January 5, 2026. Oklahoma City is a 15.5-point favorite at home against Charlotte. Chances are your whole group has the Thunder.
Charlotte wins 124-97.
Nights like that stick, which is why they wreck your picks. You remember the upset and forget the hundred favorites that rolled.
So we counted ten NBA seasons of spreads and results. Here's how often underdogs actually win.
Key findings on NBA underdogs
One in three, set by the spread
By the numbers
- 32.4% — Regular-season games won outright by the underdog (3,880 of 11,976) (Covers spreads + Basketball-Reference results, 2016-17 to 2025-26)
- 46.3% — Win rate for underdogs getting 2.5 points or fewer (Same sample, regular season)
- 9.0% — Win rate for underdogs getting 13 points or more (72 of 798) (Same sample, regular season)
- 34.8% — Home underdogs' win rate, against 30.9% for road underdogs (Same sample, regular season, neutral sites excluded)
- 41.2% — Underdog win rate in the 2026 playoffs, the highest of the ten postseasons (Same sample, 2025-26 playoffs (85 games))
The short version for your next NBA pick em sheet:
- Underdogs won 32.4% of regular-season games over ten seasons. Roughly one game in three.
- The spread tells you almost everything. Dogs of 2.5 or fewer won 46.3%, close to a coin flip. Dogs of 13 or more won 9.0%.
- Home underdogs won 34.8%, road underdogs 30.9%.
- Double-digit favorites still lost 15.1% of the time, 328 of 2,174 games.
- NBA upsets were barely more common in the playoffs overall, 33.3% against 32.4%. But the last four postseasons ran hotter, and 2026 hit 41.2%.
Here's the belief to drop: that picking favorites is the safe, boring way to win a pool. Favorites win about two of every three games. If most of your group is picking them too, the games that separate you are the ones where you went the other way.
For more on how NBA contest play has shifted, see Splash's take on NBA fantasy gaming trends.
How we built the sample
Every game, both sides, matched twice
Ten NBA seasons, 2016-17 through 2025-26. That's every regular-season, play-in and playoff game, plus three NBA Cup finals, 12,852 in all.
The point spread. We pulled every team's game-by-game results from Covers.com, which lists the point spread next to each final score. Every game shows up twice, once on each team's page, and the two spreads matched in every game listed on both pages. Three games had no spread on Covers and are left out.
Results and game type. Basketball-Reference's schedule pages supplied each game's final score, arena and game type. All 12,849 games matched.
A cross-check. TeamRankings publishes win records by closing point spread for every NBA game since 2003-04, playoffs included. Its long-run rates by spread size land within three points of ours in every bucket. Its NBA win trends data, with team records as favorites and underdogs, confirmed our 2025-26 team numbers.
Definitions used in this post
| Term | How we counted it |
|---|---|
| Underdog | The team getting points in the listed spread. Every game in the sample had one |
| Win outright | The underdog won the game, regardless of margin |
| Regular season | Games that count in the standings. Play-in games, playoffs and NBA Cup finals are counted separately |
| Home or road underdog | From the team's home or away marker. Neutral-site games, including the 2020 Orlando bubble, are left out of this split |
| Spread size | Grouped into buckets: 0.5 to 2.5, 3 to 4.5, 5 to 6.5, 7 to 9.5, 10 to 12.5, and 13 or more |
Opening or closing? Covers does not say whether its listed spread is the opening or the closing number, so we call it the listed spread. That is why the TeamRankings check matters: its numbers are labeled closing spreads.
Pushes and pick'em lines. Pushes don't apply: we count who won, not who covered, and NBA games can't end tied. A pick'em line has no underdog, so those games would drop out. Covers lists none here. TeamRankings' closing lines show 54 over the same seasons, another sign the two don't always list the same number.
These are NBA stats from finished games only. They describe the past, not tonight.
How often NBA underdogs win outright
Steady, with favorites gaining lately
Across 11,976 regular-season games, NBA underdogs won 3,880, or 32.4%. The rate barely moves year to year.
Regular-season underdog win rate by season
| Season | Games | Underdog wins | Win % |
|---|---|---|---|
| 2016-17 | 1,230 | 412 | 33.5% |
| 2017-18 | 1,230 | 389 | 31.6% |
| 2018-19 | 1,230 | 408 | 33.2% |
| 2019-20 | 1,059 | 365 | 34.5% |
| 2020-21 | 1,079 | 364 | 33.7% |
| 2021-22 | 1,230 | 392 | 31.9% |
| 2022-23 | 1,230 | 414 | 33.7% |
| 2023-24 | 1,228 | 380 | 30.9% |
| 2024-25 | 1,230 | 376 | 30.6% |
| 2025-26 | 1,230 | 380 | 30.9% |
The last three seasons were the sample's three lowest, all under 31%. Favorites have been a little more trustworthy lately, not less.
By NBA point spread size
The number is the whole story
In this data, the NBA point spread does nearly all the work. The bigger the number, the faster the underdog's chances fall off.
Underdog win rate by spread size
| Underdog getting | Games, 2016-17 to 2025-26 | Win %, regular season | Win %, TeamRankings since 2003-04 |
|---|---|---|---|
| +0.5 to +2.5 | 2,521 | 46.3% | 46.5% |
| +3 to +4.5 | 2,537 | 40.5% | 39.5% |
| +5 to +6.5 | 2,296 | 32.0% | 31.7% |
| +7 to +9.5 | 2,448 | 25.4% | 23.8% |
| +10 to +12.5 | 1,376 | 18.6% | 15.8% |
| +13 or more | 798 | 9.0% | 7.9% |
The TeamRankings column covers 2003-04 through 2025-26 and includes playoff games, so it is a check, not a twin.
Small underdogs are close to coin flips. A 5-point underdog wins about one time in three. Once the number reaches double digits, NBA underdogs win fewer than one game in five.
Underdog win % by spread size, regular season
| Underdog win % by spread size, regular season | Value |
|---|---|
| +0.5 to +2.5 | 46.3% |
| +3 to +4.5 | 40.5% |
| +5 to +6.5 | 32.0% |
| +7 to +9.5 | 25.4% |
| +10 to +12.5 | 18.6% |
| +13 or more | 9.0% |
Source: Covers spreads matched to Basketball-Reference results, 2016-17 to 2025-26 regular seasons
Home vs road underdogs
Same spread, same chances
Home underdogs won 34.8% of their games (1,554 of 4,460). Road underdogs won 30.9% (2,280 of 7,390).
That gap looks like home court at work. It mostly isn't. Road underdogs are more often big underdogs, because the spread already builds home court in. Line them up at the same spread and the difference nearly disappears.
Home and road underdog win % at the same spread, regular season
| Underdog getting | Home underdogs | Road underdogs |
|---|---|---|
| +0.5 to +2.5 | 46.6% | 46.3% |
| +3 to +4.5 | 41.0% | 40.0% |
| +5 to +6.5 | 31.8% | 32.0% |
| +7 to +9.5 | 24.7% | 25.8% |
| +10 to +12.5 | 17.9% | 18.7% |
| +13 or more | 7.3% | 9.4% |
So a 4-point home dog and a 4-point road dog are basically the same pick. Let the number decide, not the arena.
NBA favorites that lose most often
Some favorites deserve less trust
Over ten regular seasons, among teams that were favored in at least 200 games, Chicago lost the most often, 113 of 277 games (40.8%). Washington (40.2%) and Atlanta (39.6%) were next. At the other end, Philadelphia (27.3%), Oklahoma City (27.7%) and Boston (27.9%) held up best as NBA favorites.
Highest loss rate as a favorite, 2025-26 regular season (20+ games as favorite)
| Team | Games as favorite | Losses | Loss % |
|---|---|---|---|
| Chicago Bulls | 30 | 17 | 56.7% |
| Milwaukee Bucks | 29 | 15 | 51.7% |
| Dallas Mavericks | 24 | 12 | 50.0% |
| Golden State Warriors | 50 | 21 | 42.0% |
| Miami Heat | 41 | 15 | 36.6% |
| Los Angeles Clippers | 45 | 16 | 35.6% |
Chicago lost more often than it won as a favorite in 2025-26. TeamRankings' 2025-26 favorite records show the same Bulls, Bucks and Mavericks lines.
One season of 30 games is a flag, not a law. Question that team's spread before you hand it a pick.
Big favorites are safer, not safe. Teams favored by 10 or more lost 328 of 2,174 regular-season games, or 15.1%.
NBA playoff upsets vs regular season
Same rate overall, hotter lately
Playoff basketball feels different. On upsets, the totals say it isn't.
Excluding the 2020 Orlando bubble, underdogs won 250 of 750 playoff games, or 33.3%. The regular-season rate over the same seasons was 32.4%. Play-in underdogs won 11 of 37.
Playoff underdog win % by postseason
| Playoff underdog win % by postseason | Value |
|---|---|
| 2017 | 21.5% |
| 2018 | 23.2% |
| 2019 | 34.1% |
| 2020 (bubble) | 37.3% |
| 2021 | 34.1% |
| 2022 | 35.6% |
| 2023 | 34.5% |
| 2024 | 36.6% |
| 2025 | 38.1% |
| 2026 | 41.2% |
Source: Covers spreads matched to Basketball-Reference results, 79 to 87 playoff games per postseason
NBA playoff upsets have risen every postseason since 2023, from 34.5% that year to 41.2% in 2026. Over those four, underdogs won 126 of 335 games, or 37.6%. Back in 2017 and 2018, they won just over one in five.
Mind the sample. A postseason is about 85 games, so a few results swing the rate by several points. Notice the trend; don't rebuild your sheet around it.
Biggest upsets in NBA history
When the number lied
The biggest upsets in NBA history come in two kinds: single games where the spread was huge, and playoff series where a top seed went home early.
Start with series. A No. 8 seed has beaten a No. 1 seed six times.
No. 8 seeds that beat No. 1 seeds, with regular-season records from Basketball-Reference
| Season | No. 8 seed | No. 1 seed | Series |
|---|---|---|---|
| 1993-94 | Denver Nuggets (42-40) | Seattle SuperSonics (63-19) | Won in 5 |
| 1998-99 | New York Knicks (27-23) | Miami Heat (33-17) | Won in 5 |
| 2006-07 | Golden State Warriors (42-40) | Dallas Mavericks (67-15) | Won in 6 |
| 2010-11 | Memphis Grizzlies (46-36) | San Antonio Spurs (61-21) | Won in 6 |
| 2011-12 | Philadelphia 76ers (35-31) | Chicago Bulls (50-16) | Won in 6 |
| 2022-23 | Miami Heat (44-38) | Milwaukee Bucks (58-24) | Won in 5 |
The 2007 Warriors beat a 67-win Dallas team, 25 games better over the season. The 1999 Knicks and the 2023 Heat both went on to reach the NBA Finals. No eighth seed has done it since. The No. 1 seeds won every first-round series in 2024, 2025 and 2026.
Now single games. The biggest underdog to win outright in our sample was Brooklyn, at +18.5, beating Milwaukee 119-116 in the Orlando bubble on August 4, 2020.
Biggest underdog wins by listed spread, 2016-17 to 2025-26 (regular season, plus the biggest playoff win)
| Date | Underdog | Favorite | Spread | Final |
|---|---|---|---|---|
| Aug 4, 2020 | Brooklyn Nets | Milwaukee Bucks | +18.5 | 119-116 (neutral site) |
| Apr 9, 2025 | San Antonio Spurs | at Golden State | +17.5 | 114-111 |
| Apr 2, 2023 | Portland Trail Blazers | at Minnesota | +17 | 107-105 |
| Apr 2, 2023 | San Antonio Spurs | at Sacramento | +16.5 | 142-134 |
| Mar 10, 2019 | Phoenix Suns | at Golden State | +16.5 | 115-111 |
| Feb 5, 2026 | Washington Wizards | at Detroit | +16.5 | 126-117 |
| May 21, 2017 | Boston Celtics | at Cleveland | +15.5 | 111-108 (playoffs) |
That last one is the biggest playoff underdog win in the sample, Game 3 of the 2017 Eastern Conference finals.
Going back further, TeamRankings' closing-line records since 2003-04 show one win by a 19-point underdog, in 2022-23, and no wins by anyone getting 19.5 or more. Underdogs of 18 points or more won 2 of 89 games over that stretch.
Use the data in your NBA pick em predictions
Pick the spots, not the feelings
Good NBA pick em predictions start with the spread and end with the field.
Take favorites in most games. They win about two out of three. A sheet full of upsets is a sheet full of losses.
Pick your underdogs from the close games. Dogs of 2.5 or fewer won 46.3%. If you need to separate from your group, that's the cheapest place to do it.
Skip the long shots unless you're desperate. Dogs of 13 or more won 9.0%. Taking one is a late-season, need-to-catch-up move.
Don't pay extra for home court. At the same spread, home and road underdogs won at nearly identical rates.
Think about the field. In a small group, you rarely need to take chances. In a big one, a well-chosen small underdog that most players fade is how you gain ground.
Tip
Building your NBA pick em today? Check injury and rest news right before tip. A late scratch can move a spread more than anything in this post.
Looking for NBA pick em on Splash? Straight answer: Splash's team Pick'em is a football game. You call every NFL matchup against the spread or straight up, and a weekly tiebreaker settles ties. It doesn't run NBA team slates.
Basketball on Splash is a different game, the Splash QuickPicks game. Instead of picking winners, you call More or Less on player stat lines. An entry needs two or more picks from at least two teams, and you're up against other players, not the house.
Contests are open in most of the US and Canada, though availability and minimum age vary by state and province. Check where you can play first.
Know the numbers. Pick your spots. Let the rest of your group chase the upset they remember.
Frequently asked questions
Quick answers on NBA underdogs
About one game in three. From 2016-17 to 2025-26, regular-season underdogs won 3,880 of 11,976 games, or 32.4%, using the spreads Covers lists matched to Basketball-Reference results. TeamRankings' closing-spread records since 2003-04, playoffs included, put the long-run rate at 31.5%.
Yes, more than you'd think. Teams favored by 10 or more points lost 328 of 2,174 regular-season games over ten seasons, or 15.1%. Underdogs getting 13 or more won 72 of 798, or 9.0%. The biggest underdog win in the sample was Brooklyn at +18.5 over Milwaukee in 2020.
Yes, selectively. Small underdogs of 2.5 points or fewer won 46.3% of the time, close to a coin flip, so they are the cheapest way to separate from a group that takes every favorite. Big underdogs rarely win, so save them for when you need to catch up.
Not in the regular season. The three seasons from 2023-24 to 2025-26 were the lowest in the sample, between 30.6% and 30.9%. The playoffs went the other way: underdogs won 37.6% of playoff games over the last four postseasons, peaking at 41.2% in 2026.
Any game the underdog won outright, meaning the team getting points in the listed spread won the game. Margin does not matter. A 1-point underdog's win counts the same as a 15-point underdog's, which is why the spread-size table matters more than the headline rate.

