# Draft simulator > Practise the Mobile Legends ranked draft for Epic, Legend, Mythic, Mythical Honor, Mythical Glory and Mythical Immortal, or the tournament draft. The tool is interactive, so this page only describes the rules it follows and the data behind its suggestions. ## Ranked rules - Both teams ban blind in two phases: 3 bans each, then 2 (Epic: 1, then 2; Legend: 2, then 2); each slot is one player, who may skip their ban. Bans stay hidden until both teams lock; both teams may ban the same hero in one phase. - Every ban phase and pick has a 30-second timer when the timer is on (40 seconds in the tournament draft). - Then ten picks in the order 1, 2, 2, 2, 2, 1, starting with the first-pick team (a coin flip, which can be overridden). - A banned hero cannot be picked and a hero can be picked once. - Two people share one device; a hand-off screen hides bans between the teams. ## Tournament rules - Twenty turns of one hero each, every ban open to both teams. The team that bans first also picks first (F below; S is the other team). - Bans F, S, F, S, F, S; picks F, S, S, F, F, S; bans S, F, S, F; picks S, F, F, S. Each team ends with 5 bans and 5 picks. - A banned hero cannot be banned again or picked, and a hero can be picked once. A ban turn may be skipped. - Stats can come from any rank from All ranks to Mythical Glory; Immortal uses Mythical Glory data. - The tournament win chance counts matchups at a quarter of the ranked weight: ranked matchup data predicts pro results only weakly. ## Where the suggestions come from - Win, pick and ban rates and each hero's top counters come from the hero rankings endpoint, lanes and roles from the positions endpoint, and full matchup and teammate lists of picked heroes from the academy counters and teammates endpoints. - Picks are read like a coach: every hero needs a lane of its own, matchups against the direct lane opponent count most, the team needs a frontliner plus magic and physical damage, and a hero whose strong counters are still open is risky to pick early. - Blind bans are worth less on a hero the enemy is likely banning too. - A draft-only win chance for each team combines hero strength, every matchup, pairings and composition like a logistic draft model; it leaves out player skill. - The stats rank defaults to the draft rank and can be changed. The API has no Immortal filter, so Immortal uses Mythical Glory data. - Three suggestion methods: the rule-based coach (weights fitted on pro games), a machine-learning model (a conditional logit trained on what pro teams picked; in the tournament draft it also reads how often pros play and pair heroes), and a blend of both. Enemy picks are always predicted by the machine-learning model. - Accuracy is measured on held-out professional games (5-fold cross-validation): how often the hero the pros actually picked is a method's first suggestion and in its top 10. The tool shows these numbers next to each method. - Once the enemy has shown picks, ban advice aims at the lanes they still need and skips heroes they have no lane for; pick advice moves up heroes the enemy is likely to take next. - Game timing: each hero's win rate by game length in its lane (from the academy win-rate timeline endpoint), compared with the lane's average hero, tags it as an early-game (before 14 minutes), mid-game (14-18) or late-game (after 18) hero; both teams get a power curve and the picking team sees when each draft peaks. - Heroes can be filtered by lane and by role; a hero with two lanes or two roles appears under both, and team needs (a frontliner, magic and physical damage) count every role a hero has. - The learned model, pro meta and timing snapshots refresh twice a week; hero stats, matchups and pairings are read live. - A finished draft gets a pick review: every ban and pick ranked among the suggestions at that moment and graded (Best to Mistake), with each team's accuracy, and each team's damage split and frontline. - Suggestions are hints from win-rate differences, not a win prediction. On pro games the draft-only win chance picks the winner about 54 % of the time: play decides most games. --- HTML version: https://arena-academy.rone.dev/tools/draft All pages: https://arena-academy.rone.dev/llms.txt Unofficial, community-maintained; not affiliated with Moonton. Game data comes from the Rone Arena API.