A research report presents a conclusion wrapped in evidence. The two are not the same thing, and reading them as though they were is how reports mislead even when nothing in them is false.
Separate the layers
Every competent report contains four layers. Read them in this order:
- Facts. Reported figures, disclosed events, prices. These can be checked. If they cannot be, that is the first thing to note.
- Assumptions. The forecasts and the judgements behind them. Growth rates, margins, discount rates. These are where the conclusion is actually decided.
- Reasoning. How the assumptions lead to the conclusion. This is the part worth testing.
- Conclusion. A price target, a rating, an allocation. Almost the least informative layer on its own.
Most readers start at layer four. The information is in layers two and three.
Test the assumptions, not the conclusion
The useful question is not "is this conclusion right" but "what has to be true for this to work?"
Apply three checks:
- What is the single most important assumption? Usually a growth rate, a margin, or a multiple. Change it by a plausible amount and see whether the conclusion survives.
- Is the assumption sourced or asserted? "Management expects" is not evidence. "Company filings show" is.
- What is the base rate? How often do companies in this situation achieve what the report assumes? A forecast that requires an outcome that is rare in the historical record should say so.
A conclusion that survives all three is worth more than one that sounds confident.
What is usually missing
Reports are generally strong on what supports the view and thin on what would undermine it. Look for:
- The counter-case. Is there a serious version of the opposite argument, stated fairly and addressed? If not, the analysis is incomplete.
- The failure conditions. Under what circumstances would the thesis be wrong, and how would that be recognised early?
- Position sizing. A recommendation without a size is not a decision. A view held at 1% of a portfolio and the same view held at 20% are different recommendations.
- The time horizon. A thesis that is right in three years and wrong for two is a difficult position to hold.
- Costs and taxes. An expected return before these is not the return an investor receives.
Sources of bias
- Incentives. A report may be written to support a transaction, or to maintain a relationship with the company or its competitors.
- Institutional pressure to be decisive. "It depends" is often the honest answer and rarely the publishable one.
- Anchoring to consensus. Forecasts cluster around the average, which makes them similar and not necessarily correct.
- Recency. Models calibrated on a calm period understate the range of outcomes.
How to use a report
Treat a report as an argument to be examined, not a result to be applied:
- Take the facts. They are usually reliable and always useful.
- Interrogate the assumptions. They determine the answer.
- Ask what would change the author's mind. If nothing would, it is not analysis.
- Size it yourself. The report cannot know the portfolio's constraints.
- Keep the reasoning. When the position is later reviewed, the reasoning is what is being reviewed.
This is the standard Eonvera applies to its own research: evidence should be traceable, models should be challengeable, and every recommendation should come with the risks and the conditions under which it would be reassessed.