failing to update

Conviction should change when important evidence changes.

Practical Exercise

Choose one company and create an evidence ledger.

For each of these thesis components:

1. Business Quality
2. Financial Strength
3. Moat
4. Management
5. Growth
6. Valuation
7. Risk

record:

• three supporting evidence items,
• up to three contradictory evidence items,
• source,
• date,
• relevance,
• and confidence.

Then classify confidence in each component as:

• High
• Moderate
• Low
• Unavailable

Next answer:

Strongest Evidence

What single piece of evidence most strongly supports the thesis?

Weakest Evidence

Which important thesis component has the least reliable support?

Contradictory Evidence

What fact makes you most uncomfortable with the thesis?

Missing Evidence

What important information do you wish you had?

Research Priority

Which unanswered question would most change your conviction if resolved?

Finally ask:

Has my conviction changed because the evidence changed, or because the stock price and my emotions changed?

The Buffett Perspective

Conviction should come from understanding.

An investor who understands:

• how the business makes money,
• why customers stay,
• how capital is allocated,
• what financial risks exist,
• and what the business is reasonably worth

has a stronger foundation for patience during market volatility.

But understanding should never become certainty.

The investor should remain willing to change the conclusion when the facts change.

Market price can offer opportunity.

It should not dictate belief.

The RW Finance Perspective

RW Finance should make evidence and conviction explicitly connected.

A conclusion should not appear without showing what supports it.

Each major research dimension should be able to display:

• supporting evidence,
• contradictory evidence,
• evidence provenance,
• evidence age,
• evidence confidence,
• and unresolved uncertainty.

This applies to:

• Quality,
• Financial Strength,
• Moat,
• Management,
• Growth,
• Valuation,
• and Risk.

The investment thesis can then synthesize those dimensions without hiding uncertainty.

RW Finance should distinguish:

Evidence strength

from:

Thesis conviction

and both from:

Market price movement.

The Research Journal can preserve changes over time so the investor can see:

• what evidence existed,
• how it was interpreted,
• what conviction level followed,
• and why that conviction later changed.

The system should encourage the question:

What new evidence justifies changing this conclusion?

That keeps conviction accountable to reality.

Key Takeaways

• Conviction should be proportional to the strength, relevance, consistency, and durability of evidence.
• Conviction is not certainty and should not come from emotion, popularity, or familiarity.
• Price movement affects valuation but does not automatically strengthen or weaken the business thesis.
• Direct economic evidence generally deserves more weight than unsupported narrative claims.
• Independent and converging evidence can strengthen confidence.
• Contradictory evidence should be recorded and investigated rather than ignored.
• Quantitative and qualitative evidence are strongest when they reinforce each other.
• Missing or stale evidence should reduce confidence when it concerns an important thesis assumption.
• Different thesis components can deserve different confidence levels.
• Conviction should be updated when evidence changes but should not overreact to ordinary noise.
• Evidence provenance, age, relevance, and interpretation should remain visible.
• Strong conviction and intellectual humility can coexist when the investor remains willing to revise the thesis as facts change.