WHITEOUT SURVIVAL ・ PROVEN BY THE NUMBERS

Twin Star Adventure by the Numbers
— Dice needed & spend efficiency, via Monte Carlo

A data-driven breakdown of the board-game event "Twin Star Adventure" (× Kingshot collab). The conclusion is simple: skill of play barely matters. What matters is which reward you target, and how to reach it with the fewest dice and the cheapest spend.

State: 3597Gen: 5Method: Monte Carlo 50k + 3 measured samples
Not by feel — by the numbers. ・ WOS ANALYSIS

01Why skill of play doesn't matter

The rules of Twin Star Adventure:

Fact 1: The die value does NOT increase your coin income from tiles
One roll = one landing = one tile's reward. Rolling a 6 or a 1 both give you exactly one tile. The pip changes where you go, not how much you get.
Fact 2: You can't choose the pip on a normal die
A normal die is uniform 1–6, fixed expectation 3.5. Because "1" is in the mix, over many laps your landings converge to a near-uniform spread across all tiles (a random walk's uniform convergence). Deliberately hitting only the good tiles is therefore impossible with normal dice.

Together, these make coin income essentially a function of the number of rolls. Neither pip choice nor tile targeting creates an edge. The task is simply: roll every die. The only real choices are where to spend your 6 crystal dice and which pack to buy.

02Board composition (confirmed)

Tile typeCountDetail
Treasure tile (high, MAX 2000)1avg 500 coins
Treasure tile (low, MAX 800)3avg 200 coins
Item tile (the only coin-dropping one, Lv3)1avg ~29 coins
Member tile2flat 100 coins
Item tile (fire crystal / gear)~180 coins
Start tile1pass = crystals; land = more
Explore-chest tilevariesrandom spawn; tile reward + chest

Most of the board (~18 tiles) drops fire crystals and gear — zero coins. The coin engine is the four treasure tiles, especially the single MAX-2000 high tile. The explore chest spawns at a random position and pays "tile reward + chest" when landed on (chest: 2% for 3,000 coins / 8% for 2,000 / 10% for 1,500; the remaining 80% is generic badges).

High treasure tile — coin distribution

CoinsChance
2,0002%
1,0008%
80010%
60020%
30060%

The average is 500 coins per landing — that's what "avg 500" means.

03Board output: about 70 coins per roll

The sim needs one number: average coins per roll. Measured three times, the value converged.

RollsNet board coinsPer roll
26 rolls2,58099
127 rolls8,30065
159 rolls11,25071

The early "99" from 26 rolls was an overshoot: a small sample that happened to hit the 2,000 treasure early. With more rolls it settles to about 70 coins/roll. Trusting the small sample would halve the estimated dice needed, so it was pinned down with more data. "Net board coins" here strips out shark, progress, member passive income, and mission rewards — the pure landing value.

04Off-dice rewards are step-shaped

Apart from tile landings, two side rewards accumulate as you roll — and both require you to roll to gain them. This mirrors the Frozen Treasure (gacha-box) event: not linear in roll count, but a step function that jumps at each threshold.

Shark reward (cumulative damage)

Thresholds at 200/1000/2000/3000/4000/5000 pay 200/300/500/800/1200/1500 coins etc. The cap is low, and with teammates helping, normal rolling usually clears it.

Progress reward (cumulative-pip rank) — the top rungs are out of reach

Cumulative pipsRewardRolls needed
100500 coins~29
3502,000 coins~100
700unusual badge~200
1,60010,000 coins~457
3,50010 island decorations~1,000
CAUTION Even the 1,600-pip / 10,000-coin rung needs ~457 rolls; 3,500 needs ~1,000. In money that's tens of thousands of yen — unreachable in normal play. Realistically, only the 100- and 350-pip rungs (2,500 coins total) can be counted on.

05Where to spend your 6 crystal dice

UseAvg coins (6 dice)Note
High treasure tile (MAX 2000)~3,00090% range 2,100–4,500 · stable
Explore chest~2,2260–5,500 · downside 0 risk / ~840 badges
Member tile600100×6, clearly lower
Big pips to farm laps~0all 6 = ~1.4 laps = noise
VERDICT Spend all 6 on the high treasure tile. "Farm laps with big pips" only buys ~1.4 laps from 6 dice — noise, and it adds almost no coins. Just use them when a high treasure tile is within range (6 tiles ahead). Treasure tiles don't level, so there's nothing to save them for.

06Monte Carlo results

Calibrated on the three measured samples (fixed seed, 50k runs). Example target: 30 Mithril = 24,000 coins.

Total rollsMedian90% bandHit rate
159 (measured)20,24916,638–24,7318%
20024,80220,770–29,79162%
22026,29922,065–31,44580%
25028,56124,021–34,01395%
~200–231 rolls
to fully secure 30 Mithril (50%–90% hit rate)

07Cheapest spend for all 30 Mithril

Free dice (7-day total) = 132

Paid packs (dice only — no coins included)

Prices in JPY (¥). Convert to your local currency as needed.

PackPriceDicePer dieCap
Pack 1¥1502¥75
Pack 2¥5005¥100
Pack 3¥8008¥100
Pack 4¥1,60014¥114
Pack 5¥3,20026¥123
Pack 6¥8,00055¥145
Pack 7¥15,800110¥144
CHEAPEST 30 Mithril at 90% confidence (231 rolls, from zero) = 132 free + 99 paid.
Cheapest: ¥9,650 — Pack 1 ×7 (¥1,050) + Pack 2 ×6 (¥3,000) + Pack 3 ×7 (¥5,600).
Fill the cheapest packs to their caps first. Free dice alone reach only ~132 rolls, so 30 Mithril is clearly a paying goal.

08Bottom line

▶ Open the dice calculator for your own target
Figures are from a fixed-seed (seed=42) 50k-run Monte Carlo, calibrated on three measured samples (26/127/159 rolls). Board average is ~70 coins/roll (measured range 65–71), shown with confidence intervals; more samples tighten it.
Not by feel — by the numbers. ・ WOS ANALYSIS