NBA 2K27 vs 2K26 Shooting
Ask anyone who has played both and they will tell you that shooting in NBA 2K27 is far harder than it was in NBA 2K26. We had the same reaction. So we did the one thing we can do about it: we ran the same green window test in both games, one millisecond at a time, on a station that takes the human hand out of the release.
The window did shrink. It did not shrink by anywhere near as much as it feels like it did, and the distance between those two things is what this page is about. If you only read one section, read the one on why it feels harder than it measures.
What 2K has said so far
Mike Wang, the gameplay director on NBA 2K, has said more about 2K27 shooting in its first two weeks than in most full seasons. Four of his posts frame everything below. The first settles what kind of system this is.
@SSJerdo… can you confirm if it's green or miss again or is it back RNG with a pure green window attached
Mike Wang @Beluba“That clip was captured with a hack to make all shots go in. Shooting in 2K27 is green or miss on all difficulties with a pure green window just like last year.”
That is the same model as last year: there is a pure green window, hit it and the shot goes in, miss it and the shot does not. It is what makes the test on this page possible, and it is also why the second post matters so much.
Mike Wang @Beluba“There's no RNG factor that will make you randomly miss shots. If you release your shot perfectly every time, you will make every shot. Your shot speed can vary based on a number of different factors (stamina, defensive pressure, the catch, pulling up off dribble vs. triple threat, sig shot grades…”
Read the second half of that carefully, because it is the list of everything that changes your shot from one attempt to the next: stamina, defensive pressure, whether you caught the ball or pulled up, and the grades on your jumper. Each of those can move the window on the clock, and several of them, contests and stamina in particular, also shrink it. Our test holds every one of them still, which is the difference between the window you can measure and the window you feel in a game. The third post is his read on where percentages will land.
@TV_BulletIf your having trouble shooting on NBA 2K27 try changing your jumpshot speed and mix it up with your badge selection. Combining different things together might be what your missing
Mike Wang @Beluba“this is good advice. you also might need to reset your expectations. if you're shooting over 40% from 3, you're already doing well. very few are going to be able to get close to last year's shooting percentages.”
And the fourth is about the one shooting setting you choose yourself, which comes back up further down.
Mike Wang @Beluba“If you're using a custom jumper, the release speed setting will have a small impact on your green window. Very Quick will reduce your green window size by 10% and Quick by 5%. Slow and Very Slow will actually give you a tiny boost. So choose what's more important for your…”
Then, on 30 August 2026, he posted the community's actual shooting numbers for both games side by side. That table is reproduced in full in its own section below, because it is the best evidence anyone has for how much harder 2K27 has been in practice, and it lines up with what we measured in a way that is worth walking through.
What the community numbers say
Mike Wang @Beluba“2K26 vs. 2K27 community shooting stats to date in case you're interested.”
Here is the table from his data. These are every shot taken by everyone, with nothing filtered out, four days into early access. Comp 5v5 is ProAm and Proving Grounds in 2K26 and Crew HQ in 2K27. The last column is the three point drop as a share of the 2K26 figure.
| Mode | 2K26 FG% | 2K27 FG% | 2K26 3PT% | 2K27 3PT% | 3PT change |
|---|---|---|---|---|---|
| Park | 55.2% | 53.5% | 45.6% | 41.2% | -4.4 pts (-10%) |
| Comp 3v3 | 49.1% | 48.9% | 37.6% | 29.8% | -7.8 pts (-21%) |
| REC | 52.8% | 48.0% | 44.5% | 36.4% | -8.1 pts (-18%) |
| Comp 5v5 | 43.9% | 47.8% | 35.1% | 29.7% | -5.4 pts (-15%) |
He also broke out players at 85 overall and above in 2K27, which is the closest thing to a like for like with a built up 2K26 player:
| Mode | 2K27 FG%, 85+ OVR | 2K27 3PT%, 85+ OVR | vs 2K26 overall 3PT% |
|---|---|---|---|
| Park | 59.4% | 48.7% | +3.1 pts |
| Comp 3v3 | 52.8% | 38.0% | +0.4 pts |
| REC | 51.5% | 41.2% | -3.3 pts |
| Comp 5v5 | 51.1% | 35.8% | +0.7 pts |
Mike Wang @Beluba“Casual Corner numbers: FG%: 73.8% | 3PT%: 66.9%”
How we tested, and why that matters this year
Shots were taken on an automated testing station that holds the shot button for a set number of milliseconds and then releases it. The release timing is fixed by the device rather than judged by a human hand, and the result of every shot is read off the timing feedback on screen. Nothing in the chain has a reaction time.
Each game was tested in 1 millisecond increments from a 541ms hold to a 590ms hold, 100 attempts at every step, 50 steps per game, 10,000 shots in all. Both tests used the same build, the same jumpshot and the same spot on the floor, standing catch and shoot, wide open, with no badges, hot zones, takeover or boosts in play, on a full stamina bar, offline. The only thing that changed between the two tests was the game.
This year the controls are not a footnote, they are the point. Go back to Wang's list of what changes your shot speed. Every item on it changes the shot in a way the station has to hold still: some move the window on the clock, some shrink it, and some do both. A comparison that lets any of them vary is measuring the mix rather than the base window:
- Stamina. A tired player's jumper is slower than a fresh one's. Both tests ran on a full bar, so fatigue could not stretch one game's timing and not the other's.
- Defensive pressure. A contest shrinks the window and can change the animation speed. There was no defender on the floor for either test. Contests get their own section below, because they are one of the places 2K27 really has changed.
- The catch, and pulling up against triple threat. Off the catch, off the dribble and out of a standstill are three different releases. Every shot here was the same standing catch and shoot.
- Sig shot grades and release speed. The jumper's grades set both how wide the window is and how fast the animation runs. Same jumper in both games, on the same release speed setting, so neither could tilt the comparison. What the release speed setting is worth on its own is measured on the Release Speed Test.
- Latency. Online, a variable delay between your press and the game registering it does the same thing as a wandering window: it scatters your release. Both tests ran offline on a wired setup with the hold issued at the controller, so no network delay was on the clock in either game.
There is a built in check that the setup really was the same in both games. Two tests run on the same jumper should put the centre of the window in the same place. The centres landed at 563.1ms and 562.3ms, 0.8ms apart, which is inside a single sampling step. If the jumper, the spot or the setting had differed between the runs, that gap would have been tens of milliseconds and this page would not exist.
One more thing about reading the results, since it is not obvious. Because there is no RNG, a cell that made 62 out of 100 is not a 62% shot. It is the share of attempts on which the window happened to cover that millisecond, because the release animation does not land in exactly the same place every rep. That means the whole curve, not just its peak, is the window, and its width is what the two numbers at the top of the next section are measuring.
What we found
The pure green window for this shot is ~35.3ms wide in 2K27 against ~39.2ms in 2K26, measured as the full width at half maximum of each curve. The area under each curve, which is the same width read a second way, gives 31ms and 34.2ms, a gap of 3.2ms, and it runs a little low on both because neither test was taken all the way out to zero. Either way the window lost about 10% of itself between the two games.
Where the window lost that width is more interesting than how much it lost. Split the hold times into six stretches and compare the two games in each:
| Stretch | Hold | 2K26 | 2K27 | Difference |
|---|---|---|---|---|
| Early tail | 541 to 549ms | 54.1% | 48.7% | -5.4% |
| Early shoulder | 550 to 558ms | 80.2% | 67.3% | -12.9% |
| Peak | 559 to 569ms | 96.7% | 92% | -4.7% |
| Late shoulder | 570 to 579ms | 68.5% | 68.8% | +0.3% |
| Late tail | 580 to 585ms | 52% | 38.8% | -13.2% |
| Very late | 586 to 590ms | 29.8% | 24.2% | -5.6% |
| Whole Test | 541 to 590ms | 68.4% | 62% | -6.4% |
- The peak is nearly the same. Across the 11 cells at the top of the window, 2K26 makes 96.7% and 2K27 makes 92%. A well timed shot goes in just as it did last year. That is the part everyone can feel is unchanged, and it is why the complaint is never about perfect releases.
- Both shoulders lost, the early one most. The early shoulder is down 12.9 points and the late tail is down 13.2 points. Split the hold times at the centre and the early half is 8.9 points behind against a margin of 2.1, the late half 4.4 behind against 1.9. The window got narrower on both sides rather than sliding one way.
- The always green core halved. In 2K26, the 8ms stretch of hold times from 561 to 568ms went in on 95 attempts out of 100 or better. In 2K27 that stretch is 4ms, from 564 to 567ms. This is the number that matches how the game feels, and the next section explains why.
The two windows, side by side
Both games on one pair of axes, in real hold time. If 2K27 were dramatically harder to shoot in you would expect one curve to sit far inside the other. What you see instead is the 2K27 curve tucked a few milliseconds inside the 2K26 curve on each flank, meeting it at the top, with a shorter stretch along the ceiling.
Why it feels harder than it measures
So the window is a tenth smaller and everyone feels like they are shooting twice as bad. Both of those are true, and here is how a 3.9ms cut turns into the season everyone is having.
- A tenth of the window is a tenth of your makes. Your timing is a spread, not a number, and for most players it is wider than the window, so makes scale with width. A 10% narrower window turns a 45% shooter into a 40% one. That is the Park column in Wang's table almost exactly: down 10%, from 45.6% to 41.2%, in the mode with the least defence.
- The part that never missed is half the size. All 3.9ms came off the edges, so the stretch that greened every attempt went from 8ms to 4ms. Consistent shooters live in that core. A release 6ms off perfect was a certainty last year and is a coin flip now.
- Every speed up and slow down costs more. Stamina, pressure, the catch, off the dribble, sig grades: each moves your release by more than the 3.9ms the window lost, and some shrink the window as well. A 39.2ms window absorbed a lot of that drift. A smaller one lets more of it through, on every shot. Our test held all of it still, which is why the test says 10% and the box score says more.
- Contests are stronger, and they compound. A contest takes a share of the window, not a fixed amount, and it slows the shot. A share of a smaller window is smaller still. The proof is in the next section: the modes with real defence lost far more than Park did.
The window alone is worth about 10% of your makes. Contests and drift each take a share on top, and shares multiply. Carry all of them and you are closer to a fifth of your makes gone, which is what the REC and comp columns show, from a window that moved 3.9ms.
Shot contests are doing more of the work
2K27 shipped an overhauled shot contest system. 2K's own gameplay report describes a Defensive Impact Indicator under the shooter that grades a contest by how hard it was and which direction it came from, from a small white arc up to a large crimson one, and Wang's early access briefing put eight tiers on it:
- Wide Open
- Open
- Semi-Open
- Bothered
- Light Contest
- Solid Contest
- Heavy Contest
- Smothered
Our read after two weeks in the game is that contests are a little stronger than they were, in two ways. A contest of the same apparent quality knocks more off the shot than it did in 2K26, and the middle tiers, Bothered through Solid Contest, cover a lot of shots that would have registered as open last year. If you feel like you are getting contested more, you probably are, because the grading got finer and the arc now tells you about hands that used to go unmentioned.
The evidence for it that does not rely on feel is in Wang's own table. The base window is the same in every mode, so if the window were the whole story every mode would have dropped by about the same 10%. They did not. Park, the mode with the least defence, dropped 10% from three. REC dropped 18%, Comp 3v3 dropped 21% and Comp 5v5 dropped 15%. The extra loss shows up exactly where the defenders are, which is what you would expect if contests are taking a bigger bite this year and if the shot speed changes they cause are landing on a window with less room in it.
We have not yet tested contests so we cannot tell you how many milliseconds each tier removes. That test is on the list. Until it is done, treat this section as what it is: what the game looks like in play, and what the mode by mode numbers point at.
The raw numbers
Every cell we recorded, unsmoothed, so nothing above has to be taken on trust. 100 attempts each.
| Hold | 2K26 | 2K27 | Difference |
|---|---|---|---|
| 541ms | 44% | 42% | -2% |
| 542ms | 45% | 43% | -2% |
| 543ms | 47% | 45% | -2% |
| 544ms | 53% | 47% | -6% |
| 545ms | 52% | 50% | -2% |
| 546ms | 58% | 49% | -9% |
| 547ms | 61% | 52% | -9% |
| 548ms | 63% | 54% | -9% |
| 549ms | 64% | 56% | -8% |
| 550ms | 75% | 59% | -16% |
| 551ms | 77% | 61% | -16% |
| 552ms | 74% | 58% | -16% |
| 553ms | 79% | 62% | -17% |
| 554ms | 82% | 68% | -14% |
| 555ms | 83% | 71% | -12% |
| 556ms | 84% | 72% | -12% |
| 557ms | 84% | 75% | -9% |
| 558ms | 84% | 80% | -4% |
| 559ms | 91% | 82% | -9% |
| 560ms | 94% | 85% | -9% |
| 561ms | 96% | 90% | -6% |
| 562ms | 98% | 92% | -6% |
| 563ms | 99% | 94% | -5% |
| 564ms | 100% | 96% | -4% |
| 565ms | 99% | 95% | -4% |
| 566ms | 98% | 98% | 0% |
| 567ms | 97% | 96% | -1% |
| 568ms | 98% | 94% | -4% |
| 569ms | 94% | 90% | -4% |
| 570ms | 84% | 82% | -2% |
| 571ms | 77% | 76% | -1% |
| 572ms | 69% | 74% | +5% |
| 573ms | 69% | 69% | 0% |
| 574ms | 69% | 68% | -1% |
| 575ms | 66% | 65% | -1% |
| 576ms | 65% | 66% | +1% |
| 577ms | 63% | 64% | +1% |
| 578ms | 62% | 62% | 0% |
| 579ms | 61% | 62% | +1% |
| 580ms | 58% | 48% | -10% |
| 581ms | 58% | 44% | -14% |
| 582ms | 55% | 38% | -17% |
| 583ms | 48% | 35% | -13% |
| 584ms | 46% | 36% | -10% |
| 585ms | 47% | 32% | -15% |
| 586ms | 35% | 28% | -7% |
| 587ms | 32% | 25% | -7% |
| 588ms | 36% | 26% | -10% |
| 589ms | 29% | 27% | -2% |
| 590ms | 17% | 15% | -2% |
| Average | 68.4% | 62% | -6.4% |
What to do with this
Reset the target first. Wang put it at 40% from three being good, and the table backs him up: the whole population is shooting in the 30s from deep in every mode with defenders in it. If you are over 40% you are already ahead of the game, and if you are in the mid 30s you are average, not broken.
If you were a 70%+ shooter last year, that is still achievable, but not likely. Make sure you find a jumpshot that fits you and practice.
- Pick the jumper you can time, not the fastest one. Every speed up costs more than it did, and Very Quick costs a tenth of the window before a defender has done anything. Unless you are getting smothered, Normal or Quick is the better trade. The Release Speed Test has the numbers.
- Smarter Shot Selection. The peak of the window did not move. A well timed open shot goes in exactly as it did last year. The shots that got harder are the contested and hurried ones, so the difference between a 35% and a 45% season this year is shot selection more than it is timing.
- Take the meter off boost. It is still 10%, and it lands entirely on late releases, which is the side of the window this year punishes. The Shot Meter Test covers it.
- Give it the reps. The core of the window is half the size, and a core that size only gives you consistency once your release is grooved. Wang's numbers were climbing daily for a reason. Use the Jump Shot Practice Tool and stay on one jumper.
How to Shoot · Shooting Ratings Test · Mike Wang Dev Updates
Go to NBA2KLab YouTube →