What a Score Log Teaches After Forty Entries

A sector-relative score is only as useful as your evidence that it helps you. The cheapest evidence is a log: the score, its five pillar values, the calculation date, the peer group size, what you did, and what happened relative to the sector afterwards. Forty entries is about the point where the log starts to say something.

Sort by total first

If the high-score entries did not outperform the low-score entries relative to their sectors, either the weights do not fit your holding period or the holding period does not fit the weights. That is a finding about you and the tool together, and it is worth more than any single reading.

Then sort by pillar

Most people find that one pillar carried the result and one contributed nothing for their particular style. A long-term holder often finds quality and safety did the work; a shorter-horizon trader often finds momentum did. That is the only honest reason to reweight, and it should be done by writing the new weights down before the next batch of entries, so the next forty are a real test rather than a fit.

Judge outcomes against the sector

A stock that fell five percent while its sector fell fifteen was a relative win, and a score that pointed to it did its job. Judging against the index would score that entry as a loss and teach the wrong lesson. Sector-relative scores need sector-relative outcomes.

Keep the shape, not only the sum

Two entries with a total of 70 can have opposite pillar profiles. If only the total is logged, the log cannot later show that lopsided profiles behaved differently from balanced ones, which is often the most useful thing it has to say.

The pillar definitions and default weights are documented on the Stock Expert AI methodology page. Live sector-relative scores for US-listed stocks are at www.stockexpertai.com.

This is educational content, not investment advice. Past performance does not guarantee future results.

Turnarounds and Trailing Scores: Why a Rising Rank Lags a Recovering Business

A company that has fixed itself does not look fixed in a trailing score for a while. The filings that carry the improvement arrive one quarter at a time, and a sector-relative rank built from three-year growth and trailing margins moves slowly by design. This is a feature for most companies and a blind spot for turnarounds. Knowing where the blind spot sits is the whole trick.

What the score sees during a turnaround

Trailing margins still include the bad quarters, so business quality ranks low. Three-year revenue growth still includes the decline, so growth durability ranks low. Leverage taken on during the trouble keeps financial safety low. Valuation may look cheap on depressed earnings. Momentum, if the market has noticed the recovery, is the one pillar that can already be high. The total sits in the 30s while the business is visibly improving.

How to read it

Look at the direction of each fundamental pillar over the last three or four calculation dates rather than at its level. A quality rank moving from 10 to 20 to 32 is a recovery in progress, even though 32 is still a low number. The level says where the company has been; the trend says where it is going.

The pairing that matters

Rising momentum with rising quality is the market and the fundamentals agreeing on a recovery. Rising momentum with flat quality is a rally waiting for evidence. Rising quality with flat momentum is a recovery the market has not noticed, which is either an opportunity or a sign the market knows something the filings do not yet show. Each pairing is a different situation and the total cannot tell them apart.

Why the score is built this way anyway

A score that reacted quickly to one good quarter would also react quickly to one lucky quarter. Slowness is the price of not being fooled by noise, and for the large majority of companies that trade is right. For turnarounds, the reader supplies the speed by reading the trend.

How each pillar is computed and dated is on the Stock Expert AI methodology page. Live sector-relative scores for US-listed stocks are at www.stockexpertai.com.

This is educational content, not investment advice. Past performance does not guarantee future results.

Why the Same Stock Can Rank Differently on Two Research Sites

Open two research sites, look up the same company, and you will often find two different scores. Neither site is lying. The disagreement comes from four design choices made before any number is computed, and knowing them turns a confusing contradiction into useful information.

1. What the company is compared with

One site ranks against the whole market, another against the sector, a third against a hand-picked list of competitors. A software company with a P/E of 30 is expensive against the market and cheap against software. The comparison set is the biggest single source of disagreement.

2. Which metrics feed each question

Two sites can both call a pillar quality and build it from different inputs: one from return on capital, another from margins and margin trend. Both are defensible. They will not agree on every company, and they will disagree most on companies where the inputs point in different directions.

3. When the data was pulled

A score computed the day after a filing and one computed a week before it describe different quarters. If two sites show different numbers, compare their calculation dates before comparing anything else. The stale one is not wrong; it is late.

4. How the pillars are combined

An average forgives one weak pillar. A product punishes it. A weighted scheme tilts toward whichever question the designer trusts most. Two sites with identical pillar values can still produce different totals through this step alone.

What to do with the disagreement

Treat it as a prompt, not a problem. Find which of the four choices differs, and you will usually have learned something specific about the company: that it is strong against its sector but ordinary against the market, or that its latest quarter changed the picture. That is more useful than either score on its own.

The comparison set, metrics, dates and weighting used here are documented on the Stock Expert AI methodology page. Live sector-relative scores for US-listed stocks are at www.stockexpertai.com.

This is educational content, not investment advice. Past performance does not guarantee future results.

Why the Same Company Can Score Differently in Two Classification Systems

A sector-relative score depends on one thing before any metric is read: which companies count as the peer group. Two research tools using two classification systems can put the same company in different neighbourhoods, and the scores that follow will disagree. Neither is wrong. They are answering the question against different comparison sets.

Where classifications disagree

A payments company can be labelled technology in one system and financials in another. A pharmacy chain can be retail or healthcare. A tower operator can be real estate or telecom. In each case the company's own numbers are identical, but its margins, leverage and growth are ranked against a different crowd, and the percentiles move.

Why one system has to be chosen and named

A score that does not say which classification it uses cannot be checked. The methodology has to name the system, the level of granularity, industry or sector, and the rule for companies that sit on a boundary. Once named, a reader can at least see what the company is being compared with and judge whether that comparison fits.

What to do when the label looks wrong

Read the raw metrics for the four or five companies you consider the true competitors, and rank by eye. Five companies can be compared by hand in a few minutes. If the hand comparison and the score disagree, the label is the likely cause, and the score should be read as describing a different neighbourhood than the one you had in mind.

Boundary companies are worth extra attention

A company on the edge of two sectors often has the economics of both, and the market may price it against whichever comparison set is in favour that year. Its score will swing with the label more than its business does. That is a reason to look at the components rather than a reason to distrust the method.

The classification system and boundary rules used are named on the Stock Expert AI methodology page. Live sector-relative scores for US-listed stocks are at www.stockexpertai.com.

This is educational content, not investment advice. Past performance does not guarantee future results.

How Often Should You Re-Check a Stock Score? A Cadence That Matches the Data

Checking a stock score every day feels diligent and is mostly wasted. Four of the five pillars are built from quarterly filings and do not move between them. Checking it once a year misses the moments that matter. The right cadence follows the data, and the data has two speeds.

The slow speed: filings

Business quality, financial safety, valuation and growth durability all rest on reported financials. They change when a new quarterly report reaches the data feed, usually a few days to a few weeks after the company reports. Between filings they are static, apart from the valuation pillar drifting with price. Re-reading them weekly tells you nothing you did not know last week.

The fast speed: prices

Momentum is built from six and twelve month price change ranked inside the sector and moves daily. It is one pillar out of five, and it is deliberately slow-moving even so, because a six-month window does not turn on a single bad day.

A practical schedule

After each earnings season, read the four fundamental pillars for every company on your list. That is four sessions a year, and they are the ones that matter. Once a month, glance at momentum and at the total, mainly to catch a company whose sector-relative rank is diverging from its sector. Ignore the score between those points unless the company itself reports something.

The tell-tale of a stale score

Compare the calculation date with the company's most recent earnings date. If the score predates the report, it is describing the previous quarter and should be read as provisional until the feed catches up. That single check prevents most of the mistakes people make with dated numbers.

The refresh cadence for each pillar is documented on the Stock Expert AI methodology page. Live sector-relative scores for US-listed stocks are at www.stockexpertai.com.

This is educational content, not investment advice. Past performance does not guarantee future results.

Momentum Against the Sector, Not the Index: Why the Benchmark Choice Changes the Answer

Most momentum screens measure a stock's price change against the broad index. A sector-relative score measures it against the stock's own sector. The two answers disagree more often than people expect, and the disagreement is the point.

What index-relative momentum hides

When a whole sector rallies, every stock in it shows strong momentum against the index, including the laggards. When a whole sector sells off, every stock in it shows weak momentum, including the leaders. Index-relative momentum is mostly a sector bet in disguise. It tells you which industry the market likes this quarter, which is useful, but it is not information about the company.

What sector-relative momentum shows

Ranking six-month and twelve-month price change inside the sector strips the sector move out. What remains is whether the market is treating this company better or worse than its direct competitors. A stock in the 90th percentile of its sector on momentum is being singled out by other investors for a reason, and that reason is usually visible in the fundamental pillars a quarter later.

Why it is the only price-based pillar

Four pillars are built from filings and move quarterly. Momentum is built from prices and moves daily. It is included because it is the fastest signal that other investors have noticed what the fundamentals show, and it is limited to one pillar out of five because price alone is the easiest signal to overfit.

Reading it in a selloff

A high-quality company selling off with its sector on no company news usually keeps a strong sector-relative momentum rank even while its index-relative momentum collapses. That divergence is the profile of a pullback rather than a breakdown. The reverse, a stock underperforming its sector while the sector rallies, is the profile worth checking for a company-specific problem.

The momentum inputs and their refresh cadence are documented on the Stock Expert AI methodology page. Live sector-relative scores for US-listed stocks are at www.stockexpertai.com.

This is educational content, not investment advice. Past performance does not guarantee future results.

Growth Durability: Why Three Years of Revenue Beats One Great Quarter

Growth is the pillar most people want to read first and the one most easily fooled by a single quarter. The growth durability pillar in a sector-relative score is built to resist that, and this post explains what it looks at and why.

Three years, not one quarter

The primary input is three-year revenue growth, ranked inside the sector. A company that grew forty percent last quarter off a weak comparison can look spectacular on a one-quarter view and ordinary on a three-year view. The longer window is less exciting and far more informative about whether the growth is a trend or an event.

Earnings growth against revenue growth

The second input compares earnings growth with revenue growth. Earnings outpacing revenue for several years usually means margin expansion, which is real but has a ceiling; a company cannot expand margins forever. Earnings lagging revenue for years usually means the growth is being bought with price cuts or spending. Neither is disqualifying, but each changes how durable the growth is likely to be.

Organic against acquired

Growth funded by acquisitions is a different animal from growth from the existing business. Where the data allows it, the pillar flags companies whose revenue growth is mostly acquired, because that growth stops the moment the acquiring stops and it usually arrives with debt that shows up in the financial safety pillar.

Why it is ranked inside the sector

Ten percent revenue growth is sleepy for a cloud company and remarkable for a packaged-food producer. Ranking inside the sector turns the same ten percent into a low percentile in one case and a high percentile in the other, which is the only way the number can be compared across the market.

Reading it with the other pillars

High growth durability with low business quality is a company growing without earning much on its capital; watch the margins. High growth durability with low financial safety is growth funded by leverage; watch the coverage. The pillar on its own is a fragment, which is why the five are shown side by side.

The metric definitions and update cadence are documented on the Stock Expert AI methodology page. Live sector-relative scores for US-listed stocks are at www.stockexpertai.com.

This is educational content, not investment advice. Past performance does not guarantee future results.

Cyclicals at the Bottom of the Cycle: Where a Trailing Score Is Most Wrong

Every score built from reported financials is a trailing score. It describes what the company just did, not what it is about to do. For most businesses that lag is tolerable. For deep cyclicals at a turning point it is the whole story, and this is the situation where a sector-relative score is most likely to be wrong in a way that costs money.

What the score sees at the trough

Take a steel producer or a memory chip maker at the bottom of its cycle. Trailing margins are thin or negative, so business quality ranks near the bottom of the sector. Revenue fell, so growth durability ranks low. Earnings collapsed while the share price fell less, so the trailing earnings multiple looks high and valuation ranks poorly too. Momentum is weak because the stock has been falling. Four pillars out of five say avoid. The total might sit in the teens.

What actually happens next

If the cycle turns, earnings can triple inside a year. The company that scored 15 becomes the company that scores 75, and most of the price move happens before the filings that would have raised the score arrive. Anyone who waited for the number to improve bought after the recovery, not before it.

How to read the score in this case

First, look at the whole sector's score history, not only the company's. If every company in the group is scoring low at once, the cycle is the cause, not the company. Second, compare the company's rank now with its rank at the last peak. A company that was top quartile at the peak and is still top quartile at the trough is the strongest operator in a weak industry, and that is a different thing from a weak company. Third, weight financial safety more heavily than usual. The question at the trough is not who earns the most but who survives to the recovery.

What the score is still good for

Even at the trough, the within-sector rank on financial safety and business quality separates the companies that will recover from the ones that will dilute or default first. The total is misleading; the components are not.

How the pillars are built and why financial safety is kept separate from the total is on the Stock Expert AI methodology page. Live sector-relative scores for US-listed stocks are at www.stockexpertai.com.

This is educational content, not investment advice. Past performance does not guarantee future results.

Small Peer Groups: When a Sector-Relative Score Stops Meaning Much

A sector-relative stock score converts every metric into a percentile inside the company's own peer group. That works well when the group holds forty or eighty companies. It works badly when the group holds six. This post is about the second case, because it is the most common way a good method produces a misleading number.

Why small groups break percentiles

Third out of six is reported as roughly the 50th percentile. But with six companies, moving one place shifts the percentile by nearly twenty points. A single peer reporting a good quarter can push a company from 60 to 40 on a pillar while nothing at the company changed. The number looks as precise as a percentile from a group of eighty, and it is not.

How to spot it

Look at the peer group size next to the score. Anything below fifteen deserves caution. Below ten, treat the pillar values as coarse labels: top third, middle, bottom third. Do not read the second digit.

What to do instead

Widen the group one level up the classification. If the industry has six companies, rank inside the sector, which may have sixty. You lose some like-for-like precision and gain statistical stability, and for small industries that trade is worth making. Alternatively, keep the small group but look at the raw metrics beside the ranks. Six companies can be read by eye.

Where this shows up most

Niche industrials, specialty insurers, single-product biotech and small-cap regional banks. These are also the areas where a retail investor is most likely to be comparing two or three names directly, so the percentile adds the least and the raw comparison adds the most.

Peer group construction and the sector classification used are described on the Stock Expert AI methodology page. Live scores with their peer group context are at www.stockexpertai.com.

This is educational content, not investment advice. Past performance does not guarantee future results.

Dual-Channel Data: Why the AI Never Writes the Numbers

Language models are good at explaining and bad at arithmetic under pressure. Ask one for a company's debt-to-equity ratio and it will produce a plausible figure, sometimes the right one, sometimes a number from a different year or a different company. For a stock research tool, "sometimes" is not an acceptable failure rate. The design answer is to keep two channels apart.

Channel one: the figures

Prices, filings, ratios and the score itself come from market data providers and from deterministic code. They are rendered verbatim on the page. No model touches them, rewrites them or rounds them. If a figure is wrong, the cause is a data-provider error or a code bug, both of which can be traced and fixed.

Channel two: the commentary

The model is allowed to write text around the figures: what a low valuation percentile means for a company in that sector, why a momentum spike without a fundamental change deserves caution, which pillar is doing the work in a total. It reads the numbers as inputs. It is never the source of one.

Why the boundary matters for the reader

The two channels fail differently. A wrong figure looks exactly like a right one and is dangerous in proportion to how confident the page looks. Weak commentary is merely unhelpful, and a reader can ignore it. Keeping the channels separate means the dangerous failure mode is removed from the model entirely, and the remaining failure mode is the survivable one.

How to check whether a tool does this

Ask three questions. Does every figure on the page carry a source and a date? If the commentary mentions a number, does the same number appear in a table or a field that came from data, or only in the prose? When the tool is asked for a metric it does not have, does it say so, or does it produce one anyway? A tool that passes all three is keeping the channels apart, whatever its marketing says.

Stock Expert AI is built on this rule: figures are rendered from licensed data, the model writes only commentary, and every MoonshotScore carries its calculation date. The full design is described on the methodology page.

This is educational content, not investment advice. Past performance does not guarantee future results.

Rank Inside the Sector First: Why Market-Wide Screens Mostly Rank Industries

Most stock screens rank companies across the whole market. That sounds thorough. In practice it mostly ranks industries, and the list that comes out tells you which sectors have a high base rate on a metric, not which companies are good at what they do.

What a market-wide sort really shows

Sort every US-listed company by revenue growth and the top decile fills with software and biotech. Sort by dividend yield and it fills with REITs and pipelines. Sort by leverage and utilities sink to the bottom. None of this is new information. It is the shape of the economy, repeated every time you run the screen.

Ranking inside the sector first

Rank the same metric only among a company's sector peers and the industry base rate disappears. A utility in the top quartile of utilities on interest coverage is genuinely safe for a utility. A software company in the bottom quartile of software on gross margin is genuinely weak for software, even if its margin would look fine next to a grocer. The rank now describes the company rather than its neighborhood.

Why the order of operations matters for a total

Once every input is a within-sector percentile, the inputs can be averaged without one metric dominating because of its scale, and totals can be compared across sectors. A 90 in banks and a 90 in semiconductors both mean top decile of their own group. That is the property a single 0-100 number needs before it is worth publishing at all.

A practical habit

Before acting on any ranked list, ask which sectors dominate the top decile. If the answer is one or two, the list is ranking sectors. Re-run it inside the sector you actually care about and see whether the same names survive.

The ranking step, the five pillars and the calculation dates behind MoonshotScore are documented on the Stock Expert AI methodology page, and live scores are at www.stockexpertai.com.

This is educational content, not investment advice. Past performance does not guarantee future results.

What a Calculation Date Tells You That a Score Cannot

A stock score without a date is a photograph without a timestamp. It may be accurate, but you cannot tell whether it shows this quarter or the last one. The calculation date is the smallest piece of metadata on a score and, in practice, the one that prevents the most mistakes.

Two inputs, two clocks

A sector-relative score mixes two kinds of data that age at different speeds. Prices refresh every trading day, so the momentum pillar is never more than a day old. Fundamentals refresh only when a company files, so the quality, safety, valuation and growth pillars can sit unchanged for three months and then jump in a single day. A score computed the day before a filing and a score computed the day after can differ sharply while the price has barely moved.

Three questions the date answers

Is the fundamental half stale? If the calculation date is earlier than the company's most recent quarterly report, four of the five pillars are describing the previous quarter. Treat the total as provisional until it is recomputed.

Are two scores comparable? Comparing a stock scored on Monday with a peer scored three weeks later is comparing two different snapshots of the sector. The ranks were computed against different peer distributions. Only scores with close dates should be ranked against each other.

Did the score change, or did the data? When a score moves and the price did not, the date lets you check whether a new filing, a restatement or a data-provider correction is the cause, rather than a change in the business.

Why free data makes this more important, not less

Free and low-cost data sources post fundamentals days or weeks after the filing, and they occasionally restate earlier figures. A platform that hides the calculation date hides that lag. A platform that shows it lets the reader decide whether the number is fresh enough for the decision at hand.

On Stock Expert AI, every MoonshotScore is published with its calculation date next to the five pillar scores, and the refresh rules for each input are described on the methodology page.

This is educational content, not investment advice. Past performance does not guarantee future results.

Why Two Stocks With the Same Score Can Be Very Different Companies

Two stocks can carry the same 0-100 score and still be very different companies. The total is a summary, and summaries hide shape. Before acting on a number, look at the five components that produced it.

The same 70, two shapes

Imagine a stock that scores 70 because it ranks near the 95th percentile of its sector on price momentum and around the 45th percentile on business quality, financial safety, valuation and growth durability. Now imagine a second stock that scores 70 because it sits near the 70th percentile on every one of the five. The totals match. The businesses do not.

The first is a stock the market has recently fallen in love with while the fundamentals are middling for its sector. That profile can keep working, but it depends on other investors staying interested, and it tends to give back gains quickly when the crowd moves on. The second is a consistently above-average company that nobody is particularly excited about. Its risk is boredom, not a reversal.

Three questions to ask of any total

Which pillar is doing the work? If one component is far above the others, the score is really a statement about that one thing. Name it before you rely on the total.

Is the weakest pillar a deal-breaker? A stock can score in the 80s with financial safety in the bottom quartile of its sector. For a long holding period, that single weak leg matters more than the strong average.

How old is the fundamental data? Momentum updates daily, filings update quarterly. A score whose fundamental inputs predate the latest report is a different score, even if the number has not moved. Check the calculation date.

Why the sector-relative step makes this easier

Because each pillar is a percentile inside the company's own sector, the components are comparable with each other. A 90 on valuation and a 90 on momentum both mean top decile of the neighborhood. That is what lets you read the shape at a glance rather than converting a P/E, a debt ratio and a six-month return into a common scale by hand.

On Stock Expert AI, every MoonshotScore is shown with its five pillar scores and its calculation date, and the ranking method is documented on the methodology page.

This is educational content, not investment advice. Past performance does not guarantee future results.

How to Read a 0-100 Stock Score Without Fooling Yourself

A single number is seductive. Give an investor a stock score between 0 and 100 and most will treat 82 as "good" and 31 as "bad" without asking the only question that matters: good compared to what?

This post is a short field guide to reading composite stock scores, including the one we publish at Stock Expert AI, so that the number helps you instead of replacing your judgment.

1. Ask what the score is relative to

A score built on absolute thresholds (P/E under 15 is "cheap") will systematically favor banks and punish software. A score built on sector-relative ranks compares each company only to its own peer group. The second kind is comparable across industries; the first is not. Our own methodology page explains why every MoonshotScore input is ranked inside its sector before it is combined.

2. Look at the components, not just the total

Two stocks can both score 70. One is a 95 on momentum and a 45 on everything else. The other is a 70 across the board. They are different situations and deserve different position sizes. If a score does not show its components, treat it as a headline, not an analysis.

3. Check the date stamp

Fundamentals update quarterly, prices update every second. A score computed on two-quarter-old filings looks identical to a fresh one. Always find the "as of" date before acting.

4. Separate the AI from the arithmetic

When an AI model writes the commentary around a score, make sure it is not also inventing the numbers. We keep the two channels separate: figures are rendered verbatim from market data providers, and the model is only allowed to explain them. If a tool cannot tell you where its numbers come from, that is the answer.

5. Audit the score against outcomes

The honest test of any scoring system is what happened afterward. We published an audit of 6,213 of our own score records to show what a high or low score has historically meant. Ask the same of any score you use.

A score is a filter, not a verdict. Use it to shorten the list, then do the work on what is left.

This is educational content, not investment advice. Past performance does not guarantee future results.