Źródło angielskie: Flags and historical patterns

Źródło angielskie: A flag means "worth a closer look"

Źródło angielskie: Some records carry computed notes beside what the city published — most of them on procurement notices, and one on the rules timeline. Every one of them is statistical context, not a finding and not an accusation. A flag says a record has a property worth noticing. It does not say anyone did anything wrong, and there is a fair explanation for each of them — emergencies really do happen, some markets genuinely have few bidders, and name matching is never perfect.

Źródło angielskie: The method follows two published guides rather than a house theory of what looks suspicious: the Open Contracting Partnership's red-flags guide and Opentender's integrity indicators.

Źródło angielskie: What each note counts

• Źródło angielskie: ⚑ Short ad window — the days between when a notice is posted and when the answer is due. We flag it when it is 10 days or fewer and less than half the agency's own median. The median comes from that agency's last 200 notices. Short windows favor incumbents who already knew the work was coming.

• Źródło angielskie: ⚑ Non-competitive method — the notice says it will pick a vendor without a full contest. It may be a deal made through talks, a single chosen source, an urgent buy, or a test project. This can be fair at times. But it is always good to know.

• Źródło angielskie: ⚑ Repeat awards — the same vendor name shows up on 3 or more award notices at the same agency within 90 days. This can point to task orders under a blanket contract just as much as favoritism. The flag just counts them — you decide what it means.

• Źródło angielskie: Context strip — how big an award is, shown as a percentile of that agency's awards in the last 12 months (shown only when the agency has 20 or more awards in that time). It also shows the vendor's share of the agency's award dollars in the same time. We use the exact published name. We do not merge name variants here.

• Źródło angielskie: Rules adoption lag estimates the time from comments closing to adoption. The rules adoption lag formula owns its pools, unfinished-case treatment, and display rules. An estimate is never a confirmed date.

Źródło angielskie: These notes use saved source snapshots. Opening a notice does not fetch fresh numbers from the publisher. Check the record's source date and coverage before relying on a comparison.

Źródło angielskie: All of these read awards as published. Published award figures can run ahead of contract registration and well ahead of what was actually paid, so a note built on them is describing publications, not spending.

Źródło angielskie: Patterns from past records

Źródło angielskie: Elsewhere on the site, past public records are used to give a sense of how long something has taken before. These are descriptions of what has already happened, not forecasts, and legal deadlines always take precedence over them.

Czas tworzenia list kandydatów

Źródło angielskie: CityScroll compares past exams by exam number, from the filing deadline to list establishment. No applicant names, scores, or ranks are used. Small groups or weak test results reduce the display to the citywide median.

Czas sprzedaży nieruchomości

Źródło angielskie: The public sample is small. Hearing notices rarely join to later auctions by lot number. The displayed comparison measures auction publication to the scheduled event, not hearing to sale. A real sale date replaces the pattern. See the property timing formula.

Etapy sprzedaży zastawów podatkowych

Źródło angielskie: CityScroll shows how often liens at the same stage reached a past cycle's final sale list, not whether a property will be sold or foreclosed. Payment, payment plans, exemptions, corrections, or canceled sales can remove a lien. Weak evidence means group totals and the lot's current status only.

Historia spraw planistycznych

Źródło angielskie: Past cases show time ranges and outcomes, not forecasts. Legal deadlines control. Groups start with the same action type and borough, widening below 20 cases. Long or unfinished cases are excluded from the time range.

Historia wnioskodawcy

Źródło angielskie: At least 20 approved, modified, or disapproved zoning outcomes are needed to show an applicant's history beside the overall rate. Weak name matches or test results mean descriptive history or no display. See the applicant outcome formula.

Źródło angielskie: How to read one

• Źródło angielskie: Treat it as a question, not an answer. The useful next step after a flag is opening the record and its official source.
• Źródło angielskie: A pattern is not a prediction. "Cases like this have usually taken this long" is a statement about past cases.
• Źródło angielskie: An estimate never becomes a deadline. If an official date exists, it is the date.
• Źródło angielskie: Nothing here is an allegation. These notes save you arithmetic. What they mean is yours to judge, from the record itself.

Źródło angielskie: If you want the full arithmetic rather than the meaning, the formulas are published: the rules adoption lag, property disposition timing, award registration dwell, and applicant-conditioned zoning outcomes.