If you have ever stared at a gacha banner wondering why your first ten pulls felt hopeless and your fortieth suddenly felt inevitable, you already understand the gap between soft pity and hard pity even if you never learned the terms. The short answer: soft pity is a rising slope, a point where the game starts stacking the odds in your favor pull after pull. Hard pity is a hard wall, the exact pull number where the drop is no longer probabilistic at all and simply happens. Every major gacha game uses some version of both, but the pull counts, ramp speed, and real-money cost behind them vary enormously by title and publisher.
This piece walks through the actual math using one current, well-documented case study, Arknights: Endfield, whose 2026 pity numbers are consistently reported across multiple independent trackers including PCGamesN, Game8, and the game’s own community wiki. We will build a full pity-cost table, run a step-by-step expected-value calculation, compare the system to two other published gacha models, and separate the myths from the facts. Every number below traces back to a cited source, and where sources disagree or a figure could not be verified, that gap is flagged rather than guessed at. For the same kind of breakdown on other banners, see our provably-fair odds hub, including our separate math on Genshin Impact’s 50/50 system and Honkai Star Rail’s pity odds.
Soft Pity vs Hard Pity: The Plain-Language Difference
Soft pity and hard pity solve the same underlying problem for gacha publishers: pure random chance produces streaks, and streaks make players quit. A flat 0.8% chance per pull, run enough times across enough players, guarantees that some fraction of the player base goes 150, 200, even 300 pulls without a hit. That outcome reads as broken to a player even though it is mathematically expected. Pity systems exist to cap the pain.
Soft pity is the point where the base drop rate starts climbing with every additional pull that comes up empty. It is not a guarantee, it is an acceleration. You can still lose a pull during soft pity, you are just losing with worse and worse odds stacked against the house each time.
Hard pity is the ceiling. It is the specific pull number at which the probability is forced to 100%, regardless of what the formula would otherwise output. No gacha player who reaches hard pity walks away empty-handed on that pull, by design.
A third, related mechanic shows up in character-banner games: the 50/50. Once you hit a top-rarity drop, whether through luck, soft pity, or hard pity, many games then flip a coin between the featured (rate-up) unit and a random off-banner unit of the same rarity. Losing that coin flip is usually softened by yet another guarantee, a hard cap on how many top-rarity pulls you can lose the coin flip before the featured unit is forced. Understanding soft pity vs hard pity really means understanding all three layers stacked together.
Why the Ramp Matters More Than the Base Rate
Publishers love to advertise the base rate because it looks generous in isolation, something like “0.8% chance per pull.” That number alone tells you almost nothing about how many pulls you will actually spend, because a pity system means the base rate only governs a fraction of your total pulls. Once soft pity kicks in, the real driver of your outcome becomes the ramp speed and the hard cap, not the advertised base percentage.
This is where a worked case study earns its keep. Arknights: Endfield, the 2026 release from Hypergryph and its Chartered Headhunting banner system, publishes rates that are unusually consistent across independent sources: a 0.8% base 6-star rate, a soft pity ramp beginning at pull 65 that adds five percentage points per pull, and a hard pity guarantee at pull 80. Those numbers appear, matching digit for digit, on PCGamesN’s pity breakdown, the Arknights: Endfield community wiki, and the guide site allthings.how, which is the kind of cross-source agreement that makes a number safe to build math on.
The Arknights: Endfield Pity Table, Pull by Pull
The table below shows the per-pull chance of landing a 6-star operator, along with the cumulative probability of having hit one by that point, calculated directly from the published ramp. Pulls 1 through 64 sit flat at the base rate. Pull 65 is still base rate, since the ramp is described as beginning “after” 65. From pull 66 onward, the rate climbs by five percentage points every single pull until it is forced to 100% at pull 80.
| Pull number | Per-pull 6★ rate | Cumulative chance of a hit by this pull |
|---|---|---|
| 1 | 0.8% | 0.8% |
| 10 | 0.8% | 7.7% |
| 20 | 0.8% | 14.8% |
| 30 | 0.8% | 21.4% |
| 40 | 0.8% | 27.5% |
| 50 | 0.8% | 33.1% |
| 65 (soft pity begins) | 0.8% | 40.7% |
| 66 | 5.8% | 44.1% |
| 68 | 15.8% | 58.0% |
| 70 | 25.8% | 75.3% |
| 75 | 50.8% | 98.3% |
| 79 | 70.8% | 99.97% |
| 80 (hard pity) | 100% | 100% |
The base per-pull rate and the pull-65 soft pity trigger are drawn directly from the Arknights: Endfield community wiki’s pity breakdown, and the hard pity cap at pull 80 is independently confirmed by PCGamesN’s coverage of the system. The cumulative-probability column is our own calculation, run pull-by-pull off those published rates, not a number either source published directly.
Notice the shape of that curve. You cross the 50% mark around pull 68, twelve pulls after soft pity starts. You cross 98% by pull 75. The gap between “coin flip odds of having hit” and “guaranteed” is only fifteen pulls, which is exactly what a soft pity ramp is designed to do: compress the tail so almost nobody actually reaches the hard cap.
The Pity-Cost Table: What a Guarantee Actually Costs
Getting any 6-star operator is only half the question. Endfield layers a 50/50 mechanic on top: the first 6-star you pull on a limited banner has a 50% chance of being the featured unit and a 50% chance of being a random off-banner 6-star. If you lose that coin flip, the game does not leave you stranded. Community-verified reporting from pitycalculator.com and Game8’s official-adjacent guide both describe a separate, harder guarantee: reach 120 total pulls on the same banner and the featured unit is handed to you outright, independent of the 50/50.
Each pull costs 500 Oroberyl, and a 10-pull costs a flat 5,000 Oroberyl, confirmed by both the Arknights: Endfield gacha guide at arknights.win and allthings.how’s plain-language breakdown. There is no bulk discount baked into the 10-pull, the price is exactly ten times the single-pull cost. For real-money cost, a third-party breakdown of the game’s iOS store pricing found that the $69.99 premium currency pack converts to roughly 52.5 pulls worth of currency, which works out to about $1.33 per single pull. That per-pull figure is a third-party calculation based on store pricing, not an officially published exchange rate, and we treat it as an estimate throughout this article rather than a confirmed developer figure.
| Scenario | Pulls | Oroberyl cost | 10-pulls needed | Est. USD cost |
|---|---|---|---|---|
| Expected pulls for any 6★ (mathematical average) | ~54 | 26,950 | 6 | ~$72 |
| Blended expected cost for the featured unit | ~87 | 43,475 | 9 | ~$116 |
| Hard guarantee, worst case (120-pull cap) | 120 | 60,000 | 12 | ~$160 |
That “blended” row is not a number any publisher or tracker published. We built it, and the next section shows the exact steps, so you can check it or rebuild it yourself with different assumptions.
Step-by-Step: How We Calculated the Expected Value
Here is the full expected-value walkthrough, using only the published rates cited above.
- Step 1: Find the expected pulls for any 6-star. Using the per-pull ramp (0.8% for pulls 1-64, climbing five percentage points per pull from 66 onward, forced to 100% at 80), we summed the survival probability across all 80 possible pull counts. That calculation lands on an expected value of 53.9 pulls, rounded to 54, to land any 6-star operator. This is a standard expected-value method for capped pity systems: sum the probability of “still not hit” across every pull count from zero up to the hard cap.
- Step 2: Apply the 50/50. Half the time, that 6-star lands as the featured unit. In that half of outcomes, your cost to get the featured unit equals the same 54-pull average from Step 1, because the coin flip itself does not cost extra pulls.
- Step 3: Model the loss case. The other half of the time, you lose the 50/50 and must continue pulling. The documented backstop is the 120-pull guarantee: no matter how the second cycle of pity plays out, you have the featured unit in hand by pull 120 on that banner. We use 120 as the cost for this branch, which is the conservative (worst-case) published figure rather than an estimate of the second cycle’s own average.
- Step 4: Blend the two branches. (0.5 × 54) + (0.5 × 120) = 27 + 60 = 87 pulls. That is the blended expected-value figure in the table above.
- Step 5: Convert to cost. 87 pulls × 500 Oroberyl = 43,500 Oroberyl (rounding differences from the unrounded 53.9 figure account for the 43,475 shown in the table). At the estimated $1.33-per-pull rate derived from the $69.99 pack, that is roughly $116.
This blended figure is deliberately conservative on the loss branch. In reality, some players who lose the first 50/50 will get lucky on their second climb toward another 6-star and land the featured unit before hitting the full 120-pull wall, which would pull the true average somewhere between 87 and 54. We chose the worst-case assumption for the loss branch specifically because the task of estimating fair-game math should never make the odds look better than the published guarantee actually promises.
Five Worked Examples With Real Numbers
Example 1: The lucky puller. A player hits a 6-star on pull 12 and it happens to be the featured unit. Total cost: 12 pulls, 6,000 Oroberyl, roughly $16 at the estimated per-pull rate. This sits well below the 54-pull average and happens for a real, calculable slice of the player base, roughly 14.8% of players will have hit any 6-star by pull 20.
Example 2: The average case. A player pulls their first 6-star right around pull 54, matching the calculated expected value, and wins the 50/50. Total cost: 54 pulls, 27,000 Oroberyl, about $72.
Example 3: The single lost coin flip. A player hits their first 6-star at pull 54, loses the 50/50, then gets lucky and hits a second 6-star quickly, on pull 75 overall, and wins that coin flip. Total cost: 75 pulls, 37,500 Oroberyl, about $100.
Example 4: The unlucky run to the wall. A player does not land any 6-star until forced by hard pity at pull 80, loses the 50/50, and then does not hit a second 6-star before the 120-pull guarantee simply hands over the featured unit. Total cost: 120 pulls, 60,000 Oroberyl, about $160.
Example 5: The multi-banner budget. A player who wants two featured units across two separate banner runs, using the blended 87-pull average for each, should budget roughly 174 pulls, 87,000 Oroberyl, and about $232, before any free currency earned through normal play is subtracted from that total.
How Endfield’s Model Compares to Other Published Pity Systems
Endfield is not the first Hypergryph game with a documented pity system, and it is useful to see how its numbers sit next to two other published models: the original mobile Arknights, and Fate/Grand Order’s very different guarantee structure.
| Game | Currency per pull | Base top-rarity rate | Soft pity start | Guarantee model |
|---|---|---|---|---|
| Arknights: Endfield | 500 Oroberyl | 0.8% | Pull 65 | Hard pity 80 (any), featured guaranteed by 120 |
| Arknights (mobile) | 600 Orundum (≈3.33 Originite Prime) | 2.0% | Pull 50 | Ramp continues; exact hard-pity cap is inconsistently reported across trackers and is not independently confirmed here |
| Fate/Grand Order | 3 Saint Quartz per roll | Not a ramping system | No soft pity ramp | Flat 330-roll Guaranteed Summon (990 Saint Quartz total) |
The Arknights mobile figures for currency cost per pull and soft pity start come from the Arknights Wiki on Fandom and a 2026 Orundum farming guide, both of which agree on 600 Orundum per pull and a soft pity start at 50. Where trackers disagree on the exact hard-pity pull count for the mobile game, some report 99 and others report 300, we are not printing either figure as fact, because printing a number you cannot verify is worse than admitting the gap. The Fate/Grand Order figures come from a single, detailed source describing its Guaranteed Summon campaign structure, which is fundamentally different from a ramping pity system: there is no soft pity at all, just a flat count-to-330 hard guarantee, which is arguably easier to plan a budget around even though the total cost, at roughly $250 to $300 depending on regional Saint Quartz bundle pricing, sits in a similar range to Endfield’s worst-case scenario. Readers comparing systems across more games can also check our breakdowns of Zenless Zone Zero’s gacha rates, Wuthering Waves’ pity odds, and Blue Archive’s pity system, all of which use different base rates and hard-pity caps than Endfield.
Myths vs Facts About Soft Pity and Hard Pity
Myth: If you have not hit in a while, you are “due.” Fact: in a flat-rate system with no pity, that reasoning is the gambler’s fallacy and is simply wrong, past pulls do not affect future ones. But gacha pity systems are explicitly designed to break that independence on purpose. Once soft pity engages, your odds genuinely do rise pull over pull, so “I have not hit in 60 pulls, my odds are better now” is factually correct in a pity-based game, even though it would be false in a casino slot machine with flat odds.
Myth: The 50/50 is a coin flip with no backstop. Fact: in Endfield’s documented system, losing the 50/50 does not leave you exposed indefinitely. The 120-pull guarantee caps the downside, which is precisely the number this article’s pity-cost table is built around.
Myth: Buying in bulk with a 10-pull gets you a discount. Fact: Endfield’s 10-pull costs exactly 5,000 Oroberyl, ten times the 500-Oroberyl single-pull price, with no currency discount. Any advantage to pulling in batches of ten is about pity math and guaranteed rarity floors on multi-pulls in some games, not about price.
Myth: Every gacha game uses roughly the same pity numbers. Fact: base rates, soft pity triggers, and hard pity caps vary widely by publisher and even by banner type within the same game. Endfield’s 0.8% base and 80-pull hard cap are not interchangeable with another studio’s numbers, and readers should always check the specific game before budgeting.
Myth: Publishers hide their real odds. Fact: this varies by jurisdiction. Since a 2016 Ministry of Culture notice took effect in China in May 2017, games sold there are legally required to disclose randomized-item draw probabilities, a rule that academic compliance studies as recently as 2024 still describe as actively enforced, even if imperfectly followed by some publishers. Outside China, disclosure is largely a matter of platform policy and industry codes rather than statute, which is why community trackers and datamining fill the gap for games sold in most other markets.
The Regulatory Backdrop: Who Actually Requires Disclosure
China remains the only country with a nationwide statute forcing exact probability disclosure for randomized game rewards. The rule traces back to the Ministry of Culture’s “Notice on Regulating Online Game Operations and Strengthening Post-Event Supervision,” which took effect on May 1, 2017, and required publishers to disclose the names, properties, quantities, and draw probabilities of items obtainable through random draws, either in-game or on the official website. A Game Developer report from when the rule was first announced and a broader summary of consumer-protection rules against microtransactions both document the requirement, and compliance research published as recently as 2024 confirms the law is still treated as binding, with the ongoing complaint being uneven enforcement rather than repeal.
Belgium takes a different route entirely. Its Gaming Commission ruled in April 2018 that paid loot boxes in games including FIFA, Overwatch, and Counter-Strike: Global Offensive met the legal definition of a game of chance, putting them under national gambling law and exposing operators to fines up to €800,000 and potential prison time. A FAQ page from the Belgian Gaming Commission itself lays out that stance, and it held up as recently as 2025, when an Antwerp court ruling in a case against Apple’s App Store reaffirmed that loot boxes purchased for real money constitute an unlawful game of chance under Belgian law.
Elsewhere, disclosure leans on labeling rather than statute. The ESRB’s “In-Game Purchases (Includes Random Items)” label, in place since April 2020 and still standard on US and North American storefronts, flags the presence of randomized paid mechanics without mandating a specific probability figure be published, a policy confirmed by Engadget’s coverage at the time of its launch. PEGI, the age-rating body used across the UK and most of the EU, went further starting in June 2025: games featuring paid random items such as loot boxes and gacha pulls are now rated PEGI 16 by default, a change reported by the BBC when it took effect. That change ties gacha mechanics to an age gate, not to a numeric-odds mandate, so a game can be fully PEGI-compliant while still leaving players to rely on community trackers, like the ones cited throughout this article, to reverse-engineer the actual math.
Is Pulling for a Featured Unit Worth the Money?
That depends entirely on which of the five worked scenarios above you can afford to land in, and gacha math only tells you the shape of the distribution, not which draw you personally get. What the pity-cost table does give you is a real budget range: roughly $72 if you land near the average for any 6-star and win the coin flip, up to roughly $160 if you hit the full run to the 120-pull guarantee. A reasonable rule for any pity-based game: never budget for the lucky-case number, budget for the guaranteed-case number, because the guarantee is the only outcome the math actually promises you.
The other variable worth weighing is free currency. Most gacha games, Endfield included, hand out premium currency through normal play, story missions, events, and login rewards, which offsets some fraction of the cash cost calculated here. None of the sources used for this article published an exact monthly free-currency rate for Endfield, so that offset is intentionally left out of the pity-cost table rather than estimated. If you track your own free-currency income over a few weeks, you can subtract it directly from the 43,475 to 60,000 Oroberyl range above to get a realistic personal cash number.
How to Read Any Game’s Pity Rates Yourself
The method used in this article works for any gacha game with published or reliably datamined rates, not just Endfield. Four questions get you most of the way to your own pity-cost table.
- What is the flat base rate, and at what pull number does the ramp start? That tells you where the “safe zone” of low odds ends.
- How fast does the rate climb once the ramp starts, and at what pull is it forced to 100%? That gives you the true worst-case pull count, not the advertised base rate.
- Is there a coin-flip layer (a 50/50 or similar) on top of the rarity pity, and if so, is there a separate hard guarantee if you lose it? Skipping this step is the single most common mistake in gacha budgeting, since it can roughly double your real expected cost.
- What does a single pull cost in real currency, and is that number officially published or a third-party estimate from store pricing? Treat unofficial per-pull dollar figures, like the $1.33 estimate used throughout this article, as directional rather than exact.
Run those four questions against the published or datamined numbers for any banner, and you can build the same kind of expected-pulls, expected-cost, and worst-case table shown here, using nothing but real, cited rates.
Frequently Asked Questions
What is the actual difference between soft pity and hard pity?
Soft pity is a rising probability, the game increases your odds of a top-rarity drop with each pull once you cross a certain threshold. Hard pity is a fixed pull number at which the drop is no longer probabilistic and simply happens. In Arknights: Endfield, soft pity starts at pull 65 and hard pity locks in at pull 80.
How many pulls does it take on average to get a featured unit in Arknights: Endfield?
Based on the published rates, the expected number of pulls to land any 6-star is about 54. Factoring in the 50/50 coin flip and the documented 120-pull featured guarantee as a worst-case backstop, a blended expected-value estimate lands around 87 pulls, with a hard ceiling of 120 pulls if you lose the coin flip and do not get lucky on the second run.
Does losing the 50/50 waste all your pity progress?
No. Community-verified reporting on Endfield’s system describes a separate 120-pull guarantee that hands you the featured unit outright if you have not landed it by that point on the same banner, independent of how the earlier 50/50 played out.
How much does a guaranteed featured unit cost in real money?
Using a third-party estimate of roughly $1.33 per pull, derived from the $69.99 premium currency pack’s conversion rate, the full 120-pull worst-case guarantee costs approximately $160. That figure is our calculation built on an unofficial per-pull cost estimate, not a number published directly by the developer.
Is China really the only country that legally requires gacha odds disclosure?
Yes, based on current compliance research. China has required disclosure of randomized-draw probabilities since a Ministry of Culture notice took effect in May 2017. Belgium takes a stricter approach by classifying many paid loot boxes as illegal gambling outright rather than requiring disclosure, while the US and most of the EU rely on labeling requirements, like the ESRB’s random-items tag and PEGI’s age-rating rules, rather than a statutory numeric-odds mandate.
Do soft pity and hard pity numbers carry over between different banners?
It depends on the game and even on which counter you mean within the same game. For Endfield specifically, community reporting is mixed on whether the 0-to-80 “any 6-star” pity carries between banners, with some sources describing carryover and others not, while the 120-pull featured guarantee is more consistently reported as resetting per banner. Because the sources conflict on the first point, treat any claim about cross-banner carryover for a specific game as something to verify in that game’s own patch notes rather than assume.
Should I save currency for a guaranteed pull instead of pulling as I go?
The math in this article supports budgeting toward the guarantee rather than hoping for the lucky-case outcome. Since the worst-case guarantee (120 pulls, roughly $160 by our estimate) is the only outcome the published rates actually promise you, saving toward that number and treating any earlier hit as a bonus is the more reliable way to plan spending on a featured-unit banner.
