Genshin 50/50 & wish simulator
The math says 55% - your heart says otherwise. Run your pull plan through the documented rates first: every 5★, guarantee, Radiance rescue and fate point resolved wish-by-wish, then sanity-check your luck across ten thousand parallel timelines. For education and expectation management; the RNG here is not HoYoverse's RNG.
Your run
Set your situation and hit “Wish once”.
3★ · 4★ · 5★ · your target
Monte Carlo - parallel timelines
What this simulator models
Every pull uses the documented per-wish rates - 0.6% base on character/standard banners with the soft-pity ramp from wish 74 and hard pity at 90; 0.7% base on the weapon banner ramping from ~63 with hard pity at 80. 5★ resolution follows the official rules: the 50/50 with guarantee, Capturing Radiance (v5.0+) as a 10% rescue on an armed losing 50/50, and the one-fate-point Epitomized Path. 4★s use the simplified documented model (5.1% base, guaranteed every 10th).
The Monte Carlo engine is the same one the calculator pages verify against - in our test suite, 20,000 simulated runs agree with the exact dynamic-programming answer to within a percentage point. Model details and sources: methodology. Snapshot October 2026
Simulator FAQ
Can this predict my pulls?
No - nothing can. What it does is show you the shape of the randomness: how often a 160-wish budget ends without the character (about 1 run in 10), how Radiance rescues feel, and why “I lost five 50/50s in a row” is painful but possible.
Why do 10,000 runs sometimes beat 1,000 to the exact answer?
They converge: the Monte Carlo percentage drifts toward the exact dynamic-programming value shown in teal as run count grows. Comparing the two is the point - it demonstrates the law of large numbers on your own screen.
What does the seed do?
It fixes the random sequence, like replaying the same timeline. Share a seed and a friend gets the identical run - handy for comparing luck honestly.