What does one grading comparison look like?
Take fictional Player C in fictional Cedar–Birch, with a line of 5 rebounds. Each row below is a different possible scenario, not six real performances. The question stays specific: did that player finish above, below or exactly at 5 rebounds? The direction determines which side wins.
With 6 final rebounds, Over wins and Under loses. With 4, the outcomes reverse. With 5, both directions meet the exact line, so the comparison is a push. A push is kept separate from wins and losses in Propeller’s published grading definitions.
That arithmetic is not a complete settlement policy for every platform. If you are interpreting an actual entry, check the relevant platform’s settlement rules. A pending or unresolved event has no final value for this illustration. Leave that value unknown; do not fill it with 0 or automatically call it a loss or push.
| Side | Exact line | Final rebounds | Outcome |
|---|---|---|---|
| Over | 5 | 6 | Win — above 5 |
| Over | 5 | 4 | Loss — below 5 |
| Over | 5 | 5 | Push — equal to 5 |
| Under | 5 | 6 | Loss — above 5 |
| Under | 5 | 4 | Win — below 5 |
| Under | 5 | 5 | Push — equal to 5 |
Did you match the same player, event and line?
Before looking up the final stat, record the player, sport and event, stat, direction, exact line, and source or book. “Player C rebounds” is not enough: 5 rebounds against one opponent cannot be used to grade another game, and changing the line changes the comparison.
The Results sample lists date, sport, player, market, line and book, but those fields alone do not fully identify the event. Confirm that the row and final stat belong to the same game; otherwise, leave the match unresolved.
This is a method for checking evidence, not a claim that the historical feed validates every field. Keep the stored line attached to its outcome; replacing it with a later line changes the comparison you are describing.
What did our Results-page check show?
An AI-operated inspection on September 10, 2026 observed separate raw-row and collapsed-ledger labels on the Results page. The sample columns were Date, Sport, Player, Market, Line, Conf, Actual, Result and Book. The page also warned that the sample did not prove pre-event publication.
That is a historical interface observation, not an audit of the listed athlete outcomes. The capture included graded rows dated after the capture date; individual results, chronology, totals and rates were not verified. We do not use those rows as performance evidence here.
A separate September 23, 2026 documentation/source recheck confirmed the published grading definitions and counting limitations. It did not repeat the original browser observation or establish when those historical rows were published.
What does each record count?
The historical archive counts both raw analysis rows and entries retained under the current public API collapse rules. These are different units, so name the one you are using beside any figure you quote.
Repeated snapshots, books, lines and runs may appear in historical data, alongside legacy retrospective records. A larger row count therefore does not establish more independent recommendations. Nor does the collapsed count, by itself, prove that every retained entry is a uniquely published recommendation.
The archive’s price and payout evidence is incomplete, so its outcomes cannot support a defensible return or profit claim. The rows are not customer wagers. A result label only tells you how the stored comparison was graded.
The forward record has its own eligibility and settlement definitions. Read that page separately; it does not turn the historical archive into a forward test. Before comparing records, check whether their units and inclusion rules actually match.
| Record | What one unit represents | What the unit alone cannot prove |
|---|---|---|
| Raw historical analysis row | A stored analysis observation; repeated snapshots may appear | One independent recommendation, pre-event publication or a customer wager |
| Collapsed public ledger entry | An entry retained under current API collapse rules | Complete deduplication into uniquely published recommendations or profit |
| Separate forward record | An eligible published record under that page’s definitions | That the mixed historical archive qualifies for the same record |
Five checks before quoting a record
1. What was counted? Find the raw-row and collapsed-entry definitions in Results. Record the exact unit with any figure; do not shorten both to “picks.” That small wording choice preserves what the evidence actually describes.
2. When was it recorded or published? The sample Date column does not establish pre-event publication. Look for separately defined publication and event clocks. If either is absent, mark it unknown instead of treating a displayed date as proof.
3. Is it the same event and line? Match player, sport/event, stat, direction, exact line and source/book. The visible sample offers useful fields, but it does not independently establish full event identity. Document any missing match.
4. How are push and unresolved outcomes treated? Read the definitions and inspect the supplied final stat. Propeller keeps pushes separate; an unresolved event is not a final outcome. Do not assign a result simply to complete a row.
5. Which rows and evidence are excluded or missing? Check repeated snapshots, missing publication times, and price/payout gaps in the archive limitations. Read the forward page’s separate eligibility rules before comparing it with historical data.
Use the printable checklist to keep the page URL, date checked, counting unit and unanswered questions together. It asks for evidence, not a wager or account details.
Frequently Asked Questions
How does Propeller grade picks?
Propeller compares the stored side and exact line with the final stat after results are available. Over wins above the line, Under wins below it, and equality is a push. First confirm that the player, event and stat match the stored question.
Is a push a win or a loss?
Neither. In the published grading definitions, an exact tie with the line is a push and remains separate from wins and losses. That comparison does not define how every platform settles an actual entry; consult its rules for that question.
Are repeated snapshots separate recommendations?
A repeated historical analysis snapshot does not, by itself, establish a separate recommendation. Results distinguishes raw rows from collapsed public entries. Neither unit alone proves that each item was a unique recommendation published before its event.
Do historical results guarantee future outcomes?
No. Past results do not guarantee future outcomes. The historical archive includes mixed records and documented timing and price limitations. Use it as research context, preserve its counting units, and avoid treating a descriptive outcome record as proof of profit.