National Basketball Association
NBA Analytics & Game Predictions
A nightly NBA board built on pace, efficiency and availability: AI game breakdowns, market-derived win probabilities, player and team stats, and a graded record of every read.
Season note. The NBA regular season runs from October to April, with the playoffs and Finals through June. Times on this page are UTC.
NBA Game Analysis
NBA lines move more on availability than on anything else, so the breakdown reads the injury and rest report first: who is out, who is on a minutes restriction and whether either side is on the second night of a back-to-back. From there it compares pace and offensive and defensive rating to explain why the spread and total sit where they do.
What the breakdown reads
- availability and injury report
- pace and efficiency
- rest and back-to-backs
- market lines and movement
- head-to-head history
Markets covered
- point spread
- moneyline
- game total
Each breakdown carries a data-quality tier — rich, partial or weak — based on how much of the list on the left the feed actually delivered for the game. The tier is passed to the language model so the write-up hedges when the package is thin, and it is the first thing to check when a read looks off. Read more about how the AI analysis is produced.
AI NBA Predictions
The win probability on every NBA game page is derived from the market, not from the language model. The prices are converted into implied probabilities, the bookmaker’s margin is removed so the set sums to 100%, and the result is shown as each side’s fair win chance.
NBA lines move late and often, and the probability on the page follows the current line: a page that shows 58% at noon may show 61% at tip-off after a scratch. The breakdown itself is pre-generated for the slate and cached, so it is the number, not the write-up, that tracks the move — which is why a read is worth checking against the line it was written on.
What the AI adds is the explanation, not the number. The breakdown reads the same data package the statistics section shows — form, availability, head-to-head, rest and the line, as far as the basketball feed provides them — and writes out why the market sits where it does and what it may be underweighting. A data-quality tier is attached — rich, partial or weak, depending on how much of that package the feed delivered — and the read is published before the game starts. It is graded from the official result afterwards, and nothing is edited after the fact. AI breakdowns are pre-generated for the upcoming slate and cached; the win probability on the card is derived from the current market line. The write-up is not rewritten when the line moves.
A probability is a statistical estimate. A 65% favourite is expected to lose roughly one game in three; a month in which it loses more than that is not evidence that the number was wrong, and a month in which it loses fewer is not evidence that it was right. The prediction-tracking table further down this page and the Pick Tracker exist so that the method can be judged on volume, not on memory.
The full board — every game on the slate with its probability and status — lives on the NBA predictions page.
NBA Statistics
The NBA statistics section is organised around the categories the breakdown actually reads. Every number comes from the public scoreboard and box-score feeds and the per-game data package built on game day; where a category is not provided for a competition, the page says so rather than showing a blank.
Offensive rating
Points scored per 100 possessions — the pace-adjusted view of an offence that the breakdown uses instead of raw points per game.
Defensive rating
Points allowed per 100 possessions, so a slow, stingy team and a fast, leaky one are compared on the same scale.
Pace
Possessions per 48 minutes for both teams; the expected pace of the matchup is the first input to the total.
Net rating and recent form
Offensive minus defensive rating over the season and the last ten games — where the breakdown looks for a team playing better or worse than its record.
Home and road splits
Record and net rating at home versus on the road, because the home effect in basketball is real and varies widely by team.
Rest situation
Days since the last game and back-to-back flags for both sides — one of the inputs that most often explains a line that looks off.
NBA Matchups
The NBA plays enough games that recent form is a real signal — the last ten games say more than the season record. Head-to-head results are used for style context (the pace of previous meetings, how a matchup handled a particular defence) rather than as a predictor, and rest is treated as a first-class input because back-to-backs measurably change scoring.
Form, head-to-head and rest are inputs to the breakdown, not the verdict. The market line is the anchor, and the matchup context is what the analysis uses to explain agreement or disagreement with it — a read that contradicts the line has to say which of these inputs the market appears to be missing, and by how much.
Every game page shows the matchup view in full: the participants’ recent results, the head-to-head record where one exists, the schedule context and the public line, drawn from the same data package the AI reads. Start from the NBA games list or from today’s board below.
NBA Team Performance
Team profiles carry the crest, abbreviation and display name and refresh daily from the schedule feed. Record, net rating, recent results and home and road splits are not stored on the profile; they are assembled on each game page from that night’s data package, so the numbers behind the breakdown sit next to the line they explain.
Team profiles are rebuilt from the schedule feed every day, but they hold identity, not statistics: the crest, the abbreviation and the display name that keep a team consistent across game pages, the pick tracker and the breakdown. The performance numbers — record, recent results, the next scheduled game and the market’s number for it — are assembled on each game page from the per-game data package, because they only mean something next to an opponent and a date.
Two things the performance view does not do. It does not project a team’s next result; that is the breakdown’s job, and it is graded. And it does not backfill history the feed never delivered: a team with a short record is shown with a short record. The NBA team directory lists every team on the board with its profile, and each card opens that team’s upcoming games.
NBA Player Statistics
Points, rebounds, assists, minutes and three-point volume feed the breakdown’s read of the matchup — a starter on a minutes restriction changes the line more than any team statistic — and are shown where the feed provides them. Player props are covered for MLB only: there are no NBA prop reads, and the published NBA board read is the moneyline.
Points, rebounds and assists
Per-game production and recent-window averages for rotation players — the base numbers behind any discussion of who carries the offence when a starter is out. Read as matchup context; player props are covered for MLB only.
Minutes and usage
Minutes played and usage rate, read together because a minutes restriction or a shifted role changes a team’s output before any skill number does. The breakdown notes both when the availability report moves them.
Three-point volume
Attempts and makes per game; volume shifts with rotation changes, and the breakdown notes when it does.
NBA Prediction Tracking
Published NBA reads settled in each of the last 3 calendar months (UTC). The published NBA read is the moneyline favourite, graded on the final score including overtime. Postponed games are removed from the record rather than counted.
| Month | Settled | W-L-P | Hit rate | ROI |
|---|---|---|---|---|
| September 2026 | 0 | 0-0-0 | — | — |
| August 2026 | 88 | 59-29-0 | 67% | -6.5% |
| July 2026 | 71 | 49-21-1 | 70% | -3.7% |
Hit rate is wins divided by wins plus losses; pushes are excluded. ROI is measured at a flat stake per read, so one large result cannot dominate a month. 3 months is a small sample — read it as a window into the method, not as a forecast of the next month — and remember that past results do not predict future results.
Today’s NBA Games
Every NBA game on today’s board (UTC) in start-time order, with the public spread, total and moneyline where the feed provides them. The NBA regular season runs from October to April, with the playoffs and Finals through June.
There are no NBA games on today’s board (UTC). Either the schedule has a gap — common in every sport — or the feed has not returned the day yet; the board refreshes several times a day and fills in on its own.
Upcoming games appear in the next section as soon as the feed lists them, and the NBA games page keeps the full schedule window.
Upcoming NBA Games
Games scheduled over the next 3 days on the public feed, after today. Lines are shown as soon as the market posts them; games without a line yet are listed with their start time only.
Nothing further is listed for NBA in the next 3 days on the public schedule feed. The NBA regular season runs from October to April, with the playoffs and Finals through June.
The NBA games page shows the widest schedule window available, and the sports index lists which boards have events today.
NBA Analytics FAQ
How does Sportslyx analyze an NBA game?
The breakdown starts from the market line, then layers availability, rest, pace and efficiency numbers and recent form. The output is a market-derived win probability, an AI-written matchup summary and a data-quality tier; the read is graded after the final buzzer.
Does the NBA analysis account for injuries?
Yes. The availability report is part of the data the model reads before writing a breakdown, and a late scratch is one of the main reasons a line moves.
Which NBA stats matter most for predictions?
Net rating, pace, rest and availability explain more of the spread than raw points per game. The Stats section shows the numbers the model actually weighs.
Can I see how past NBA predictions performed?
Every settled read appears on the public Pick Tracker with hit rate and ROI by month.
Is this betting advice?
Sportslyx produces statistical estimates and analytical breakdowns for information and entertainment. Nothing on this page is betting advice, and no outcome is certain.
NBA Breakdowns for Every Game — 7 Days on Us
Full Pro access: the AI breakdown and market-derived probability for every NBA game on the board, the daily reads, Combo Builder, Studio and the pick tracker. A card is required, nothing is charged today, and you can cancel before the trial ends.
7-day trial · $0 today · card required · cancel anytime before the trial ends
Sportslyx produces statistical estimates and analytical breakdowns for information and entertainment. Nothing on this page is betting advice, and no outcome is certain. 18+ only. Please play responsibly — see our Responsible Play page.