How Round Robin Sports Algo works
We build picks for fantasy and prediction markets the way a careful trader would: start from what the market already knows, look for where our data adds real lift, and pass when it doesn't.
Who builds it
Round Robin Sports Algo is proprietary. It was built by a data professional with 20 years in the fintech data industry, and spent three years in development, tested, backtested and retrained through many iterations, before the models went live. Picks are built from a consensus of models with calibrations that adjust as the season moves. It is not a chatbot answering questions.
- Consensus, not one opinion: each pick draws on several independent models and signals, weighed together rather than one opinion.
- Dynamic calibration: each model's confidence is checked against real results and recalibrated as the season moves.
- Walk-forward testing: a rule is tested on games the model never saw before it is allowed to go live.
What goes into a pick
- Form and role: recent stat lines, playing time, usage and depth-chart changes.
- Matchups: the opposing pitcher or defense, ballpark, weather and game script.
- Availability: injuries and inactives, and how a teammate's absence shifts opportunity.
- The market: the live price on Kalshi and the books. It is our starting point, not something we ignore.
Free, Members, All-Access
Free picks are our highest hit-rate calls, each one bettable at a standard line. Members get the full card per game plus value plays, which hit less often by design and are judged in units. All-Access adds game picks: spreads and moneylines when our signals agree.
How we grade
Every pick is locked with its price when it posts. Props settle at the official result (Kalshi's settlement for NFL props, the box score for MLB); spreads and moneylines at the final score. Records read "X of N": picks won out of picks graded. Backtests are labelled and kept apart.
Questions
What is Round Robin Sports Algo?
A picks service for fantasy players and prediction-market traders. AI models read hundreds of data points per player every game and we publish only the picks that clear our bar, then grade every one in public.
How are picks graded?
Player props settle at the official result: Kalshi's settlement for NFL props and the official box score for MLB. Spreads and moneylines settle at the final score. Each pick is graded at the price we posted, wins and losses.
Is this a chatbot or an off-the-shelf AI tool?
No. The models are proprietary, built over three years by a data professional with 20 years in the fintech data industry. Picks come from a consensus of models with in-season recalibration.
What is a backtest?
A replay of the model on past games using only data available before each game. Backtests are labelled on every page and never mixed into the live record.
What do members get?
Free picks are our highest hit-rate calls. Members get the full card for each game, value plays priced to win over a season, and anytime TDs for NFL; the top tier adds game picks. Plans are on Winible.
Do you post every day?
We post every MLB slate and every NFL kickoff window, but we don't force picks. If nothing clears our bar, the edition says so.
Picks are informational, not betting advice. 21+. Please play responsibly. Gambling problem? Call 1-800-GAMBLER.