Compare R&D Expense With Revenue and Gross-Profit Trends Using Claude and FMP

By Pranjal Saxena·Oct 21, 2026·21 min read

Data in Action

When a company increases research and development expense, you need to know how that increase compares with the rest of the business. Are revenue and gross profit rising too? Is R&D taking a larger share of revenue? Those questions help you identify years that deserve closer review before drawing conclusions about research spending.

This workflow uses Claude and Financial Modeling Prep's MCP server to turn five annual income statements into a repeatable R&D direction and intensity screen. The first prompt retrieves and checks the financial inputs. The second calculates growth rates and R&D ratios, labels each year's direction, and records whether revenue and gross profit recovered in the following year. You get a set of tables you can use to compare periods consistently and prioritize questions for your company research.

Apple Inc. (AAPL) provides the worked example. From FY2021 to FY2025, reported R&D expense rose from $21.9 billion to $34.6 billion. R&D expense, revenue and gross profit increased together in three of the four annual comparisons, producing an overall Aligned classification. Yet R&D also rose from 6.0% to 8.3% of revenue. The business grew, while R&D expense grew faster than revenue.

Reading those findings together is the point of the screen. An Aligned label describes direction; it does not establish R&D efficiency, innovation productivity or a causal relationship between research expense and subsequent sales.

Key takeaways

Apple's R&D expense, revenue and gross profit all rose in FY2022, FY2024 and FY2025. Three of four comparisons are Aligned.
FY2023 is Weak under the screening rules: R&D expense rose 14.0%, while revenue fell 2.8% and gross profit fell 1.0%.
R&D expense as a share of revenue increased in every comparison, including the three Aligned years.
FY2023 contained 53 weeks, while the other years contained 52. The annual labels survive an average-week sensitivity check, but unequal year lengths still limit comparability.
Revenue and gross profit recovered in FY2024 after the Weak FY2023 result. That sequence does not measure how long Apple's R&D took to produce results.

What the R&D screen measures

The screen answers one question per year: did reported R&D expense, revenue and gross profit move in the same direction? It adds R&D intensity ratios to show whether expense grew faster or slower than either output measure. These two views serve different purposes. Direction helps you identify a divergence; intensity reveals a change that an Aligned label can conceal.

Apple reports R&D separately on its consolidated statements of operations under US GAAP. That makes the line suitable for a consistent five-year comparison. The window also includes a year of declining revenue and a 53-week fiscal year, so the example shows how the screen handles both a divergence and a period-length difference.

The input is accounting expense, rather than research cash outflow or total investment in innovation. Innovation-related costs can also appear in capital spending, acquisitions or other expense lines. Operating income is included for context, but it is already calculated after R&D expense and cannot independently confirm what that expense achieved.

Measures and calculation rules

Measure	Calculation	How to use it
R&D share of revenue	R&D expense ÷ revenue × 100	Tracks R&D expense relative to sales
R&D share of gross profit	R&D expense ÷ gross profit × 100	Tracks R&D expense relative to gross profit
Change in intensity	Current ratio minus prior ratio, in percentage points	A positive change means R&D expense grew faster than the denominator
Year-over-year growth	(Current value ÷ prior value − 1) × 100	Determines the direction of R&D expense, revenue and gross profit
Operating-income growth	Same growth formula	Context only; excluded from direction labels

Calculate from unrounded inputs, apply the rules to unrounded results, then round percentages and percentage-point changes to one decimal place for display. Percentage points measure the difference between two percentages, not their relative growth rate.

A percentage on a negative base can reverse the apparent direction. For example, a loss narrowing from $10 million to $5 million produces −50% under the usual formula. The screen therefore does not calculate growth percentages on zero or negative bases.

Condition	Treatment
Prior R&D expense, revenue or gross profit is zero or negative	Show Not calculable: nonpositive base; label the comparison Data issue
A required value is unavailable	Show Not calculable: unavailable; never replace it with zero
Prior operating income is zero or negative	Show its dollar change and Up or Down direction; do not use it to set a label
Revenue or gross profit is nonpositive in an intensity ratio	Leave the ratio uncalculated and flag it for review
Growth is exactly zero	Treat it as not growing

Annual direction labels

Apply these tests in order to each year after the baseline:

Label	Condition	Interpretation
Data issue	A required growth rate cannot be calculated	The comparison cannot be assessed
Not evaluated	R&D expense growth is zero or negative	There is no R&D expense increase to compare
Aligned	R&D expense, revenue and gross profit all grow	All three rose; relative pace can differ
Mixed	R&D expense grows; exactly one of revenue or gross profit grows	The output measures diverge
Weak	R&D expense grows; revenue and gross profit are flat or lower	Higher R&D expense coincides with no growth in either output measure

These are illustrative screening labels, not industry benchmarks or assessments of management performance. A Weak year identifies a research question, not a finding that the research program failed.

Following the next year's direction

For each comparison year, reuse the following year's revenue and gross-profit growth. Label the next-year observation Positive if both grow, Partial if only one grows, and None if neither grows. Use Outside window when the following year was not retrieved, and Not calculable when its required data is invalid.

A Mixed or Weak year followed by a Positive observation counts toward the Subsequent recovery rule. This identifies a sequence in the financial statements. It does not attribute the recovery to earlier R&D or estimate a product-development lag.

Overall classification

Order	Test	Result
1	Any comparison has a Data issue, or an input row fails validation	Difficult to Interpret: data issue; stop
2	At least three of four comparisons are Aligned	Aligned
3	At least one Mixed or Weak year is followed by a Positive observation, with no persistent deterioration	Subsequent recovery
4	At least two comparisons are Weak, with no qualifying subsequent recovery	Weak
5	No earlier rule applies	Difficult to Interpret: inconclusive evidence

Persistent deterioration means two or more consecutive years in which both revenue and gross-profit growth are zero or negative. Decision order matters: an overall Aligned result takes precedence over a subsequent recovery observation. A five-year window with two or more Not evaluated comparisons falls into the inconclusive category under these rules.

Accounting for 52-week and 53-week years

Apple's fiscal year ends on the last Saturday of September. FY2023 contained 53 weeks; FY2021, FY2022, FY2024 and FY2025 each contained 52. The extra week was in the first fiscal quarter, which includes the holiday season. The FY2023 and FY2024 growth rates therefore span unequal periods.

Keep the reported values unchanged, then apply an average-week sensitivity check. A 53-week year is approximately 1.9231% longer than a 52-week year. A 52-week year is approximately 1.8868% shorter than a 53-week year. Those percentages differ because their denominators differ.

Comparison	Growth rates flagged At risk
Current year longer than prior year	Positive growth up to and including (current weeks ÷ prior weeks − 1) × 100
Current year shorter than prior year	Growth from (current weeks ÷ prior weeks − 1) × 100 up to and including zero
Equal-length years	Check not needed

A comparison is At risk if any of its required R&D, revenue or gross-profit rates falls in the relevant range. Otherwise it passes. This is a sensitivity check under an average-week assumption, not a measurement of the extra week's actual contribution. Seasonal sales and expense patterns can make that contribution different from an average week.

For Apple, FY2023 revenue and gross profit fell despite the longer year, while FY2024 revenue and gross profit rose despite the shorter year. R&D growth also stays outside the relevant risk ranges. Neither annual label changes under this sensitivity check.

Confidence describes the data checks and application of the rules. It is High when the inputs pass validation and all comparisons use equal-length years; Medium when unequal periods occur but all sensitivity checks pass; and Low when a check is At risk. No confidence rating is assigned if the data gate fails. Apple receives Medium, which is not a rating of the likelihood that R&D produced growth.

Retrieve the inputs through FMP MCP

The workflow uses the FMP Income Statement API. Through the MCP connector, Claude calls the statements tool with endpoint=income-statement, symbol=AAPL, period=annual and limit=10.

Measure	FMP field	Apple reporting line
Revenue	revenue	Total net sales
Gross profit	grossProfit	Gross margin, in dollars
R&D expense	researchAndDevelopmentExpenses	Research and development
Operating income	operatingIncome	Operating income
Fiscal year	fiscalYear	Fiscal year
Period end	date	Last day of the fiscal year
Annual flag	period	FY
Currency	reportedCurrency	USD

Use the dollar amount of Apple's gross margin, not its gross-margin percentage. Keeping the financial inputs in one annual dataset avoids mixing quarterly, trailing-twelve-month and differently normalized records. FMP supplies the values in dollars; dividing by one million produces the exact USD million inputs below.

Fiscal-year lengths come from Apple's filings and are supplied separately in the prompt. The filing references also let you compare the retrieved values with the reported statements.

Fiscal year	Filing reference	Financial-value check	Fiscal-year length source
FY2021	Form 10-K for year ended 2021-09-25	FY2023 Form 10-K, Item 7: net sales, gross margin and operating-expense tables; operating income calculated as gross margin minus total operating expenses	FY2023 Form 10-K, Item 7, Fiscal Period
FY2022	Form 10-K for year ended 2022-09-24	FY2023 Form 10-K, Item 7: net sales, gross margin and operating-expense tables; operating income calculated as gross margin minus total operating expenses	FY2023 Form 10-K, Item 7, Fiscal Period
FY2023	Form 10-K for year ended 2023-09-30	FY2023 Form 10-K, Item 7: net sales, gross margin and operating-expense tables; operating income calculated as gross margin minus total operating expenses	FY2023 Form 10-K, Item 7, Fiscal Period
FY2024	Form 10-K for year ended 2024-09-28	FY2024 Item 7 net sales; FY2025 Item 8 statements of operations, prior-year column	FY2025 Form 10-K, Note 1
FY2025	Form 10-K for year ended 2025-09-27	FY2025 Item 8, statements of operations	FY2025 Form 10-K, Note 1

The five-year example is fixed at FY2021 to FY2025, using data retrieved October 15, 2026. Requesting ten records brought all five target years into that retrieval. For later runs, confirm that the requested history still reaches FY2021; a fixed record limit will eventually exclude it. Historical records can also change after revisions, so retain the dated inputs when you need to compare runs.

To connect Claude, follow the FMP MCP setup instructions. With an active FMP API key and access to custom connectors, open Claude's Settings, choose Connectors, then Add custom connector. Replace the placeholder in this URL with your key:

https://financialmodelingprep.com/mcp?apikey=YOUR_FMP_API_KEY

Save the connector and start a new conversation. Run Prompt 1 first, check its inputs against the filings, then use Prompt 2 in the same conversation to calculate the screen.

Prompt 1 Retrieve and check the annual inputs

Use this prompt to select the fixed five-year window and keep retrieval separate from calculation. The example tables show the October 15, 2026 snapshot; the reusable prompts include the period-length sensitivity rules described above.

Objective

Retrieve Apple Inc. (AAPL) annual income-statement data through the FMP MCP

connector for the fixed example window FY2021 to FY2025, validate it, and

return the unrounded inputs used by Prompt 2. Do not calculate growth rates,

ratios or classifications in this step.

Data rules

- Use the FMP connector only. Call the statements tool with

endpoint=income-statement, symbol=AAPL, period=annual, limit=10.

- Make one call. Keep only records whose fiscalYear is 2021, 2022, 2023,

2024 or 2025. Ignore the other returned years.

- Map fields exactly: fiscalYear, date, period, reportedCurrency, revenue,

grossProfit, researchAndDevelopmentExpenses, operatingIncome. Do not

substitute another field, a quarterly record, a TTM record, web search

or a filing value.

- Convert USD values to USD millions by dividing by 1,000,000. Do not round.

- Fiscal-year length in weeks from Apple's Form 10-K filings:

FY2021 = 52, FY2022 = 52, FY2023 = 53, FY2024 = 52, FY2025 = 52.

- If the call fails, or a required field in a kept record is missing or

non-numeric, retry the same call once. If the problem remains, mark that

record "Review Required: unavailable" and leave the value blank.

Never treat a missing value as zero.

Validation checks

1. Each fiscal year from 2021 to 2025 appears exactly once.

2. period = FY for every kept record.

3. reportedCurrency = USD for every kept record.

4. Revenue, gross profit, R&D expense and operating income are present

and numeric in every kept record.

Output

Table 1. Retrieval log

Columns: Item | Result

Rows: Connector and tool | Endpoint | Parameters | Retrieval date |

Records returned | Fiscal years returned | Records kept |

Duplicate fiscal years | Missing or non-numeric required fields |

Retries | Reporting currency

Table 2. Annual inputs, fixed example window

Columns: Fiscal year | Period end | FMP period |

Fiscal-year length (weeks) | Revenue (USD M) | Gross profit (USD M) |

R&D expense (USD M) | Operating income (USD M) | Input status

One row per kept fiscal year, oldest first.

Input status = Validated only when all validation checks pass for the

record. Otherwise show Review Required with the reason. Include a blank

row for any missing target year; never silently omit a missing year.

After the tables, state in one sentence whether all four validation

checks passed. Do not interpret the figures.

Apple's annual inputs

The retrieval returned ten annual records, FY2016 to FY2025, and retained the five target years. All four validation checks passed, with no retries.

Table 1. Retrieval log

Item	Result
Connector and tool	FMP connector, statements tool
Endpoint	income-statement
Parameters	symbol=AAPL, period=annual, limit=10
Retrieval date	October 15, 2026
Records returned	10: FY2016 to FY2025
Records retained	5: FY2021 to FY2025
Duplicate fiscal years	None
Missing or non-numeric required fields	None
Retries	0
Reporting currency	USD for all retained records

Table 2. Annual financial inputs in USD millions

Fiscal year	Period end	Weeks	Revenue	Gross profit	R&D expense	Operating income
FY2021	2021-09-25	52	365,817	152,836	21,914	108,949
FY2022	2022-09-24	52	394,328	170,782	26,251	119,437
FY2023	2023-09-30	53	383,285	169,148	29,915	114,301
FY2024	2024-09-28	52	391,035	180,683	31,370	123,216
FY2025	2025-09-27	52	416,161	195,201	34,550	133,050

All records have period=FY, use USD, and passed the input checks. Fiscal-year lengths are 52, 52, 53, 52 and 52 weeks, respectively. The amounts match the filed figures to the USD million; year lengths are the only inputs supplied separately from FMP.

Prompt 2 Calculate direction and R&D intensity

Prompt 2 uses only the checked annual-input table, including fiscal-year lengths. It calculates ratios and growth, applies the annual labels, and checks next-year observations before determining the overall result. It makes no additional FMP calls.

Objective

Using only Table 2 from Prompt 1, test whether Apple's reported R&D expense,

revenue and gross profit moved in the same direction in each fiscal year

from FY2022 to FY2025. This is a direction screen. It does not measure

whether output kept pace with R&D, R&D efficiency or innovation

productivity, and it must not state that R&D caused any change.

Data rules

- Use the unrounded Table 2 values. Apply every threshold to unrounded

results, then round percentages and percentage-point changes to one

decimal place for display.

- Growth = (current year / prior year - 1) x 100, where the prior year is

the immediately preceding fiscal year.

- R&D expense, revenue and gross profit: if the prior-year value is zero

or negative, show "Not calculable: nonpositive base"; if either value is

unavailable, show "Not calculable: unavailable". Label that year

Data issue.

- Operating income is context only, because it is reported after R&D

expense. If the prior-year value is zero or negative, show the USD M

change and its direction (Up or Down) instead of a percentage. Operating

income never changes a label.

- R&D % of revenue = R&D expense / revenue x 100.

R&D % of gross profit = R&D expense / gross profit x 100.

If the denominator is zero or negative, show "Not calculable: nonpositive

denominator" and set Data or comparability review = Required.

- A growth rate of exactly zero counts as not growing.

Step 1. Week-length check

For each comparison of years with different lengths, use the

average-week sensitivity thresholds below for each R&D, revenue and

gross-profit growth rate:

- Current year longer: threshold = (current weeks / prior weeks - 1)

x 100. A positive rate up to and including this threshold is At risk.

- Current year shorter: threshold = (current weeks / prior weeks - 1)

x 100. A rate from this negative threshold up to and including zero

is At risk.

- Otherwise show Passed. Equal lengths show Not needed.

For 53 versus 52 weeks, the positive threshold is about +1.9231%.

For 52 versus 53 weeks, the negative threshold is about -1.8868%.

A comparison is At risk if any required rate is At risk. This is an

average-week sensitivity check, not a measurement of the extra week's

actual contribution. Do not adjust the reported inputs or growth rates.

Step 2. Annual labels, FY2022 to FY2025, checked in this order

- Data issue: a required growth rate is not calculable.

- Not evaluated: R&D growth <= 0.

- Aligned: R&D growth > 0, revenue growth > 0 and gross-profit growth > 0.

- Mixed: R&D growth > 0 and exactly one of revenue growth or

gross-profit growth > 0.

- Weak: R&D growth > 0, revenue growth <= 0 and gross-profit growth <= 0.

Labels describe direction only. Aligned does not mean R&D became more

efficient.

Step 3. Next-year direction, FY2022 to FY2025

Reuse the following fiscal year's revenue and gross-profit growth from

Step 2. This is not a measure of a development lag.

- Positive: both next-year growth rates > 0.

- Partial: exactly one > 0.

- None: neither > 0.

- Outside window: the next fiscal year is after FY2025.

- Not calculable: a next-year growth rate is not calculable.

Used in Subsequent recovery rule = Yes only when the row's own label is Mixed or Weak

and its next-year direction is Positive. Otherwise No.

Step 4. Overall classification, applied in this order

1. Data gate: if any year is Data issue, or any Table 2 row is Review

Required, the result is Difficult to Interpret (data issue). Stop.

2. Aligned: at least 3 of the 4 years are Aligned.

3. Subsequent recovery: at least one row has Used in Subsequent recovery rule = Yes, and there is

no persistent deterioration. Persistent deterioration means two or

more consecutive years in which revenue growth <= 0 and gross-profit

growth <= 0.

4. Weak: at least 2 years are Weak, and no row has Used in Subsequent recovery

rule = Yes.

5. Otherwise: Difficult to Interpret (inconclusive evidence). This

includes windows with two or more Not evaluated years.

Step 5. Confidence in data and rule application

This rates the inputs and the rule application only. It does not rate

whether R&D produced growth.

- High: data gate passed and every comparison is between years of equal

length.

- Medium: data gate passed, at least one comparison spans unequal

fiscal-year lengths, and every week-length check is Passed.

- Low: data gate passed, but any week-length check is At risk.

- Not assigned: data gate failed.

Data or comparability review = Required when the data gate fails, confidence is Low, or an

intensity ratio is not calculable. Otherwise Data or comparability review = No exception identified.

Step 6. Follow-up action for the overall result

- Aligned: test whether the same direction holds in the next reported

fiscal year, and compare it with the intensity trend in Table 3.

- Subsequent recovery: review segment and product disclosures for the Mixed or Weak

year and the year after it.

- Weak: review the filing's explanation of higher R&D expense alongside

lower revenue and gross profit.

- Difficult to Interpret (data issue): resolve the named field or record,

then rerun Prompt 1.

- Difficult to Interpret (inconclusive evidence): extend the window or

review the years labeled Not evaluated or Mixed.

Output

Table 3. R&D intensity

Columns: Fiscal year | R&D % of revenue | R&D % of gross profit |

Change in R&D % of revenue (pp) | Change in R&D % of gross profit (pp)

FY2021 change cells = Baseline.

Table 4. Year-over-year direction

Columns: Fiscal year | Weeks compared | R&D growth | Revenue growth |

Gross-profit growth | Operating-income growth (context) |

Week-length check | Annual label

FY2021 row: growth, check and label cells = Baseline.

Table 5. Next-year revenue and gross-profit direction

Columns: Fiscal year | Annual label | Next fiscal year |

Next-year revenue growth | Next-year gross-profit growth |

Next-year direction | Used in Subsequent recovery rule

Rows FY2022 to FY2025.

Table 6. Overall result

Columns: Item | Result

Rows: Aligned years | Mixed years | Weak years | Not evaluated years |

Data issue years | Positive next-year observations, all years |

Positive next-year observations after a Mixed or Weak year |

Persistent deterioration | Data gate | Overall classification |

Deciding rule | Confidence in data and rule application |

Data or comparability review | Follow-up action

Finish with no more than three sentences that describe direction only.

Do not use "caused", "drove", "return on R&D" or "kept pace".

R&D intensity results

R&D took a growing share of revenue in every comparison year. Its share of gross profit rose in three of four comparisons; FY2024 was the exception. Apply thresholds and calculate percentage-point changes before rounding the displayed figures.

Table 3. R&D intensity

Fiscal year	R&D % of revenue	R&D % of gross profit	Change in revenue ratio, pp	Change in gross-profit ratio, pp
FY2021	6.0%	14.3%	Baseline	Baseline
FY2022	6.7%	15.4%	+0.7	+1.0
FY2023	7.8%	17.7%	+1.1	+2.3
FY2024	8.0%	17.4%	+0.2	−0.3
FY2025	8.3%	17.7%	+0.3	+0.3

A positive ratio change means R&D expense grew faster than the denominator that year. FY2024 is the only comparison in which gross profit grew faster than R&D expense. No year shows revenue growing faster than R&D expense.

Annual direction and year-length results

Table 4. Year-over-year growth and annual labels

Fiscal year	R&D growth	Revenue growth	Gross-profit growth	Operating-income growth, context	Annual label
FY2021	Baseline	Baseline	Baseline	Baseline	Baseline
FY2022	19.8%	7.8%	11.7%	9.6%	Aligned
FY2023	14.0%	−2.8%	−1.0%	−4.3%	Weak
FY2024	4.9%	2.0%	6.8%	7.8%	Aligned
FY2025	10.1%	6.4%	8.0%	8.0%	Aligned

Fiscal year	Current versus prior weeks	Average-week sensitivity check
FY2021	52; baseline	Baseline
FY2022	52 versus 52	Not needed
FY2023	53 versus 52	Passed
FY2024	52 versus 53	Passed
FY2025	52 versus 52	Not needed

All required growth rates fall outside the At risk ranges. The check leaves the reported growth rates unchanged and supports the same annual labels under the average-week assumption.

Next-year observations

The next-year table reuses the revenue and gross-profit rates above. It distinguishes a recovery after a Weak or Mixed year from another positive year in an already Aligned sequence.

Table 5. Next-year revenue and gross-profit direction

Fiscal year	Annual label	Next year	Next-year revenue growth	Next-year gross-profit growth	Next-year direction	Qualifying recovery
FY2022	Aligned	FY2023	−2.8%	−1.0%	None	No
FY2023	Weak	FY2024	2.0%	6.8%	Positive	Yes
FY2024	Aligned	FY2025	6.4%	8.0%	Positive	No
FY2025	Aligned	FY2026	Outside window	Outside window	Outside window	No

Both FY2023 and FY2024 have Positive next-year observations. Only FY2023 qualifies for the Subsequent recovery rule because its own label is Weak. FY2025's next year lies outside the retrieved window, which is different from invalid data.

Overall result

Table 6. Overall classification and follow-up

Item	Result
Aligned comparisons	3: FY2022, FY2024 and FY2025
Mixed comparisons	0
Weak comparisons	1: FY2023
Not evaluated comparisons	0
Data issue comparisons	0
Positive next-year observations	2: FY2023 and FY2024
Positive next-year observations after a Mixed or Weak year	1: FY2023
Persistent deterioration	No
Data gate	Passed
Overall classification	Aligned
Deciding rule	Three of four comparisons are Aligned; this rule takes precedence over Subsequent recovery
Confidence in data and rule application	Medium: unequal fiscal-year lengths, with both sensitivity checks passed
Data or comparability review	No exception identified
Follow-up	Add FY2026 when available and compare its direction with the R&D intensity trend

The overall Aligned classification rests on direction alone. R&D expense, revenue and gross profit rose together in three of four comparisons. The FY2023-to-FY2024 recovery is visible but does not determine the overall label because the Aligned rule is met first.

How to use the results in company research

Apple's R&D expense rose from 6.0% to 8.3% of revenue and from 14.3% to 17.7% of gross profit. An analyst who interprets Aligned as output keeping pace with R&D would miss that intensity trend. The measures moved in the same direction in most years, while R&D absorbed a larger share of revenue throughout the window.

Operating income followed the same growth signs as revenue and gross profit, falling in FY2023 and rising in the other comparisons. It helps describe overall profitability, but its growth reflects other expenses as well as R&D. It cannot isolate R&D's contribution.

Use the screen to decide where to investigate. For a Weak or Mixed year, read management's explanation of R&D expense and the factors affecting sales and gross profit. Reviewing revenue quality across the financial statements and product-mix concentration can help you assess whether consolidated growth hides different product or segment trends. A broader cost decomposition can place the R&D line alongside other changes in profitability.

These are follow-up analyses, rather than conclusions supplied by the direction label. The screen does not identify successful research programs, calculate a return on R&D, or establish that one year's expense produced the next year's growth.

When an input or comparison needs review

Resolve missing or duplicate fiscal years, nonannual records, unavailable fields, currency mismatches and nonpositive growth bases before interpreting a classification. A nonpositive intensity denominator also requires review, as does an At risk period-length check. For a coverage list, these checks can sit within your financial-data validation process.

A Weak or Mixed year, a subsequent recovery, or a rising R&D share of revenue is an analytical observation rather than a data failure. Keep the reason for review explicit so an unfavorable result is not confused with unusable inputs.

Apply the screen to another company or reporting window

Change the symbol, target fiscal years and fiscal-year lengths in Prompt 1, then ensure the history request reaches all five years. Preserve the same definitions and thresholds if you want comparable screens across a coverage list. Companies reporting under different accounting frameworks may recognize or capitalize development costs differently, so assess the accounting basis before comparing their R&D ratios.

The Apple example contains no Mixed year, reduction in R&D expense or negative operating-income base. Those rules remain useful safeguards when another company's records encounter those conditions; Apple's results do not demonstrate how often they will occur.

You can pair the annual observations with a long-term compounder screen to examine consistency across the same revenue and gross-profit history. Before comparing later runs, also account for revisions and restatements in financial data, since a fixed fiscal-year filter does not guarantee unchanged historical inputs.

For Apple, the next step is to add FY2026 when its annual data is available. Check both whether the three measures continue to rise together and whether R&D continues to take a larger share of revenue. Those questions preserve the distinction that makes the workflow useful: direction tells you what moved together; intensity tells you how their relative pace changed.

Tags:Model Builds

About the Author

Pranjal Saxena

Financial APIs, Claude MCP, and AI-driven research workflows

Pranjal Saxena writes technical content focused on financial data APIs, Claude MCP workflows, AI-driven research systems, and Python-based market analysis. For FMP, his work centers on turning structured financial data into practical, workflow-driven content for developers, analysts, and fintech teams. He combines experience in data science, NLP, generative AI, and financial API workflows to show how APIs, automation, and AI-assisted systems can support modern financial research and analysis.

View all articles

Data in Action