Same-game parlay

Default HTTP is POST /v7/cs2/sgp/quote. Full route list: Route Index.

Combine several outcomes from one CS2 match into a single price. Every leg is evaluated on the same simulated match scenarios, so the price reflects how those outcomes move together instead of pretending they are independent bets.

Choose a match, pick at least two legs, send them to /v7/cs2/sgp/quote, receive the combined price, then record the ticket and read its settlement. Builder plan and above; pre-match only.

POST/v7/cs2/sgp/quote

Param	Type	Required	Description
match	string	Yes	CS2 kr_match_id.
legs	array	Yes	2 or more legs: {market, selection, map, maps, line, direction}.
pricing.source	string	-	model (default) or book. book takes leg levels from live prices; correlation stays the model’s.
pricing.books	object	-	book → weight, e.g. {"pinnacle": 1}. Weights are relative, not percentages.
pricing.min_sources	int	-	Books that must price a leg on both sides. Default 1.
operator.margin_pct	float	-	0-100. Applied to the joint probability, not per leg.
operator.max_legs	int	-	2-8. Default 8.
operator.allowed_markets	array	-	Restrict which markets a ticket may use.
operator.max_decimal	float	-	Price cap. A capped price says price_limited.
operator.price_override_decimal	float	-	Fixed price override.
operator.quote_ttl_seconds	float	-	1-600. Sets expires_at.
scenarios	int	-	Joint tape size, 1,000-100,000. Default 20,000.
seed	int	-	Deterministic seed. Same seed and legs give the same price.

Legs are match_winner, map_winner, total_maps, total_rounds and player_kills. Team markets take a kr_tm id as selection; line markets take direction: over|under plus a half-point line; kill legs take a kr_pl id plus maps (for example [1, 2] for kills across maps 1 and 2).

cURLPOST

curl -X POST -H "X-API-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  "https://kashrock.up.railway.app/v7/cs2/sgp/quote" \
  -d '{"match":"kr_cs2_bojong-vs-sinqu-04-11-2026","legs":[{"market":"match_winner","selection":"kr_tm_87b794cec83a"},{"market":"map_winner","selection":"kr_tm_87b794cec83a","map":1}]}'

200 · live (trimmed)

{
  "available": true,
  "sport": "cs2",
  "kr_match_id": "kr_cs2_bojong-vs-sinqu-04-11-2026",
  "best_of": 3,
  "quote_id": "sgq_1f4c0b2a9d6e71",
  "legs": [
    {
      "index": 0,
      "market": "match_winner",
      "selection": "kr_tm_87b794cec83a",
      "model_probability": 0.47595,
      "resolves": "series_winner"
    },
    {
      "index": 1,
      "market": "map_winner",
      "selection": "kr_tm_87b794cec83a",
      "map": 1,
      "model_probability": 0.4847,
      "resolves": "map_1_winner"
    }
  ],
  "pricing": {
    "source": "model",
    "fair_decimal": 2.789,
    "fair_american": 179,
    "priced_decimal": 2.789,
    "priced_american": 179,
    "margin_pct": 0,
    "model_joint_probability": 0.35855,
    "priced_probability": 0.35855,
    "independence_probability": 0.230693,
    "correlation_ratio": 1.5542,
    "correlation_delta": 0.127857
  },
  "simulation": {
    "version": "cs2-sgp-joint-v1",
    "method": "measured_joint_scenarios",
    "n_scenarios": 20000,
    "seed": 0,
    "standard_error": 0.003391,
    "p_ct": 0.5,
    "p_t": 0.4917,
    "map_name": "Ancient",
    "team_side_sample": 58,
    "live_conditioned": false
  },
  "in_play": {
    "supported": false,
    "reason": "no_live_conditioning"
  }
}

Reading the price

model_joint_probability is the share of simulation.n_scenarios scenarios in which every leg won together. independence_probability is what multiplying the individual probabilities would have given; correlation_ratio is the joint divided by that product. Above 1 the legs help each other, below 1 they work against each other. In the example above the model joint is 35.9% against an independence product of 23.1%: a 1.55x ratio and fair_decimal 2.789 instead of 4.335.

fair_decimalis the model's price, which is not a sportsbook's executable ticket price. With operator.margin_pct the returned priced_decimal is the model joint with that margin applied. With pricing.source: "book"the leg levels come from the named books' live two-sided devigged prices and only the correlation comes from the model; pricing.source always says which one the number is.

What is refused, and why

A refused ticket returns available: false with a reason and one entry per bad leg in unsupported_legs. Whole-number lines are refused because they can push. Maps that are only played if the series reaches them (map 3 at 1-1 on a Bo3) are refused because such a leg voids rather than loses. Contradictory legs, duplicate outcomes, markets the operator disallows, players with no stored kill sample, stale or suspended prices, and matches that have already started all come back with their own reason rather than a quietly different price.

Record and settle

POST /v7/cs2/sgp/ticket takes the same body plus an optional stake. The price is recomputed server-side and stored with the legs, quote id, simulation version and seed, so a caller cannot record a price of its own choosing. GET /v7/cs2/sgp/ticket/{ticket_id} returns the ticket with every leg marked hit, miss, void, pending or unresolved, the actual value it settled on, and the final ticket status. Settlement reads KashRock's stored match, map and player-map rows, so it is the same data the result pages show. A leg that must be voided on an otherwise-live correlated ticket is reported as needing a requote rather than silently repriced.

cURLPOST

curl -X POST -H "X-API-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  "https://kashrock.up.railway.app/v7/cs2/sgp/ticket" \
  -d '{"match":"kr_cs2_bojong-vs-sinqu-04-11-2026","stake":10,"legs":[{"market":"match_winner","selection":"kr_tm_87b794cec83a"},{"market":"map_winner","selection":"kr_tm_87b794cec83a","map":1}]}'