Gains:
- Understand how artificial intelligence is used as drafting and calculation control in financial statement analysis, ratio calculation and cash flow interpretation.
- Ability to recalculate the financial interpretation produced by artificial intelligence with source data and question assumptions and limits as an expert
- Ability to understand the uncertainty of forecasting and scenario analysis and maintain that investment/credit decisions belong to people.
A financial analyst's job is to read the story inside the charts. A company's balance sheet, income statement and cash flow; It tells a story about liquidity, profitability, debt and growth. From these numbers, the analyst calculates rates, interprets trends, constructs scenarios and provides input into a credit/investment decision. In this process, artificial intelligence saves time in table reading, rate calculation draft and interpretation draft. But the heart of financial analysis is uncertainty: every interpretation is based on an assumption, every forecast is based on a scenario. AI can say with overly certainty that “this company is absolutely solid”; whereas true analysis is about questioning assumptions and pointing out uncertainty. The final credit/investment decision, with its uncertainty, is human.
In this unit, we will see how to use AI as a draft and account control in financial statement analysis, how to recalculate and query its output, and how to manage forecast/scenario uncertainty.
What is financial analysis based on?
Basic tools:
- Ratio analysis: Liquidity (current ratio, acid-test), profitability (net profit margin, return on equity), indebtedness (debt/equity), efficiency (receivables turnover ratio).
- Trend analysis: Change between years/quarters; growth or contraction?
- Cash flow: Whether a seemingly profitable company generates cash—profit on paper is not money in the cash register.
- Scenario and sentiment: Assumption-based projections such as "What will happen if sales drop 10%?"
Tip: Profitability and cash flow are different things. While a company is showing an accounting profit, it may be short of cash. If the AI slips into a shortcut like “profitable, therefore solid,” it is your job to catch that trap.
Positioning artificial intelligence correctly
Quest
AI role
control/decision
Ratio calculation sketch from the table
account draft
Recalculate from source
Trend summary
Summary draft
Expert reviews
Industry comparison note
draft
Source and currency confirmation
Scenario editing
Assumption draft
Expert questions assumption
Credit/investment advice
(draft only)
man decides
AI does not make a “final decision” on any line; It's your job to uncover the assumption and uncertainty underlying every comment.
Recalculation and assumption questioning
- Verify rates from source. Calculate yourself the current ratio, margin, debt/equity ratio given by artificial intelligence with table values. The model may mix denominator/numerator.
- Input accuracy. Are the table values read by the model correct? Wrong input gives wrong results even with the correct formula (garbage in, garbage out).
- Make the assumption visible. Ask explicitly for the assumption on which each forecast is based ("sales remain constant", "rate does not change") and question its realism.
- Break excessive precision. Replace expressions such as “absolutely,” “certainly,” with language of uncertainty (“according to available data, assuming”).
Four copyable templates
1) Ratio calculation sketch (with step display):
Calculate the following ratios with the table values below and show the formula and intermediate steps for each: current ratio, net profit margin, debt/equity. Just do the math; Making a "strong/weak" judgment.Values: [current asset, short-term liability, net profit, sales, total debt, equity]
2) Trend and cash flow summary:
Your role: summary assistant to the financial analyst. Neutrally summarize trends from the verified three-quarter data I give you. Evaluate profitability and cash flow SEPARATELY; If the two contradict each other, highlight this. Making definitive investment decisions. Data: [figures]
3) Assumption and uncertainty control:
What assumptions are based on the following financial commentary? List each assumption clearly and indicate which ones would change the outcome if they were not realized. Flag statements that are overly precise ("absolutely", "guaranteed"). Text: [comment]
4) Scenario (sensitivity) draft:
Your role: scenario drafting assistant. Set up two scenarios based on base data: (1) if sales fall 10%, (2) if interest expense increases 20%, how will base rates be affected? Write down the assumption of each scenario clearly. This is a draft; I will make the final loan/investment decision.Base data: [figures]
Weak prompt / Strong prompt
Weak prompt:
Look at the statements of this company and decide whether we should give a loan, I want a clear answer: yes or no.
It offloads the decision to the model, hides assumptions, ignores uncertainty, and invites excessive precision.
Powerful prompt:
Your role: decision support assistant, not decision maker. Calculate ratios with formula, evaluate trend and cash flow separately. Write clearly the assumption on which each comment is based and state any uncertainties. Do not use expressions such as "absolutely." I will make the credit decision, stating the uncertainty.
The strong will makes the calculation transparent, makes the assumption and uncertainty visible, and leaves the decision to the human.
three mini cases
Case 1 — Time savings. An analyst has AI calculate the underlying ratios of a portfolio of 15 companies in a rough draft, then verifies them all from the source. Manual 4-hour calculation is reduced to 1 hour; The analyst devotes his time to interpretation and questioning assumptions.
Case 2 — Recalculation catches the error. Artificial intelligence gives a company's current ratio as "2.4". The analyst recalculates: the model used total debt instead of short-term debt; the real current ratio is 1.1 and there is liquidity concern. Validation prevents a false impression of "solid".
Case 3 — Excessive precision breaks. Artificial intelligence says "the company is very solid, the loan should definitely be given". The analyst looks at cash flow: although the company appears profitable, the last two quarters have produced negative operating cash flow. The comment is rewritten by adding uncertainty as "profitable but cash generation is weak, additional collateral and close monitoring are recommended." The decision is presented to the committee with this nuance.
The importance of context: same ratio, different story
In financial analysis, numbers do not speak alone; It gains meaning with context. While the same "current ratio of 1.2" value is quite normal for a retail chain, it may be alarming for a contracting company with a long collection period. AI knows general patterns, but often not the specific context of your industry, this company, and this period; so it can slip into shortcuts like “low rate = bad.” You are the one who adds context.
Ask three questions when considering context:
- What is the industry norm? Compare the ratio to the industry average and competitors; Relative position, not absolute value, is important.
- What is the trend? The rate of a single period is misleading; Is the rate improving or worsening? Direction is often more important than level.
- Are there any one-time effects? An asset sale, provision for a lawsuit, or unusual income can make the period look good or bad and hide the real picture. See the "cleansed" image.
Tip: When having AI interpret a ratio, ask "what industry assumption are you talking about?" ask. If the model cannot explain its assumption, its interpretation is without context; unreliable until tested with your industry knowledge.
Common mistakes
- Not recalculating the rate. Not confirming the ratio given by the model with the formula; This is where the numerator/denominator confusion comes in.
- Confusing profit with cash. Falling for the “profitable, therefore solid” shortcut.
- Hiding the assumption. Not asking what assumption the prediction is based on.
- Accepting excessive certainty. Expressions such as "absolutely solid" are included in the report as they are.
- Leaving the decision to the model. Trying to delegate responsibility by having the model say the final loan/investment judgment.
Caution: The most dangerous outcome in financial analysis is a confident but based on false assumption. Trusting a prediction without knowing what "if" is behind it is like trusting a building whose foundations cannot be seen.
In summary
Financial analysis reads the story behind the numbers with ratio, trend, cash flow and scenario tools. Artificial intelligence produces account drafts and commentary drafts; but it is for the expert to recalculate the ratios from the source, validate the input, make assumptions visible, and replace excessive precision with the language of uncertainty. The final credit/investment decision, with its uncertainty, rests with the human being. In one sentence: AI summarizes the numbers; The competent analyst owns the interpretation, assumption and decision.
Application task
Prepare a small financial statement (current assets, current liability, net profit, sales, total liabilities, equity). Have the 1st template calculate the ratios and compare each by recalculating them manually. Then evaluate profitability and cash flow separately with template 2. Also print a deliberately overly precise comment and show how it is softened by checking for assumption/uncertainty with the 3rd template.
checklist
- [ ] I recalculated the rates given by the model from the source using the formula.
- [ ] I confirmed that the input table values were read correctly.
- [ ] I evaluated profitability and cash flow separately.
- [ ] I have made visible and questioned the assumption on which each interpretation is based.
- [ ] I replaced overly precise statements with language of uncertainty.
- [ ] I took the final loan/investment decision and responsibility.