Middles
Find middle opportunities where two sportsbooks have different lines, creating a gap where both bets can win.
GET /api/v1/opportunities/middlesAuthenticationPermalink for this section
Requires API key. Pro tier or higher required.
Query ParametersPermalink for this section
| Parameter | Type | Default | Description |
|---|---|---|---|
sport | string | all | Filter by sport(s), comma-separated (e.g. basketball, football). Sport ids only; a league id such as nba goes in league (details). |
league | string | all | Filter by league(s), comma-separated (e.g. nba, nfl, nhl) |
sportsbook | string | tier-allowed | Filter by sportsbook(s), comma-separated. Tier limits enforced. |
market | string | all | Filter by market type(s), comma-separated (e.g. point_spread, total_points). Supports category aliases — see Odds: Market Category Aliases. |
min_profit | number | — | Minimum ROI % — filters by each middle’s roi_percentage |
is_live | boolean | — | true = live only, false = prematch only, omit = both |
max_odds_age | number | — | Max odds age in seconds. Filters out stale opportunities. |
sort | string | quality | Sort field: quality (default), ev, probability, middle_size, event (start time), market (market id, natural order). Prefix - for descending and + for ascending (in a URL send %2B: a bare + is read as a space). With no prefix, quality, ev, probability and middle_size are best first, and event and market are ascending. Equal values are ordered by quality (best first), then by id. An unknown value sorts by quality and its prefix is ignored. |
limit | integer | 50 | Results per page (max 200) |
offset | integer | 0 | Pagination offset |
Example RequestsPermalink for this section
cURL
curl -X GET "https://api.sharpapi.io/api/v1/opportunities/middles?league=nfl&min_profit=1" \
-H "X-API-Key: YOUR_API_KEY"ResponsePermalink for this section
Success (200)Permalink for this section
Real response for GET /api/v1/opportunities/middles?league=mlb&is_live=false&sort=middle_size&limit=1 (captured 2026-07-02; alt_side2 truncated for brevity; each side’s deep_link, sent since 2026-11-15, added with an illustrative value):
{
"data": [
{
"alt_side1": [
{
"book": "kalshi",
"external_event_id": "KXMLBTOTAL-26JUL022210SDLAD",
"market_id": "KXMLBTOTAL-26JUL022210SDLAD-4",
"odds": -1329,
"selection_id": "KXMLBTOTAL-26JUL022210SDLAD-4_yes"
},
{
"book": "novig",
"external_event_id": "019f1eb1-8730-7851-aab2-7d433205d1fc",
"market_id": "019f20bb-5b05-7841-9c2c-e4fabf16bdcf",
"odds": -3471,
"selection_id": "019f20bb-5b05-7841-9c2c-e500de0ba6e2"
}
],
"alt_side2": [
{
"book": "kalshi",
"external_event_id": "KXMLBTOTAL-26JUL022210SDLAD",
"market_id": "KXMLBTOTAL-26JUL022210SDLAD-8",
"odds": 182,
"selection_id": "KXMLBTOTAL-26JUL022210SDLAD-8_no"
},
{
"book": "novig",
"external_event_id": "019f1eb1-8730-7851-aab2-7d433205d1fc",
"market_id": "019f1eb1-8767-7393-9a85-b234a45f6e58",
"odds": 170,
"selection_id": "019f1eb1-8767-7393-9a85-b25a26e03f2c"
}
],
"away_team": "San Diego Padres",
"best_case_profit": 89.75,
"break_even_percent": 5.4,
"detected_at": "2026-07-02T18:24:11Z",
"event_id": "mlb_dodgers_padres_2026-07-02_b3",
"event_name": "San Diego Padres @ Los Angeles Dodgers",
"expected_value": 28.32,
"guaranteed_roi": null,
"home_team": "Los Angeles Dodgers",
"id": "102f0c34a65c64b6",
"is_guaranteed_profit": false,
"is_live": false,
"is_player_prop": false,
"is_team_total": false,
"key_number_probability": 0,
"key_numbers": [],
"league": "mlb",
"league_label": "MLB",
"market_label": "Total Runs",
"market_overround": 1.054,
"market_type": "total_runs",
"middle_numbers": [
4,
5,
6,
7
],
"middle_probability": 0.3525,
"middle_size": 4,
"odds_age_seconds": 546,
"player_name": null,
"quality_score": 1.84,
"roi_percentage": 28.32,
"side1": {
"book": "polymarket",
"deep_link": "https://api.sharpapi.io/api/v1/deeplink/206720022460212?state=pa",
"external_event_id": "mlb_dodgers_padres_2026-07-02_b3",
"fair_probability": 0.655,
"line": 3.5,
"market_id": "pm_2763202",
"odds_age_seconds": 244.6,
"odds_american": -223,
"odds_decimal": 1.448,
"odds_probability": 0.6904024767801857,
"selection": "Over",
"selection_id": "2763202_over",
"selection_type": "over",
"stake_percent": 65.5
},
"side2": {
"book": "sx_bet",
"deep_link": null,
"external_event_id": "L18646245",
"fair_probability": 0.345,
"line": 7.5,
"market_id": "sx_0x5e051a0f483fc08b7b3d9d5893ca1970a90d693b8072be89ae272e5684be2c6d",
"odds_age_seconds": 546,
"odds_american": 175,
"odds_decimal": 2.75,
"odds_probability": 0.36363636363636365,
"selection": "Under",
"selection_id": "0x5e051a0f483fc08b7b3d9d5893ca1970a90d693b8072be89ae272e5684be2c6d_2",
"selection_type": "under",
"stake_percent": 34.5
},
"sport": "baseball",
"start_time": "2026-07-03T02:11Z",
"stat_category": null,
"team_name": null,
"warnings": [
"HIGH_PROBABILITY",
"LARGE_GAP",
"STALE_ODDS"
],
"worst_case_loss": -5.13,
"worst_case_pnl": -5.13
}
],
"pagination": {
"count": 1,
"has_more": true,
"limit": 1,
"next_offset": 1,
"offset": 0,
"total": 31
},
"updated_at": "2026-07-02T18:24:11.737562235Z"
}Response HeadersPermalink for this section
X-RateLimit-Limit: 300
X-RateLimit-Remaining: 296
X-RateLimit-Reset: 1707401000
X-Data-Delay: 0
X-Request-Id: 1782526326424224-82519Error ResponsesPermalink for this section
401 Unauthorized
No key supplied returns missing_api_key; a bad key returns invalid_api_key.
{
"error": {
"code": "missing_api_key",
"message": "API key required. Pass via X-API-Key header, api_key query parameter, or Bearer token.",
"docs": "https://docs.sharpapi.io/en/authentication#authentication-methods"
}
}403 Tier Restricted
{
"error": {
"code": "tier_restricted",
"message": "This endpoint requires Pro tier or higher. Upgrade to Pro for $229/mo: https://sharpapi.io/dashboard/billing/upgrade/pro",
"docs": "https://sharpapi.io/pricing",
"tier": "hobby",
"required_tier": "pro",
"upgrade": "https://sharpapi.io/dashboard/billing/upgrade/pro",
"upgrade_note": "Upgrade to Pro for $229/mo"
}
}Response FieldsPermalink for this section
| Field | Type | Description |
|---|---|---|
id | string | Unique middle identifier (hash) |
event_id | string | Event identifier |
event_name | string | Human-readable event name |
sport | string | Sport identifier |
league | string | League identifier |
market_type | string | Market type (point_spread, total_points, player prop types) |
market_label | string | Human-readable market name |
league_label | string | Human-readable league name |
home_team | string | Home team name |
away_team | string | Away team name |
start_time | string|null | Game start time (ISO 8601) |
side1 | object | First side of the middle bet |
side2 | object | Second side of the middle bet |
middle_size | number | Size of the middle gap in points |
middle_numbers | number[] | Integer values that would hit the middle |
middle_probability | number | Estimated probability of the middle hitting. Sport- and market-specific: a per-sport multiplicative calibration factor (backtested against settled outcomes and refreshed as data accrues, most recently 2026-07-02) is applied to the raw per-number sum, then clamped to a ceiling of 0.50. |
expected_value | number | EV in dollars per $100 total wagered |
roi_percentage | number | EV as percentage of total stake |
worst_case_loss | number | worst_case_pnl clamped to <= 0 — the downside when the middle misses (0 on guaranteed-profit middles, negative otherwise) |
worst_case_pnl | number | Raw worst-case profit-and-loss per $100 total stake. Positive on guaranteed-profit middles (arb + middle — the worst case still profits), negative otherwise. Equals worst_case_loss only when negative. |
best_case_profit | number | Profit when middle hits (both bets win) |
break_even_percent | number | Minimum hit rate needed to break even |
odds_age_seconds | number|null | Age of stalest underlying odds at detection time |
is_guaranteed_profit | boolean | True when worst_case_loss >= 0 (arb + middle) |
guaranteed_roi | number|null | Worst-case ROI% when guaranteed (null otherwise) |
key_numbers | number[] | Key numbers in middle range (e.g., NFL 3, 7) |
key_number_probability | number | Combined probability of key numbers only |
quality_score | number | Multi-factor quality metric (0-100) |
market_overround | number | Combined implied probability across both sides |
is_live | boolean | Whether the game is live: true when our game tracking shows the game in progress, or otherwise when the sportsbooks pricing it offer live prices. |
is_player_prop | boolean | Whether this is a player prop middle |
player_name | string|null | Player name (if player prop) |
stat_category | string|null | Stat type (if player prop, e.g., points, rebounds) |
is_team_total | boolean | Whether this is a team-total middle |
team_name | string|null | Team name (if team total) |
alt_side1 | object[] | Other books offering side 1’s selection at the same line — {book, odds, selection_id, market_id, external_event_id} per entry, sorted best price first. Empty array when no alternates. |
alt_side2 | object[] | Other books offering side 2’s selection at the same line (same shape as alt_side1) |
warnings | string[] | Warning flags (see below) |
detected_at | string | When the middle was detected (ISO 8601) |
Side ObjectPermalink for this section
| Field | Type | Description |
|---|---|---|
book | string | Sportsbook for this side |
selection | string | The selection name |
selection_type | string|null | Which side of an over/under middle this is: "over" on side1 and "under" on side2 for totals, team totals and player props. null on both sides of a spread-like middle (point spread, puck line, game spread, Asian and set handicaps, half and quarter spreads), where selection names the team. Always present. Same meaning as selection_type on an /odds row, limited here to over and under. Read the side from this field, not from selection: selection is the sportsbook’s own label, which can be the player’s name, a localized label or Yes/No on a prediction market. |
line | number | The line value at this sportsbook |
odds_american | number | American odds |
odds_decimal | number | Decimal odds |
odds_probability | number | Implied probability (0-1) |
fair_probability | number | Fair probability adjusted for overround |
stake_percent | number | Recommended stake allocation (% of total) |
odds_age_seconds | number|null | Age of this side’s odds at detection time |
selection_id | string|null | Sportsbook’s native selection/outcome ID |
market_id | string|null | Sportsbook’s native market ID |
external_event_id | string|null | Sportsbook’s native event ID |
deep_link | string|null | Redirect address for this side’s bet, https://api.sharpapi.io/api/v1/deeplink/{id}?state=XX — the Deep Link Redirect, built the same way as an arbitrage leg’s deep_link. The key is always present: null when the sportsbook gave no link. The ?state= suffix is the request’s state parameter, pa when omitted. |
home | object|undefined | Reserved — not currently emitted on this endpoint. Nested home-team reference: {id, numerical_id, name, abbreviation, logo, city, mascot, conference, division}. See Entity reference IDs. |
away | object|undefined | Reserved — not currently emitted on this endpoint. Nested away-team reference (same shape as home). |
sport_ref | object|undefined | Reserved — not currently emitted on this endpoint. Nested sport reference: {id, numerical_id, name}. |
league_ref | object|undefined | Reserved — not currently emitted on this endpoint. Nested league reference: {id, numerical_id, label}. |
market_ref | object|undefined | Reserved — not currently emitted on this endpoint. Nested market reference: {id, numerical_id, label}. |
sportsbook_ref | object|undefined | Reserved — not currently emitted on this endpoint. Nested sportsbook reference for this side’s book: {id, numerical_id, label}. |
Nested reference blocks on each side (reserved)Permalink for this section
The side schema reserves six nested reference objects (home, away, sport_ref, league_ref, market_ref, sportsbook_ref) mirroring the blocks on /odds. As of 2026-07-02 the middles engine does not stamp them, so live side1 / side2 objects never include these blocks — today they appear only on /odds rows for atlas-mapped books. Treat every block (and every inner field) as optional and never require them; when emitted they will be purely additive — the flat top-level fields (sport, league, home_team, away_team, market_type) and per-side book / selection strings are unchanged. See Entity reference IDs for the shape and guarantees.
Warning FlagsPermalink for this section
| Warning | Meaning |
|---|---|
HIGH_PROBABILITY | Calibrated middle probability exceeds 35% — at or beyond the highest empirically realized hit rate; treat as notable (and double-check the lines) |
LARGE_GAP | Middle size exceeds 3 points — verify lines are correct |
STALE_ODDS | One or both odds are older than 300 seconds (5 minutes) |
LIVE_GAME | Game is currently in progress |
GUARANTEED_PROFIT | Worst-case loss is positive — this is also an arbitrage |
Detail EndpointPermalink for this section
Fetch a single middle opportunity by its ID.
GET /api/v1/opportunities/middles/:id| Parameter | Type | Description |
|---|---|---|
id | path | Required. 16-character hex hash (e.g., a1b2c3d4e5f67890) |
Returns { "data": <middle>, "updated_at": "<ISO>" } — the middle uses the same schema as the list items. Returns 404 if the ID is unknown or malformed.
curl -X GET "https://api.sharpapi.io/api/v1/opportunities/middles/a1b2c3d4e5f67890" \
-H "X-API-Key: YOUR_API_KEY"What Are Middles?Permalink for this section
A middle occurs when two sportsbooks offer different lines on the same market, creating a numerical gap where both bets can win simultaneously.
Unlike arbitrage (which guarantees profit), middles give you a chance to win both sides of a bet if the final result lands in the “middle” of the two lines.
Spread Middle ExamplePermalink for this section
DraftKings: Chiefs -2.5 @ -110
FanDuel: Bills +7.5 @ -108You bet:
- Chiefs -2.5 at DraftKings
- Bills +7.5 at FanDuel
Possible outcomes:
| Chiefs Win By | Chiefs -2.5 | Bills +7.5 | Result |
|---|---|---|---|
| 1-2 points | Lose | Win | Split (small loss) |
| 3, 4, 5, 6, or 7 | Win | Win | MIDDLE HIT |
| 8+ points | Win | Lose | Split (small loss) |
If the Chiefs win by 3-7, both bets win. The numbers 3 and 7 are NFL key numbers (field goal and touchdown), making this middle especially valuable.
Key NumbersPermalink for this section
Certain sports have scoring increments that spike the probability distribution:
| Sport | Key Numbers | Why |
|---|---|---|
| NFL | 3, 7, 10, 14 | Field goal (3), touchdown (7), combinations |
| NHL | 1 | Most games decided by 1 goal (~47%) |
| MLB | 1 | Most games decided by 1 run (~30%) |
| Soccer | 1 | Most matches decided by 1 goal (~38%) |
When a middle contains key numbers, the key_numbers and key_number_probability fields highlight this. An NFL middle spanning 3 and 7 has a combined key number probability of ~24.2%.
Guaranteed Profit (Arb + Middle)Permalink for this section
When is_guaranteed_profit is true, the middle is also an arbitrage — the worst case still yields a profit. This is the rarest and most valuable type of middle. The guaranteed_roi field shows the minimum return even if the middle misses.
Middle Size GuidePermalink for this section
| Middle Size | Quality | Typical Markets |
|---|---|---|
| 0.5 points | Narrow | Rare opportunities, low hit rate |
| 1.0 points | Standard | Common in spreads and totals |
| 1.5 - 2.0 points | Good | Look for these actively |
| 2.5+ points | Excellent | Uncommon but highly profitable |
Sort by middle_size to surface the widest gaps first, and use min_profit for a minimum-ROI floor. Larger middles have a higher probability of landing in the gap, making them more consistently profitable over time.
Risk ConsiderationsPermalink for this section
Middles are not risk-free like arbitrage (unless is_guaranteed_profit is true). When the result falls outside the middle, you lose the vig on the losing side (typically 4-5% of one leg). However, when a middle hits, you win both bets, which more than compensates over time.
Odds FreshnessPermalink for this section
The odds_age_seconds field shows how old the underlying odds were when the middle was detected. Each side also has its own odds_age_seconds. If the stalest side exceeds 300 seconds (5 minutes), the STALE_ODDS warning is added. Stale odds may have already moved — act quickly on fresh opportunities.
Best PracticesPermalink for this section
- Focus on larger middles — Sort by
middle_sizeand prefer wider gaps for better hit rates - Watch for key numbers — Middles containing NFL 3/7 or NHL/MLB/soccer 1 have elevated hit rates
- Prioritize guaranteed profit — Filter for
is_guaranteed_profitto find risk-free opportunities - Check odds freshness — Low
odds_age_secondsmeans the lines are current and actionable - Combine with the summary endpoint — Use
/middles/summaryfor dashboard overviews - Act quickly — Lines converge as books adjust, closing the middle window
- Use streaming — Subscribe to
middles:detectedevents via SSE or WebSocket for real-time alerts
Related EndpointsPermalink for this section
- Middles Summary — Aggregate statistics without the full list
- +EV Opportunities — Positive expected value bets
- Arbitrage Opportunities — Guaranteed profit opportunities
- SSE Stream — Real-time
middles:detectedandmiddles:expiredevents - Best Odds — Compare odds across all books