A complete guide to understanding every feature, every number, and every term. No experience needed — we'll break it all down.
ORACLE is a free sports prediction tool that analyzes player performance data from multiple sources to help you make smarter decisions about player props. It compares odds from 8+ sportsbooks, runs a 7-factor prediction model, and shows you where the data says the value is.
Think of it like this: instead of guessing whether LeBron will score over 25 points tonight, ORACLE looks at his last 68 games, checks if key teammates are injured, sees how fast tonight's opponent plays, factors in the day of the week and whether it's a home game, and tells you there's a 72% chance he goes over — with a Grade A confidence rating.
No signup, no payment, no hidden costs. PickFinder charges $15/mo. Sharp App charges $49/mo. ORACLE is $0 forever.
Not just averages — we factor in injuries, pace, Vegas lines, rest days, recent form, venue, and regression patterns.
Finds bets where the math is in your favor. The same technique professional bettors use to profit long-term.
Compares lines from DraftKings, FanDuel, BetMGM, Caesars, Bovada, BetRivers, PrizePicks, and Underdog side by side.
A player prop (short for "player proposition") is a bet on whether a specific player will go OVER or UNDER a specific number for a stat. It's not about who wins the game — it's about individual player performance.
| Market | What it means | Example |
|---|---|---|
| Points | Total points scored | Brunson Over 27.5 pts |
| Rebounds | Total rebounds grabbed | Towns Over 10.5 reb |
| Assists | Total assists recorded | Brunson Over 6.5 ast |
| 3-Pointers | Three-point shots made | Curry Over 4.5 threes |
| PRA | Points + Rebounds + Assists combined | Jokic Over 45.5 PRA |
| Steals | Steals recorded | Edwards Over 1.5 stl |
| Blocks | Shots blocked | Wembanyama Over 3.5 blk |
The line is the number set by the sportsbook — it's the dividing line between OVER and UNDER. When ORACLE shows "27.5", that's the line. The .5 exists to prevent a "push" (tie) since you can't score half a point.
Every term you'll see on ORACLE, explained simply.
The average line across all sportsbooks. If FanDuel says 27.5 and DraftKings says 28.5, the consensus is 28.0. This represents what the market "thinks" the number should be.
How many sportsbooks are offering this prop. More books = more data to compare = more confident in the line. If only 1 book has it, be cautious. If 6+ have it, the line is well-established.
The difference between the highest and lowest line across books. A spread of 0 means all books agree. A spread of 3+ means books disagree significantly — this is where edges hide.
Negative odds (like -110) mean you risk that amount to win $100. So -110 means bet $110 to win $100.
Positive odds (like +150) mean you win that amount on a $100 bet. So +150 means bet $100 to win $150.
The further from -100/+100, the more the book is leaning one direction.
The percentage of games where a player would have gone OVER a specific line. If Brunson's hit rate for Over 27.5 pts is 62%, that means he scored 28+ points in 62% of his games this season.
The player's average over their last 5 or last 10 games. This shows recent form. If L5 is much higher than the season average, the player is on a hot streak.
How differently a player performs at home vs. on the road. Some players score 3-5 more points at home due to crowd energy, familiarity, and less travel fatigue.
A player's average performance on each day of the week. Surprisingly, some players consistently score more on Fridays and Saturdays (national TV games, higher energy) than on Tuesday mid-week games.
When a team plays two games on consecutive days. Players typically score 2-4 fewer points on the second night due to fatigue. ORACLE detects B2B games and adjusts projections down.
The sportsbook's built-in profit margin. Standard vig is -110 on both sides, meaning you risk $110 to win $100. The sportsbook keeps $10 regardless of outcome. ORACLE's +EV engine removes the vig to find the "fair" probability.
A bet where the math is in your favor long-term. If the true probability of winning is higher than what the odds imply, you have +EV. This is how professional bettors make money — they don't need to win every bet, they need to consistently find +EV spots.
The #1 metric pro bettors track. It measures whether you got a better number than where the line closed at game time. If you took OVER 27.5 and the line closed at 26.5, you have +1.0 CLV — you got an easier number to hit. Consistently beating the closing line is the single best predictor of long-term profitability.
When a line moves in the opposite direction of where the public is betting. Example: 80% of bets are on the OVER, but the line moves UP (making OVER harder to hit). This means sharp money — big professional bets — is coming in on the UNDER, overriding the public volume. RLM is the #1 way sharps find edges.
When a line moves 1+ point across multiple sportsbooks within minutes. This signals large coordinated sharp money hitting the market. Steam moves happen fast — by the time most people notice, the value is already gone. ORACLE detects these in real-time.
A mathematical formula that tells you the optimal bet size based on your edge. Instead of betting the same amount every time, Kelly sizes bets proportionally — bigger when your edge is bigger, smaller when it's marginal. Most pros use "quarter Kelly" (25% of the formula's suggestion) for safety.
A parlay where the legs are statistically connected. If Game Total goes OVER, Player Points is more likely to go OVER too (more scoring = more individual stats). Smart parlays use positive correlations. Bad parlays combine uncorrelated or negatively correlated legs.
The process of removing the sportsbook's profit margin (vig) from the odds to find the "true" fair probability. If a book offers -110 on both OVER and UNDER, the vig makes both sides slightly negative EV. Devigging reveals the actual 50/50 line. ORACLE uses the Power method — the same math used by professional syndicates.
ORACLE doesn't just look at averages. It combines 10 different factors to generate a projection for every prop. The first 7 are the foundation, and the 3 newest factors (Minutes, Game Script, Opponent History) are what push accuracy to the next level. Here's each one:
Your last 3 games matter more than a game from 2 months ago. ORACLE gives 50% weight to the last 3 games, 30% to games 4-10, and only 20% to the rest of the season. This catches hot and cold streaks faster than a simple average.
When a key teammate is OUT, the remaining players see a usage spike. If Karl-Anthony Towns is injured, Brunson's scoring projection goes UP because he'll handle the ball more and take more shots. ORACLE tracks 85+ injuries across 28 NBA teams and automatically adjusts.
The Vegas game total tells you how many points are expected in the game. A total of 235 = fast, high-scoring game = more stats for everyone. A total of 208 = slow grind = fewer stats. ORACLE also checks the spread — if a team is favored by 15+, starters may sit the 4th quarter (fewer minutes = lower stats).
Different teams play at different speeds. The Indiana Pacers play at 104 possessions per game (fast). The Cleveland Cavaliers play at 97 (slow). More possessions = more opportunities to score. ORACLE calculates the expected game pace by averaging both teams' pace ratings.
When a teammate is out, their "usage" (percentage of team plays they were involved in) gets redistributed to the remaining players. Points are most affected (1:1 boost), assists moderately (0.6x), rebounds least (0.3x). This catches the "X is out so Y will eat" plays that casual bettors miss.
If a player is performing 1.5+ standard deviations above their average (hot streak), they're statistically likely to cool off. If they're 1.5+ below (cold streak), they're likely to bounce back. ORACLE detects these extremes and adjusts the projection toward the player's true average.
Players perform differently at home vs. away. Some have a +5 point home advantage. ORACLE checks whether tonight's game is home or away and applies a 15% weighted adjustment based on the player's venue-specific history.
Minutes played is the #1 predictor of player stats — more minutes = more stats, period. ORACLE projects expected minutes by checking the Vegas spread (15+ point favorites = starters sit Q4, losing 4-6 minutes), back-to-back games (starters lose ~2.5 min from fatigue), and teammate injuries (remaining players absorb more minutes). A player projected for 28 min instead of their usual 35 means roughly 20% fewer stats — this single adjustment prevents the most common reason picks miss.
The Vegas game total and spread predict the entire game environment. A total of 235 with a 1.5 spread = fast, competitive shootout where starters play full minutes and everyone's stats are boosted. A total of 210 with a 12 spread = slow blowout where starters sit early. ORACLE calculates separate multipliers for points, rebounds, and assists based on the predicted game script.
Some players consistently dominate certain opponents. Brunson might average 25 points overall but 31 points vs the Celtics — that's a +6 opponent-specific advantage. ORACLE checks the player's game log for previous matchups against tonight's opponent and applies a 20% weighted adjustment when there's a meaningful historical pattern (2+ game sample, 2+ point difference from season average).
This is the single most important concept in sports betting. Most people think you need to pick winners 70% of the time to profit. That's wrong. You need to find bets where the odds are in your favor — that's called Positive Expected Value (+EV).
Imagine a coin flip. It's 50/50. If someone offers you +150 odds on heads (bet $100 to win $150), you should take it every time — even though you'll lose half the time. Over 100 flips, you'll win about 50 × $150 = $7,500 and lose about 50 × $100 = $5,000. Net profit: $2,500. That's +EV.
Now imagine the opposite: someone offers -200 on heads (bet $200 to win $100). Even though heads hits 50% of the time, you'll lose money over time. That's -EV.
ORACLE uses a technique called "devigging" — removing the sportsbook's profit margin from the odds to find the true probability of an outcome. Then it compares that fair probability to each sportsbook's odds. If a book is offering better odds than the fair probability suggests, that's a +EV bet.
You don't need to win every bet. Over 100 bets with a 5% average edge, you'll profit significantly even if you only win 55% of them. This is how professional bettors think.
These are the exact techniques professional sports bettors use daily. Sharp App charges $49/mo and OddsJam charges $99/mo for these features. ORACLE gives them to you free.
Professional bettors don't obsess over their win percentage — they track CLV. If you consistently get better numbers than where the line closes, you will profit long-term. Period. It doesn't matter if you lose tonight's bet.
ORACLE snapshots every prop line every 15 minutes. When the game starts, it compares the line you could have taken vs. the closing line. If you bet early and the line moved in your direction, you had CLV — you got value the market hadn't priced in yet.
Reverse Line Movement is the most powerful signal in sports betting. It tells you where professional bettors are putting their money — even when they're going against the crowd.
ORACLE detects RLM by comparing the direction of line movement against the juice (which indicates where public money is flowing). When the line moves opposite to public sentiment, sharp money is on the other side.
Most people bet the same amount on every play. Pros don't. They use the Kelly Criterion to size bets proportionally to their edge. A 5% edge gets a bigger bet than a 2% edge. A 0% edge gets $0.
ORACLE calculates Kelly for every +EV bet it finds. It recommends quarter-Kelly (the safest option) with a hard cap at 5% of bankroll to prevent overexposure.
Most parlay builders treat each leg as independent — but they're not. If the game total goes OVER, individual player points are more likely to go OVER too. That's a positive correlation (r=0.72). ORACLE identifies which props move together so you can build smarter parlays.
Top correlations to use:
| Combo | Correlation | Why It Works |
|---|---|---|
| Game Total OVER + Player Points OVER | +0.72 (strong) | More total scoring = more individual scoring |
| Close spread + Starter stat OVER | +0.55 | Close games = full 4th quarter minutes |
| Key player OUT + Teammate OVER | +0.62 | Usage redistributes to remaining players |
| Fast pace + OVER rebounds | +0.58 | More shots = more rebound opportunities |
Avoid: Both opposing stars OVER points (-0.30 correlation). If one team dominates, the other star typically underperforms.
A steam move happens when a line shifts 1+ points across multiple sportsbooks within minutes. This means coordinated professional money is hitting the market. ORACLE scans for steam moves every 15 minutes and alerts you to the direction sharp money is flowing. Act fast — steam value disappears quickly as books adjust.
The main view. Shows all available player props across 8+ sportsbooks, compared side by side. Click any row to expand and see every book's line + odds. Filter by stat type (Points, Rebounds, Assists, 3PT, PRA) or by signal type (Demons, Goblins, Edges). Each prop is enriched with hit rate, L5/L10 averages, and confidence scoring.
ORACLE's best picks of the day. The 18-factor prediction model analyzes every prop, scores them by confidence (0-95%), and surfaces the top 8 with the strongest signals. Each pick shows the projection, hit rate, reasoning, and a letter grade (A+ through C). Updated every 15 minutes.
Props ranked by "trending score" — a composite of multiple signals converging on the same prop. A prop gets a high trending score when it has Demon detection (high book consensus), line movement (lines shifting in one direction), AI confidence, and edge detection all pointing the same way.
Tracks every pick ORACLE makes and grades them against final box scores. Shows your hit rate over time, broken down by confidence tier, stat type, and pick direction (OVER vs UNDER). This is how you track whether the model is actually performing.
Live and upcoming games from ESPN. Shows start times, broadcast channels, and team matchups. Click any game to get an AI prediction for the winner.
Same game list as Scores, but focused on game outcome predictions. Click "AI Predict" to get a detailed analysis of each game including win probability, key matchups, and scoring projection.
Current season standings pulled from ESPN. Shows wins, losses, win percentage, and conference/division rankings for all teams.
ORACLE detects three types of line anomalies that represent different kinds of opportunities:
A Demon is detected when 6+ sportsbooks agree on a line AND there's a 2+ point spread between books. This means the market has strong consensus but at least one book is offering a significantly different number — that outlier book is where the value is. Demons are the highest-conviction plays.
A Goblin is a warning signal. It appears when there's a 3+ point line spread between books, suggesting the market is confused or a sharp bettor has moved one side aggressively. Goblins can be traps set by sportsbooks. Approach with caution.
An Edge is found when the line spread between books is 1.5+ points. One book's number is significantly different from the others, creating a mathematical edge for bettors who take the right side at the right book.
A Smash Spot appears when a player's position faces a defense ranked in the bottom 10 (worst) for allowing that stat. For example, if the Hornets allow the 3rd most points to point guards, and Brunson is a PG playing the Hornets tonight — that's a Smash Spot. Target these aggressively.
| Sport | Data Source | Features |
|---|---|---|
| 🏀 NBA | ESPN + Odds API | Props, scores, standings, DvP, predictions, +EV |
| 🏈 NFL | ESPN + Odds API | Props, scores, standings, predictions |
| ⚾ MLB | ESPN + Odds API | Props, scores, standings |
| 🏒 NHL | ESPN + Odds API | Props, scores, standings |
| ⚽ EPL / La Liga | ESPN | Scores, standings |
| 🏀 NCAAM / 🏈 NCAAF | ESPN | Scores, standings |
| 🎮 CDL | PandaScore + BreakingPoint | Match predictions, per-map kill props, standings |
| 🎯 Valorant / 💣 CS2 | PandaScore | Matches, standings, predictions |
| ⚔️ LoL / 🛡️ Dota2 | PandaScore | Matches, standings, predictions |
ORACLE doesn't just make picks — it tracks whether they actually hit. Every 30 minutes, the auto-grader checks finished games and compares our picks against the final box scores.
| Grade | Confidence | What it means |
|---|---|---|
| A+ | 80%+ | Strongest possible signal. Multiple factors aligned. Target aggressively. |
| A | 70-79% | Strong pick. Most factors agree. High conviction. |
| B+ | 60-69% | Good pick. Solid edge detected. Worth consideration. |
| B | 55-59% | Decent pick. Some factors align. Moderate conviction. |
| C+ | 50-54% | Marginal edge. Proceed with caution. |
| C | Below 50% | Weak signal. Skip or reduce bet size. |
You can test the model's historical accuracy for any player using the backtest endpoint. It runs the prediction model against every game in the season and reports how often it would have been right. This is how we validate and improve the model over time.
The Sharp Dashboard at /sharp is a dedicated page for serious bettors. It puts all of ORACLE's most powerful tools in one place. Here's what each tab does:
Shows every bet that has positive expected value right now. ORACLE scans all props across all sportsbooks every 10 minutes, removes the vig to find the true probability, and flags any book offering better odds than the math says they should. Each card shows the exact edge percentage, how much you'd make per $100 bet, which sportsbook has the edge, and the optimal Kelly bet size. Best during pre-game hours (11 AM - 7 PM ET) when lines are most active.
Shows where professional bettors are putting their money — by detecting when lines move in the opposite direction of where the public is betting. If 80% of people bet OVER but the line moves UP (making OVER harder), that means big sharp money is on UNDER. Each alert shows the public side, the sharp side, the movement amount, and how strong the signal is. This is the single most reliable signal in sports betting.
Catches moments when a line moves 1+ point across multiple sportsbooks within minutes. This happens when coordinated professional money hits the market simultaneously. Each alert shows the direction, how fast it moved, how many books shifted, and which side the sharp money is on. Act fast when you see these — the value disappears as books adjust.
An interactive calculator that tells you exactly how much to bet based on your bankroll, your edge, and the odds. Enter your numbers and it instantly shows the recommended bet amount, your risk percentage, potential win/loss, and whether the edge is strong enough to bet on. Uses quarter-Kelly by default (the safest option), with a 5% bankroll cap to prevent overexposure. No more guessing bet sizes.
A personal bet logging system. Enter every bet you make — player, market, line, direction, odds, amount — and track your results over time. Mark each bet as Hit, Miss, or Push. The tracker automatically calculates your total profit/loss, hit rate, and shows your complete betting history. All data is saved in your browser so it persists between visits. This is how you know if you're actually profitable.
Shows which props tend to hit together (positive correlation) and which ones work against each other (negative correlation). Use this to build smarter parlays. The strongest correlation: game total OVER + player points OVER (r=0.72). The biggest trap: betting on both opposing team stars to go OVER points (r=-0.30). Includes 5 actionable parlay tips.
Shows all lines being tracked for Closing Line Value. ORACLE snapshots every prop line every 15 minutes. After games start, it compares where the line was when you could have bet vs where it closed. If the line moved in your direction after you bet, you had CLV — the market confirmed your read. This is the single best predictor of long-term profitability according to professional bettors.
A "middle" is when two different sportsbooks have different lines for the same prop, creating a gap where BOTH bets can win. For example, Book A has OVER 25.5 and Book B has UNDER 28.5 — if the player gets 26, 27, or 28, both bets win. ORACLE scans every prop across all books and flags middling opportunities with the gap size, which books to use, and the exact window where both bets hit. Even if the middle misses, you only lose the vig on one side.
An arbitrage (arb) is a guaranteed profit. It happens when two sportsbooks disagree on odds so much that you can bet both sides and make money no matter what. ORACLE calculates the exact stakes for each side to guarantee equal profit. Arbs are rare and disappear fast — when you see one, act immediately. The tab shows the exact bet amounts, which books to use, and your guaranteed profit on $100.
Sometimes one sportsbook sets a line that's way off from what every other book has. For example, if the consensus is OVER 27.5 but one book has OVER 24.5 — that book is giving you a 3-point easier number for the same bet. ORACLE finds these outlier lines automatically and tells you exactly which book has the value, which direction to bet, and how many points easier it is compared to the market.
Visit /parlay for the full interactive experience. Five tabs in one page:
Add legs one by one and ORACLE automatically checks correlation between every pair. Same-game legs in the same direction (both OVER) have positive correlation — this actually increases your real probability above what the naive math suggests. Opposing stars both going OVER is negatively correlated — avoid it. You get combined odds, payout per $100, adjusted probability, EV calculation, and a letter grade. Grade A = positively correlated + positive EV.
Real-time scoreboard for every game today. Shows LIVE with quarter and clock, or FINAL with scores. Refreshes every 30 seconds so you can track your bets as they happen.
ORACLE's complete public track record. Every pick is graded against final box scores. See total hits/misses, hit rate overall and for the last 7 days, broken down by grade (A+ through C), by market (Points, Rebounds, Assists), and by direction (OVER vs UNDER). Full transparency — you can audit everything.
Generate a beautiful branded card for any pick — shows player name, direction, line, confidence, and grade on a dark ORACLE-branded background. Screenshot it and post to Twitter or Instagram. Free marketing that builds community.
Set thresholds for conditions you care about: "+EV edge ≥ 5%", "POTD convergence ≥ 80%", "hit rate ≥ 70%", or "middle gap ≥ 3 points." Alerts are stored in your browser and checked against incoming data.
ORACLE uses 18 distinct factors to generate projections — more than any tool on the market at any price. Here's every single one:
| 1. Weighted Recency | Last 3 games = 50%, games 4-10 = 30%, rest = 20% |
| 2. Recent Form (L5) | Average of last 5 games — hot/cold streak detection |
| 3. Hit Rate | % of games where player exceeds the specific line |
| 4. Injury Impact | Teammate injuries redistribute usage to remaining players |
| 5. Vegas Environment | Game total + spread predict scoring pace |
| 6. Pace/Tempo | Team pace ratings (possessions per 48 min) affect all stats |
| 7. Regression to Mean | Extreme hot/cold streaks revert toward the average |
| 8. Home / Away Venue | 15% weighted adjustment based on venue-specific history |
| 9. Consistency Score | Low-variance players are more predictable → higher confidence |
| 10. Minutes Projection ✦ | #1 predictor — spread (blowout risk), B2B fatigue, teammate injuries |
| 11. Game Script ✦ | Total + spread → shootout vs defensive battle vs blowout |
| 12. Opponent History ✦ | Player's historical stats vs tonight's specific opponent |
| 13. Pace × Usage ✦ | High usage + fast pace = exponential stat boost |
| 14. Rest Days ✦ | 0 days = -4%, 2 days = +2%, 3-5 days = +4% |
| 15. Garbage Time Filter ✦ | Blowout = starters sit Q4, bench gets 25% boost |
| 16. Weather (NFL/MLB) ✦ | Cold, wind, rain impact outdoor sports stats |
| 17. Travel Distance ✦ | Coast-to-coast (2000+ mi) = -3% fatigue penalty |
| 18. Defensive Efficiency ✦ | Position-specific DvP — pts allowed per 100 possessions |
✦ = New factors added in the latest update
A 55% win rate at +EV odds makes more money than a 70% win rate at bad odds. Check the +EV tab and prioritize bets with 3%+ edge.
Don't bet every pick. Focus on the highest-confidence selections where multiple factors (injuries, pace, trend, hit rate) all point the same direction.
ORACLE auto-adjusts for injuries, but check the injury report anyway. A star player going from "Questionable" to "Out" right before game time changes everything.
Research shows rebounds are less variable than points and assists. Rebounding props are the easiest to predict because they correlate strongly with position, minutes, and matchup. If you're new, start here.
If a team is favored by 12+ points, starters may sit the entire 4th quarter. That's 12 minutes of stats gone. ORACLE flags blowout risk — if you see it, lean UNDER on the favorite's starters.
ORACLE shows you every book's line. ALWAYS take the best available. If FanDuel has Over 27.5 at -110 and DraftKings has Over 26.5 at -110, DraftKings is strictly better. One point can be the difference between hitting and missing.
The History tab tracks everything. After a week, check your hit rate by confidence tier. If Grade A+ picks are hitting at 65%+, your process is working. If not, adjust your filters.
Ready to start finding edges?
Open ORACLE →