Disclosures

01Who we are

Loep is operated by ADJL Capital LLC. Loep is a publisher of research software. It manages no money, executes no trades, holds no assets for anyone and tailors nothing to a person's circumstances. Nothing on it is offered as the service of an investment adviser or a broker, and reading it creates no relationship of that kind. Write to laskowskidanny@gmail.com.

02Research, never advice

Everything on Loep is research, never advice. In full:

• It is not investment advice, tax advice or legal advice, and it is not a substitute for any of them.
• There is no fiduciary relationship. Using Loep, paying for it or writing to it does not make the operator your adviser, your agent or your fiduciary, and the operator owes you no duty of that kind.
• Nothing is a recommendation to buy, sell or hold any security, property or other asset, and nothing is an offer or a solicitation of one.
• Past results are not indicative of future results. A figure that describes what a price, a rent or a company did is a description of the past. Loep shows no performance of its own and forecasts nothing.
• Every figure is software output on data you supplied or a provider reported. It does not know your circumstances, your holdings, your tax position, your time horizon or your tolerance for loss, and it forms no view about them.
• Verify before acting. Read the source stamped beside a figure, check it where it matters, and take your own advice. Do not buy a building, or a share of anything, because a screen said a number.

The sheets carry this in one line, Research, never advice., and a verdict carries it in full: "This is research, never advice. It does not know your situation, and every input above is yours to change."

03Retrieved, modeled, assumed

Every figure on a sheet is one of three things, and the sheet says which:

• Retrieved: taken from a record or a table somebody else published, such as a parcel's assessed taxes, a state's energy rate or a company's filed revenue. Accurate as of when it was read, and no more current than its source, which is stamped beside it with a date.
• Modeled: produced by Loep's own arithmetic on retrieved figures and assumptions, such as the energy a building of that size uses or a coverage ratio. A reasonable estimate, and an estimate.
• Assumption: a value you set yourself, such as vacancy, the loan terms or the exit year. The defaults are published beside each input, and changing one changes the result.

Where a figure could not be read, the sheet shows a gap and says why. A gap is never a zero, and silence from a source is never a finding.

Clocks. Market prices, options and estimates arrive on a delayed clock and are stamped with it; nothing on Loep is a live quote. Filings are shown as filed, by form, accession number and date. Public tables carry their vintage. Cached provider answers are reused for the periods the privacy page states.

04The mechanical equity signal

Beside a filings report, the market research sheet prints a one-word signal for a named security: BUY, NEUTRAL or SELL. A single word about a specific stock deserves more than a general disclaimer, so this is exactly what it is.

It is a published rule, applied mechanically to prices. Three conventional indicators are computed from the security's daily closing prices, as the market data provider reports them on a delayed clock:

1. Trend: where the close sits against its 200-day simple moving average, and whether the 50-day average is above or below it.
2. Momentum: a 14-day Wilder Relative Strength Index, read as mean reversion, with 30 and 70 as its thresholds.
3. Confirmation: the sign of the histogram of a 12/26/9 MACD.

Each contributes −1, 0 or +1. They are added. A total of +2 or more reads BUY, −2 or less reads SELL, and everything between reads NEUTRAL, which is the reading about two thirds of the time. That is the entire rule. The sheet prints it, with each component's own reading and the thresholds, on the same screen as the word, every time; the screen cannot show one without the other. Where there is too little price history to compute it, the sheet says how much it has and how much the rule needs, and prints no signal.

What it is not:

• Not advice, and not a recommendation. It is the output of an arithmetic rule.
• Not an opinion. Nobody at Loep forms or publishes a view of the security, and the rule is not one.
• Not personalized. It is computed only from published daily price data. It considers nothing about the reader: not your circumstances, your holdings, your tax position, your horizon or your objectives. Every viewer of the same security on the same day sees the same word.
• Not Loep's trading system. The rule is a textbook combination of indicators documented publicly since the 1970s. It is published here precisely because it is nobody's secret and reveals nothing.
• Not a price target, a valuation or a forecast. It says nothing about what the security is worth or what it will do.

Two things the rule does badly, stated because they are real. It is a trend-and-momentum rule with no knowledge of the business under the price, so it reads a company whose fundamentals are deteriorating as BUY for as long as the price holds up. And because its momentum component treats a stretched market as a reason for caution, it returns NEUTRAL most of the time. A rule that mostly declines to commit is behaving as designed, not failing.

The sheet states this beside every signal, verbatim: "This signal is mechanical and identical for every viewer. It is computed only from published price data, considers nothing about the reader or their circumstances, and is neither advice nor a recommendation. It is not Loep’s trading system."

05The filings report, positioning and forecasts

Under every filings report: "This is a report on what a company filed. It is not a recommendation, not advice, and not a view on the security." Beside it, four things with their own limits:

• Headwinds and tailwinds are three different kinds of claim and are never merged. Figures measured from the filings, by Loep's arithmetic on what the company filed with the SEC; the company's own risk-factor headings from its latest annual report, verbatim and unweighted; and news as a third-party vendor classified it, with the vendor's own sentiment score, which Loep neither verifies nor endorses. Each tier sits under its own heading with its own basis. There is no combined count and no net reading across them.
• Positioning is a record of what already happened, not a prediction. Institutional holdings come from quarterly filings that arrive weeks late and on different dates for different holders. Options open interest measures what is positioned, not what anyone expects. Insider selling is reported in full but only buying is read as a lean, because officers sell constantly for reasons unconnected to the company. Congressional disclosures come in bands and are counted, never summed. Each source leans by its own published threshold, and the sheet shows how many lean each way without combining them into a score.
• The forecasts are other people's, and are labeled as theirs. Consensus estimates, the range, the number of analysts behind each figure and how many revised up against down are reported as what analysts currently say. Loep publishes no price target, no fair value and no estimate of its own for any security, and nothing on it derives one. Analysts are frequently wrong and revise together.
• Where a source cannot be read, the sheet says so rather than showing a zero. Unavailable is not neutral, and unmeasured is not steady.

06The underwriting model

The property scanner's underwriting is arithmetic on assumptions you confirm. Nothing about it is a judgment of the property.

• You confirm the price before it runs. The price a data provider holds and the price on the listing in front of you are frequently different. Confirming it is your judgment, not Loep's.
• Every assumption is on the screen with its default beside it: the loan terms, vacancy, the growth rates, the exit year and the rest. The defaults are conventional starting points, not predictions, and each one is yours to change. Two people underwriting the same address with different assumptions get different results, and both are correct arithmetic.
• Each line of the cost stack is marked retrieved, modeled or assumption, at the granularity of the line. A line can be retrieved for its rate and modeled for its usage, and the sheet says so. Where a local figure could not be verified, the model says what it fell back to rather than presenting it as local knowledge.
• A verdict is a threshold test, not advice. The word on the stamp means a set of ratios cleared a set of numbers that are printed beside it. It is not a recommendation to anybody to buy anything, and it cannot render without its disclaimer: "This is research, never advice. It does not know your situation, and every input above is yours to change."
• Nothing is invented. A provider miss is a gap with a reason. The model refuses to fill in a figure it was not given.

07The written note

Under a survey or a filings report, Loep can write a note. That note is written by a language model, and what it is given is deliberately narrow.

• It is written from the confirmed figures alone. The brief is the arithmetic already on the sheet: the inputs you confirmed, the retrieved figures and the results. For a filings report the brief carries the filed figures and omits the share price. The model is instructed to speak as Loep, to treat the note as research and never advice, and to use no figure the brief did not contain.
• It says nothing the brief did not. It is prose about numbers, never a source of numbers. Nothing in it is parsed back into a figure, and no figure on the sheet comes from it.
• It can be wrong in the way prose is wrong: an emphasis misplaced, a connection drawn that the figures do not support, a plain sentence that reads as more than it is. Read it against the figures, not instead of them.
• It is labeled as written by a language model, with the time it was written. If the model declines to write, or stops before it is finished, no note is shown; a half note is refused rather than shown.

08The indicators

An indicator on Loep is a TradingView script, invite-only, that you add to your own TradingView charts. Loep hosts no chart; the reading happens where your charts already are, on TradingView's data, under TradingView's terms.

• Each one ships with its method written out, so you can check the reading yourself. A reading you cannot check is not on the sheet.
• An indicator is not a trading system. It is a computation on prices, drawn on a chart. It issues no instruction, and a reading is not a recommendation to trade.
• No performance is claimed. Loep publishes no backtest as an expectation of returns, no hit rate, no track record and no projection for any indicator. Anything an indicator did on past prices says nothing about what prices will do.
• Access is a permission on TradingView, granted by hand to a username and withdrawn when the plan that includes it ends.

09Past results and forward-looking statements

Past results are not indicative of future results, and Loep shows none of its own. Where a sheet describes what a market, a property or a company may do, or shows what a rule would have read on past prices, that is a description of a rule or a record, not a forecast. Actual outcomes differ, and may differ substantially. Nobody undertakes to update any of it.

10Data sources

Every figure on a sheet carries its source in the margin, and this is the list of those sources: what each one supplies, and what its license or usage policy asks of Loep in return. Loep selects them carefully, checks them where it can, and does not warrant that anything they supply is accurate, current or complete.

The names appear on this page and nowhere else on the site, by design. Everywhere else a sheet says "the property data provider" or "the market data provider", because the licenses allow Loep to present the figures inside its own instrument, and the build fails if a licensed provider's name lands on any other page.

1. RentCast

   The parcel record, the for-sale listing, the rent estimate and the value estimate for an address the person asks about. Answers are cached by address for seven days, listing status for twenty-four hours.

   AttributionLicensed. The license lets Loep present the figures inside its own instrument without naming the provider on the sheet; the provider is named here, and the figures are marked as retrieved.

2. Alpha Vantage

   Daily prices, options open interest, earnings estimates, company profile data and news sentiment scores for a ticker, on a delayed clock. The mechanical signal is computed from the daily closes it reports.

   AttributionLicensed. The provider is named here rather than on the sheet; a figure from it carries a delayed-clock stamp. The sentiment scores are the provider’s own classification and are labeled as such.

3. SEC EDGAR

   Company filings and the financial facts companies file in XBRL: the annual and quarterly reports behind every filings report, the risk-factor headings, insider transactions and institutional holdings.

   AttributionPublic domain. The SEC’s fair-access policy requires every request to carry a contact address, so Loep’s server identifies itself by name and email on each call. Filings are quoted with form, accession number and filing date.

4. U.S. Energy Information Administration

   Residential electricity and natural gas rates by state, used for the energy line of the cost model.

   AttributionPublic domain. Cited on the sheet as EIA with the series and the annual average used.

5. U.S. Census Bureau, American Community Survey

   Median gross rent by bedroom count for a ZIP (table B25031, five-year estimates), with its margin of error. Where the Census releases no figure for a ZIP the sheet says so instead of guessing.

   AttributionPublic domain. Cited on the sheet with the table, the vintage and the reliability of the estimate.

6. U.S. Census Bureau, Building Permits Survey

   Residential building permits by county, for the supply reading.

   AttributionPublic domain. Cited on the sheet with the vintage.

7. U.S. Census Bureau Geocoder

   A fallback for turning a street address into coordinates when the primary geocoder finds no house.

   AttributionPublic. Nothing about the person is sent; the address text is.

8. Tax Foundation

   County property tax rates, derived by the Tax Foundation from Census ACS five-year estimates, for the tax line of the cost model.

   AttributionCited on the sheet as Tax Foundation, from Census ACS, with the vintage.

9. FEMA National Risk Index

   Expected annual loss from natural hazards by county, as a rate, a score and FEMA’s own rating.

   AttributionPublic domain. Cited on the sheet as FEMA National Risk Index with the release date.

10. NCES Common Core of Data

    The public school directory for the covered markets: schools, charters and enrollment by county.

    AttributionPublic domain. Cited on the sheet as NCES with the school year.

11. NCES IPEDS

    Campus enrollment for the colleges in a covered market, from the Integrated Postsecondary Education Data System.

    AttributionPublic domain. Cited on the sheet as IPEDS with the term reported.

12. Zillow Home Value Index (ZHVI)

    The state price cycle table: the all-homes index, smoothed and seasonally adjusted, by state and month, from which Loep computes where a state sits against its own history.

    AttributionZillow Research publishes the index for use with attribution. The cycle table is labeled as computed from ZHVI, Zillow Research, with the month of the release, and the index itself is not redistributed.

13. OpenStreetMap contributors, via Photon and Nominatim

    Address search and reverse geocoding: the text a person types becomes a place, and a point on the map becomes an address to confirm. Photon is run by komoot; Nominatim by the OpenStreetMap Foundation. Answers are cached by query for thirty days.

    AttributionOpen Database License (ODbL). “© OpenStreetMap contributors” is shown wherever a result from either service appears, and Loep’s server keeps within each service’s usage policy by caching and throttling on the person’s behalf.

14. OpenFreeMap

    The vector map tiles under the property scanner (the Liberty style, on the OpenMapTiles schema, drawn from OpenStreetMap data). The browser fetches tiles from OpenFreeMap directly.

    AttributionFree, no key, fair use. The map shows the attribution its style requires: OpenFreeMap, OpenMapTiles and © OpenStreetMap contributors.

15. X

    Public posts mentioning a ticker, for the pulse strip, only while Loep’s connection to X is configured. Without it the sheet says posts are off; nothing is invented in their place.

    AttributionPosts are shown as posts, with the author, the time and a link to the original, and are never summarized, scored or ranked by Loep. X’s terms govern their display.

16. Stripe

    Checkout, the monthly subscription, invoices and the billing portal. Card details are entered on Stripe’s pages and never reach Loep.

    AttributionNone required. Stripe’s privacy policy governs what Stripe collects on its own pages.

17. Supabase

    The database and the sign-in service: accounts, profiles, the lookup log, saved surveys, the watchlist, the waitlist and the caches of provider answers, in a project hosted in the United States.

    AttributionNone required.

18. Netlify

    Hosting for the site and the serverless functions every lookup passes through. Ordinary request logs, including IP addresses, are kept by Netlify as by any web host.

    AttributionNone required.

19. Anthropic

    The language model that writes the note under a survey or a filings report, from the confirmed figures alone. It is the one outside service Loep sends a brief to, and the brief carries no name, email or share price.

    AttributionNone required. Every note is labeled as written by a language model, with the time it was written.

11Contact

Effective 2026-09-17. Questions about any of this go to laskowskidanny@gmail.com.