Skip to content
Tool Corner

Dice Probability Simulator: Advantage, Keep-Highest and Success Pools

See the exact odds behind the dice rolls used in many tabletop and video games: a single check with advantage or disadvantage, keeping the highest or lowest dice, and counting successes in a pool.

Illustration of colourful glass dice tumbling in mid-air on a purple background
Illustration
Built and verified by Jogeswara, Founder and Engineer, MSc, PMP — Tool CornerModel stated in full belowReviewed

Press Roll to roll one example with the seeded dice.

Skip to content
Tool Corner

Dice Probability Simulator: Advantage, Keep-Highest and Success Pools

See the exact odds behind the dice rolls used in many tabletop and video games: a single check with advantage or disadvantage, keeping the highest or lowest dice, and counting successes in a pool.

Illustration of colourful glass dice tumbling in mid-air on a purple background
Illustration
Built and verified by Jogeswara, Founder and Engineer, MSc, PMP — Tool CornerModel stated in full belowReviewed
{{ f.face }}

{{ rollTxt }}

{{ mcTxt }}

Seed {{ seedTxt }}. The exact odds never change with the seed; only the sampled rolls do.

Try this: {{ tryTxt }}

Headline

{{ headline }}

{{ statLab }}
Mean
{{ cMean }}
Standard deviation
{{ cSd }}
{{ cSdP }}
Median
{{ cMed }}
{{ cMedP }}
Minimum
{{ cMin }}
Maximum
{{ cMax }}
{{ cHitLab }}
{{ cHit }}
Chance to reach each target
Show the numbers (exact chance of each result)
{{ rowHead }}
Exactly this
This or more
Reaches target
{{ row.v }}
{{ row.p }}
{{ row.ge }}
{{ row.hit }}

For very wide spreads, results with less than a 0.05% chance are left out of this table; the charts and cards use every result.

Honesty note

These are pure probabilities for fair dice. No real game's special rules are included, and the page makes no claim about how any specific game works. It is for entertainment and learning, not gambling advice.

What probability means for dice

A fair die gives every face the same chance. On a 20-sided die each number comes up 1 time in 20, or 5%. A probability of 55% for a check means that over a very long run of identical rolls, about 55 in every 100 succeed. It says nothing about the next roll: dice have no memory, and a run of failures does not make a success more likely.

Everything on this page except the sampled rolls is exact: it comes from counting the ways each result can happen, not from simulation. "Roll 1,000 times" shows what sampling looks like next to those exact answers. The white marks land close to the exact bars and wobble around them, because 1,000 rolls is a lot, but not infinite.

Why advantage helps most in the middle

Rolling two dice and keeping the higher only fails if both dice fail, so the chance of success becomes 1 − (1 − p)2. Disadvantage, keeping the lower, only succeeds if both succeed: p2. The gain from advantage is p(1 − p), which is biggest when p is one half. A check you would pass half the time jumps by 25 percentage points, to 75%.

At the extremes it barely matters. A check that needs a 20 on a d20 goes from 5% to only 9.75%; one you pass 95% of the time goes to 99.75%. With the defaults, a d20 plus 5 against 15, normal is 55%, advantage 79.75% and disadvantage 30.25%.

Why sums of several dice bunch up near the middle

One die is flat: every face is equally likely. Add a second and the middle totals can be made in many more ways than the extremes. Two six-sided dice make 7 in six ways but 2 in only one. With three dice the shape is already close to a bell: 3d6 averages 10.5, and a total of 15 or more happens only 9.26% of the time.

Why keeping the highest raises the average and narrows the spread

Roll four six-sided dice and drop the lowest. The dropped die is usually a low one, so the kept three average 12.24 instead of 10.5. The spread also shrinks a little, to a standard deviation of 2.85 against 2.96 for plain 3d6, and the shape leans towards the top: the median is 12 and a total of 15 or more comes up 23.15% of the time. Keeping the lowest does the opposite and pulls the results down.

Advantage is the simplest case: two d20s, keep the highest one. Mode 2 with 2d20 keep highest 1 gives exactly the same chances as advantage in mode 1, which is a useful check that both calculations agree.

How success pools work

In a success pool you roll a handful of dice and count how many show at least a set number. Each die is a separate yes-or-no trial with the same chance, so the count follows the binomial distribution. With six ten-sided dice succeeding on 8 or more, each die has a 30% chance, the average is 1.8 successes, and the chance of at least two is 57.98%. Adding dice moves the whole distribution up; raising the number each die needs moves it down.

How it works: the formulas

need = DC − b; k = S if need ≤ 1, 0 if need > S, else S − need + 1
p = k / S (natural extremes: p kept between 1/S and (S − 1)/S)
advantage = 1 − (1 − p)2; disadvantage = p2
pool: P(X = x) = C(N, x) qx (1 − q)N − x, q = (S − T + 1) / S
  • Mode 1: S sides, bonus b, target DC. A roll succeeds if the face plus b is at least DC; k is the number of faces that succeed. With natural extremes on, the top face always succeeds and a 1 always fails.
  • Mode 2: N dice of S sides, keep the K highest or lowest, add the bonus. The exact distribution is built by going through the face values from best to worst and counting, for every number of dice placed so far and every running kept total, how many ways the dice can fall. No random numbers are used.
  • Mode 3: N dice of S sides; a die succeeds if it shows T or more, so each die succeeds with chance q. The number of successes X is binomial; its mean is Nq.
  • Mean, standard deviation, median, minimum and maximum are read straight from the exact distribution. The median is the smallest result with at least a 50% chance of being at or below it.
  • Sampled rolls use a seeded random-number generator (mulberry32), so a link with the same seed repeats the same rolls.

Worked example: the default settings

  • Check: d20 + 5 against 15 needs 10 or more on the die, which is 11 faces of 20. Normal 11/20 = 55%; advantage 1 − 0.452 = 79.75%; disadvantage 0.552 = 30.25%.
  • Keep highest: 4d6 keep 3 has 1,296 equally likely rolls. A total of 18 can only come from three or four sixes, which happens 21 ways, 1.62%. Mean 12.24, standard deviation 2.85, median 12, chance to reach 15 23.15%.
  • Success pool: 6d10 succeeding on 8+ gives q = 0.3. No successes: 0.76 = 11.76%. At least two: 57.98%. Average 6 × 0.3 = 1.8.
  • 1,000 sampled rolls of 4d6 keep 3 with seed 1 average 12.16, close to the exact 12.24.

How to use this simulator

  • Pick a mode with the tabs. Each tab starts from its own sensible defaults.
  • Move the sliders, or tab to one and use the arrow keys. The die-size buttons jump to the common sizes.
  • Roll shows one example roll with its dice. Roll 1,000 times overlays a sample on the exact bars; each press uses the next seed. Show the numbers opens a table of every result for screen readers or copying.
  • Copy link saves every setting, including the mode and seed. Save picture downloads the main chart with the settings and headline underneath.

Checks, kept dice and success pools in numbers

The tables are exact. They are built by the page's own formulas and dynamic programme, with no random rolls, so they match what the simulator shows for the same settings.

One d20 check: normal, advantage and disadvantage (natural extremes off)
CheckFaces that succeedNormalAdvantageDisadvantageAdvantage adds
d20 vs DC 1011 of 2055.00%79.75%30.25%24.75 points
d20 vs DC 156 of 2030.00%51.00%9.00%21.00 points
d20 vs DC 201 of 205.00%9.75%0.25%4.75 points
d20 + 3 vs DC 159 of 2045.00%69.75%20.25%24.75 points
d20 + 5 vs DC 1016 of 2080.00%96.00%64.00%16.00 points
d20 + 5 vs DC 1511 of 2055.00%79.75%30.25%24.75 points
d20 + 5 vs DC 206 of 2030.00%51.00%9.00%21.00 points
d20 + 8 vs DC 1514 of 2070.00%91.00%49.00%21.00 points
d20 + 8 vs DC 209 of 2045.00%69.75%20.25%24.75 points

Advantage helps most in the middle: a check that succeeds on 11 of 20 faces (d20 + 5 against 15) goes from 55% to 79.75%, a gain of 24.75 points. A check needing a 20 goes from 5% to 9.75%, and one needing a 15 from 30% to 51%.

Natural extremes at the edges (d20, no bonus)
CheckNatural extremesNormalAdvantageDisadvantage
d20 vs DC 25Off0.00%0.00%0.00%
d20 vs DC 25On5.00%9.75%0.25%
d20 vs DC 1Off100.00%100.00%100.00%
d20 vs DC 1On95.00%99.75%90.25%

With natural extremes on, a roll that could never succeed still has the 1-in-20 chance of the top face, and a roll that always would still fails on a 1: 100% becomes 95%, and advantage turns 5% into 9.75%.

Keeping dice

Keep-highest and keep-lowest schemes (exact, from the page's dynamic programme)
RollMeanStandard deviationMedianRangeTotal of 15 or more
3d610.502.96103 to 189.26%
4d6 keep highest 312.242.85123 to 1823.15%
5d6 keep highest 313.432.60143 to 1837.71%
4d6 keep lowest 38.762.8593 to 182.78%
2d20 keep highest 113.834.71151 to 2051.00%
2d20 keep lowest 17.184.7161 to 209.00%
3d20 keep highest 115.493.87161 to 2065.70%

Rolling four d6 and dropping the lowest lifts the mean from 10.50 to 12.24 and trims the spread from 2.96 to 2.85. A total of 15 or more comes up 23.15% of the time against 9.26% for plain 3d6. Keeping the highest of two d20s is advantage, and the table's 13.83 mean on the 2d20 line is the average of that roll.

Success pools

Success pools on d10s: chance of at least 2 successes, by pool size and target
PoolSucceed on 6+Succeed on 8+Succeed on 9+Succeed on 10
2 dice25.0%9.0%4.0%1.0%
4 dice68.8%34.8%18.1%5.2%
6 dice89.1%58.0%34.5%11.4%
8 dice96.5%74.5%49.7%18.7%
10 dice98.9%85.1%62.4%26.4%
The same pools succeeding on 8 or more
PoolChance per dieAverage successesNo successes
2 dice30%0.6049.00%
4 dice30%1.2024.01%
6 dice30%1.8011.76%
8 dice30%2.405.76%
10 dice30%3.002.82%

At 8 or more each die has a 30% chance. Six dice average 1.80 successes and have an 11.76% chance of none, and 57.98% of at least two. Raising the target to 10 cuts the chance of two or more from 57.98% to 11.43% for the same six dice.

Six mistakes when working out dice odds

  • Treating advantage as a flat bonus. It adds up to 25 points in the middle and under 5 at the extremes, as the first two tables show. A bonus adds the same 5 points per +1 until it hits 0% or 100%.
  • Adding the dice instead of keeping them. Three dice summed and four dice with the lowest dropped are different: the mean rises from 10.5 to 12.24 and the spread falls.
  • Expecting a run of luck to even out. Every roll is independent. Ten misses do not make the next roll more likely to hit, and the page assumes fair dice throughout.
  • Forgetting what natural extremes do. It is a single house rule for mode 1. Switched on, 100% becomes 95% and 0% becomes 5%, so check the setting before comparing.
  • Using the average of a success pool as a promise. Six dice at 30% average 1.8 successes but have an 11.76% chance of none at all.
  • Applying the numbers to a game's special rules. Rerolls, exploding dice, critical hits and ties are not modelled. The page gives exact odds for plain fair dice.

Embed this calculator on your site

You can put this tool on your own page, blog or lesson site for free. Copy the code below and paste it into your page. Please keep the credit line under the tool. The embedded version shows no ads, loads no analytics and sets no cookies.

<iframe src="https://toolcorner.net/embed/dice-probability-simulator" title="Dice Probability Simulator" width="100%" height="1820" style="border:0;max-width:760px" loading="lazy"></iframe>
<p style="font-size:13px">Calculator by <a href="https://toolcorner.net/labs/dice-probability-simulator">Tool Corner</a></p>

Cite this page

If you use this tool in a report, assignment or article, you can cite it like this.

Tool Corner. (2026). Dice Probability Simulator [Online calculator]. https://toolcorner.net/labs/dice-probability-simulator

Dice Probability Simulator. Tool Corner. https://toolcorner.net/labs/dice-probability-simulator. Retrieved [date].

Assumptions & limitations: what the model leaves out

  • Every die is fair and every roll is independent.
  • No rerolls, exploding dice, critical doubling, botches, ties or other special rules from any game.
  • Natural extremes is a single on-or-off house rule for mode 1 only.
  • Success pools count one success per die; no die counts twice and none subtracts.
  • It says nothing about how any particular game, system or video game uses dice.

Frequently asked questions

How much does advantage help on a d20?

It depends on how hard the check is. If you would succeed with chance p, advantage gives 1 − (1 − p)². A 55% check becomes 79.75%, a 50% check becomes 75%, but a check needing a natural 20 only rises from 5% to 9.75%.

Is advantage the same as adding a bonus?

Not exactly. In the middle of the range it adds up to 25 percentage points, about the same as a bonus of 5 on a d20, but near the extremes it adds much less, while a flat bonus adds the same 5% per point until it hits 0% or 100%.

What is the average of 4d6 keep the highest 3?

About 12.24, against 10.5 for plain 3d6. Out of the 1,296 equally likely rolls, 21 give the maximum of 18, and a total of 15 or more comes up about 23.1% of the time.

How do I work out a dice success pool?

Each die succeeds with chance q, the share of faces at or above the threshold, so the count of successes is binomial. Six ten-sided dice succeeding on 8 or more give q = 0.3, an average of 1.8 successes and about a 58% chance of at least two.

Why do 1,000 sampled rolls not match the exact answer?

Because a sample always wobbles. With 1,000 rolls the share of successes is usually within about three percentage points of the exact chance. The exact numbers are what a never-ending run of rolls would settle on.

Related calculators

Sources and references

Last updated
Was this result right?
What next?
Educational only

This simulator shows exact probabilities for fair dice. It does not describe the rules of any game and is not gambling advice.