Decision 03 · Week 3 · 4 August 2026

A Reroll Is Not Free

Opportunity cost and option value in the one pre-game choice the mode allows

The reroll is the only choice the mode hands back before the game starts, and it is the one most often spent badly. Not because players reroll champions they should keep — sometimes they should reroll — but because the button is pressed before anybody has worked out what is being bought and what is being paid.

What the button actually costs#

Three prices, and only the first is visible.

A reroll is spent currency. It accumulates slowly, across games. Using one now is a decision not to have it in some future game where the roll is worse. The cost is not paid in this game at all, which is exactly why it is underweighted.

A reroll discards a known quantity for an unknown one. You are trading a champion you can evaluate for a draw from a distribution. If the champion in hand is already acceptable, most of that distribution is not an improvement.

A reroll changes the team, not just you. Whatever you were contributing — the only frontline, the only waveclear, the second engage — leaves with the champion you discarded. Week 6 takes this up properly; the point here is that the reroll is evaluated as a personal decision and is not one.

The choice, priced#

A worked case. You are given Azir. Your team is filling in around you: Azir, Maokai, Ashe, Seraphine, Renekton.

The interesting variable is not on this table, and it is the one that decides the answer: how well do you play Azir?

A champion’s strength is a property of the champion. Its value to you is a property of the pairing. A strong champion played at sixty per cent of its requirement is frequently worse than a modest champion played at full. The population data cannot see this, because it is averaged over players who are not you.

The trap: maximising#

The failure mode has a name outside this subject. A maximiser searches for the best available option; a satisficer takes the first option above a threshold and stops. In a setting where search is free, maximising wins. In one where search costs something and the distribution is wide, it usually does not.

Rerolling is search with a price. The player who rerolls an acceptable champion because a better one exists somewhere in the pool is paying a real cost for an expected improvement that is, most of the time, small — and is sometimes negative, because the roster contains worse outcomes than the one they discarded.

The discipline is to set the threshold before seeing the champion. Something like: I keep anything I can play competently that covers a function my team needs. Then apply it. A threshold invented after the roll is not a threshold, it is a justification.

When rerolling is right#

Three cases, and they are narrower than they feel:

1. You genuinely cannot play it. Not “I would prefer something else” — cannot. Self-knowledge, honestly applied, is the strongest reason on the list.
2. The champion is functionally redundant and the team has an obvious hole your reroll could plausibly fill.
3. You are holding several rerolls near the cap, in which case the future cost of using one has already been paid.

Note what is not on the list: the champion having a poor win rate.

The counterargument#

Some champions really are unplayable in this mode, and pretending otherwise in the name of decision discipline is its own kind of error. A melee champion with no gap-closer, dropped into a bridge full of ranged poke, is not a satisficing problem. It is a bad hand, and the reroll exists for it.

The argument here is not never reroll. It is that the button is pressed reflexively, on champions that clear the bar, by players who have not stated the bar. The correction is not to reroll less. It is to be able to say why.

Takeaway#

The reroll is an asset with a price, and it buys a draw rather than an outcome. Spend it against a threshold you set before you looked.