Knicks vs Spurs Game 3 Preview: The Signal Stack Picks the Underdog

New York hosts Game 3 up 2-0. Three venues favor them; both our signal layers lean the Spurs.

GhostLeg, 2026-06-08, 7 min

Data in this preview reflects a snapshot from Sunday, June 7 at 8:12 PM ET. Lines and signals refresh roughly every 30 minutes; the live intel dashboard carries the current state of every number cited below.

New York walks into Game 3 favored. Our signal stack is on the underdog.

The Knicks stole both games in San Antonio. Game 1 was a 10-point road statement (105-95). Game 2 was a one-point escape (105-104) that felt more like a possession-level street fight than a blowout, but a win is a win. New York comes home up 2-0, riding a 10-game win streak with a +17.7 average margin across that run. Madison Square Garden awaits on Monday, June 8 at 8:40 PM ET.

Now the interesting part: the three biggest probability-setting venues on the planet are within 1.1 percentage points of each other on the Knicks side. The cross-venue alignment is tight. Inside that consensus, our signal stack is leaning a different way.

The signals behind that read are available programmatically through the GhostLeg Data API.

What the h2h market shows

The moneyline consensus has New York as the home favorite at -130, available at four of the seven books in our feed. Pinnacle is the friendliest number for Knicks backers at -125. BetMGM is the most expensive at -135.

The Spurs side is mostly flat across books. Five of seven post San Antonio at +110, with Pinnacle nudging up to +111 and FanDuel at +112. If you want the Spurs, FanDuel and Pinnacle pay a touch more than the rest. If you want the Knicks, Pinnacle's -125 saves you a meaningful amount of juice over BetMGM's -135 on the same wager.

GhostLeg's composite score for the h2h sits at 61 on a 0-100 signal-strength scale, and the lean is on San Antonio, not New York. The composite stack weights line movement, sharp action, prediction-market consensus, and historical patterns; on net it reads the underdog as the better play even with the form gap. The XGBoost model pushes harder in the same direction. Its raw output for this game puts the Spurs at a 57 percent win probability, which is well above the market's roughly 44 percent implied for SAS at +110. On the calibrated 1-99 edge display, that lands at 65, comfortably above the neutral midpoint of 50.

When both signal layers agree on the underdog while three pricing venues converge on the favorite, the read is direct: our stack thinks the market is overweighting the Knicks' recent form. The intel dashboard surfaces both layers side by side so you can see exactly how they line up.

Pinnacle's implied probability on the Knicks is 55.6 percent. Since the open it has drifted 0.68 percentage points lower, with Pinnacle leading the move by 120 minutes ahead of the rest of the market. The drift is small, but the direction matches our model: the sharpest book in the world has been quietly easing off the Knicks all afternoon.

The prediction markets weigh in

Kalshi has the Knicks at 55%. Polymarket has them at 54.5%. Pinnacle sits at 55.6%.

The maximum gap across all three venues is 1.1 percentage points. GhostLeg's DGI summarizes it plainly: "San Antonio Spurs at New York Knicks: Sportsbooks, Kalshi, and Polymarket are all aligned within 1pp — efficient market."

That convergence is worth sitting with. When three independent price-discovery mechanisms, a regulated US exchange, a crypto-native global market, and the sharpest sportsbook in the world, all land within a single percentage point, it tells you the public consensus, the sharp consensus, and the prediction-market consensus have all arrived at essentially the same number. The exploitable divergence in this game is not between those venues. It is between those venues and our signal stack.

For background on how Kalshi and Polymarket prices relate to traditional sportsbook lines, the Kalshi vs Polymarket explainer covers the conversion math and fee structures in depth.

Polymarket's total volume on this market is $6.41M across two sides, with 42 snapshots per side in GhostLeg's feed. Kalshi shows deep, balanced two-sided liquidity across 43 snapshots per side. Volume on both platforms is competitive, which reinforces the efficient-market read for the venues themselves. Thin markets produce noisy prices. This is not a thin market.

Whale activity: split with high confidence

GhostLeg's whale tracker aggregates large-wallet behavior from Polymarket. For this game, the read is: split.

1,936 total wallets are active. 992 wallets on one side, 944 on the other. Dollar volume leans modestly toward the numerically-smaller wallet pool ($225K vs $161K), but the confidence tag on the consensus is high and the direction is listed as split. The gap is narrow enough that this does not resolve into a directional call. Monitor the whale feed for any late-session shift before tip. Track it live in the intel dashboard's whale tracker.

A note on perspective: the h2h Polymarket market for this game has a majority of unresolved perspective snapshots in GhostLeg's data. That means we cannot confirm with certainty which team the "Yes" and "No" pools map to. We report wallet counts and dollar volume only, not a directional conclusion.

Spread and total

The spread has moved since open. NYK's cover probability slipped 1.85 percentage points, and several books have rolled the line from NYK -2.5 to NYK -2 as a result.

That half-point matters in a series this tight. Game 2 was decided by a single point. The best price on -2.5 is BetMGM at -102, near coin-flip juice. The best price at -2 is Pinnacle at -108. The choice is a half-point of cover difficulty versus six cents of juice. Pick your spot.

The spread composite is 58, leaning New York -2.5. That is a different read from the h2h composite, which leans the Spurs outright. Spread and moneyline composites can split when the model sees a competitive game rather than a clear winner; that is likely what is going on here.

On the total, the Over is the lean with a composite score of 60. The line sits at 215.5 to 216.5 depending on the book.

DraftKings and FanDuel both offer Over 216.5 at -105, tied for the best Over price in the market if you are willing to lay the extra half-point. Pinnacle's -106 at 216 is the tightest juice on the sharper number. The Over implied probability at Pinnacle is 51.5 percent. The line is flat on net since open, but Pinnacle led the only movement (120 minutes ahead of the rest of the market), so any future drift will likely come there first.

What the team form data shows

The Knicks are running hot by any recent-form measure. 10-0 across the last 10, a +17.7 point average margin, and 8-1 against the spread in that same window. The is_hot flag is active. A 10-game win streak going into a home game after two road wins is as clean a momentum narrative as the data produces.

The Spurs enter on a two-game losing streak, 2-3 over the last five, with a +2.2 average margin in that stretch. Their last 10 is 5-5 and 5-4 ATS, which is not terrible, but it contrasts sharply with where New York is right now.

The market knows all of this. The 10-game win streak and two road wins are not hidden information. The NYK price at -130 (not -160 or -180) tells you the market is weighting regression risk, home-court adjustment for San Antonio, and series variance. Our composite sees those same factors but reads them differently: the small spread (NYK -2 to -2.5) and the model's 57 percent on the Spurs both suggest "modest favorite" is closer to a coin flip than a sustained-form premium would imply.

The line movement layer

Understanding why Pinnacle moved first matters. When Pinnacle leads a line move by 120 minutes, that typically signals sharp, early action rather than public betting patterns. The move was modest and directional toward San Antonio. Notably, it ran in the same direction as our composite and our model output. Sharp money plus our stack pointing the same way is a coherent signal: three independent sources all suggesting the Spurs side has more in it than the market price implies.

For a deeper look at how closing line value and line movement mechanics work together, the CLV explainer is worth a read before tip.

For live line movement on this game and tonight's full NBA slate, the GhostLeg intel dashboard tracks price changes in real time. The Data API carries the same opening and closing snapshots for any quantitative work.

Game 3 at MSG: the summary read

Three pricing venues within 1.1 percentage points of each other on the Knicks side. A modest sharp-led drift toward the underdog. A home team on a 10-game streak that the market is calling a modest favorite. A spread that has rolled from -2.5 toward -2 as books calibrate.

The cross-venue read is honest: there is no structural mispricing between sportsbooks, Kalshi, and Polymarket. The divergence is one layer up. Our composite (61) leans the Spurs, our XGBoost model lands the Spurs at 57 percent win probability (well above the market's ~44 percent implied), and the calibrated edge display sits at 65 on the 1-99 scale. Three venues see a Knicks favorite; both of our signal layers see something closer to a coin flip with a Spurs edge. That is the bet to track.

Watch the spread. The roll from -2.5 to -2 across most books, and BetMGM's -2.5 at -102 outlier, are the most actionable pieces of the board. If you believe this game is as close as the model suggests, the spread is where it matters more than the moneyline.

Follow the live signal feed in the NBA intel dashboard and pull the data behind it through the GhostLeg Data API.

For entertainment purposes only. Past performance does not indicate future results.