THE IDEA BEHIND SBOS
Start with a question.
Keep an honest record.
Stars Ball On Sundays began with Ry’s theory: a team’s stars may perform differently on Sundays. Our job is to examine that idea, share selected research, and make the record easy to question.
RY’S SUNDAY THEORY
A hypothesis has to earn its place.
Do the stars show up differently on Sundays?
The day of the week is a research question, not evidence by itself. Differences can reflect opportunity, player role, opponents, availability, or the season being observed. A promising pattern can motivate further study; it does not establish a durable advantage.
Sunday Picks keeps Ry’s original idea at the center. Daily Board extends the research across the week. Ry’s personal favorites and SBOS research remain clearly attributed.
HOW WE APPROACH THE RESEARCH
Evidence before confidence.
Check the source and the context.
NBA statistical history informs the research. Saved sportsbook observations provide the exact offered terms and price. We consider the timing and relevance of information, and recognize that incomplete availability or changing roles can limit an interpretation.
Respect the exact proposition.
A player, game, stat, threshold, operator, period, and sportsbook define a selection. An 8+ milestone and over 8 do not have the same result at eight. Unknown overtime or settlement terms remain unconfirmed.
Separate exploration from validation.
Patterns in a selected sample do not prove that future outcomes are predictable. Forward results and comparisons with meaningful baselines are needed before stronger claims are justified. Passing software tests establishes correctness of checks, not predictive skill.
Keep the inconvenient outcomes.
The posted record includes misses, pushes, and unresolved results alongside hits. Price observations and statistical outcomes serve different purposes. Neither a winning selection nor a short run establishes profitability.
WHAT YOU CAN VERIFY
Public evidence. Private implementation.
Our public research shows selected propositions, saved prices when available, source times, attribution, saved probabilities and projections when available, and recorded statistical outcomes. The guide explains how those outcomes are checked and which limitations apply. The source of a result is available in the expanded pick row.
The algorithm code is proprietary. We publish its selected estimates and results while keeping the implementation, input transformations, weights, tuning, and calculation rules private. We aim to make the published record inspectable without distributing a recipe for generating it.
This distinction also limits what the public record proves: readers can assess the selections and their outcomes, but cannot independently reproduce or audit the private algorithm from this website.
CURRENT EVIDENCE STATUS
Exploratory research, clearly labeled.
SBOS research remains provisional. The public archive is a record of selected picks, not a complete evaluation of every candidate the engine considered. It does not establish calibrated probabilities, a causal Sunday effect, or a sustained betting edge.
Before presenting a model as validated, a research summary should identify the evaluation dates, sample size, comparison baseline, outcome definitions, and uncertainty. It should distinguish results collected after a selection was recorded from historical exploration. We do not substitute a hit count for that evidence.