# The expert mindset
Stop evaluating runs by outcome (“I won, so I played well”). Evaluate by decision quality: given what you knew at the time, was each choice the highest-EV option? A run can be lost to variance after perfect play, and won despite sloppy play. Only by judging decisions, not results, do you actually improve.
# Decision theory under uncertainty
Every choice is a bet with incomplete information. Four ideas make those bets sharper:
Expected value, not best case
Weight an outcome by how likely it is. A 70-damage line that bricks half the time is worse than a 45-damage line that's guaranteed.
Plan for the variance
Build lines that are fine on an average draw, not just the perfect draw. Resilient plans beat fragile combos over a long climb.
Information has value
Scouting the map, reading full intent, and counting enemy HP all reduce uncertainty before you commit — gather it before deciding.
Sunk cost is sunk
The HP you already spent on an elite doesn't make finishing a doomed run correct. Re-evaluate from the current board state every time.
# Risk management & HP as currency
HP is not a survival bar to protect — it's a resource to spend. The question is never “can I avoid this damage?” but “is this the best place to spend HP for what I get back?” Trading HP for an early relic, a free remove from an event, or a faster kill on a scaling enemy is often the correct buy.
Bank HP
When the deck is still assembling and a boss looms. You can't spend HP you don't have.
Spend HP
On relics, removes, and faster kills when you can recover before the next checkpoint.
Race
Eat a hit to kill a ramping or summoning enemy this turn instead of blocking and letting it snowball.
# Reading the map
Pathing is planned at the start of an act, not improvised node by node. Look at the whole act first and route backward from the boss: what does my deck need to beat it, and which path delivers that?
Route priorities
- Line up a rest after an elite, not before.
- Bias to shops/events when the deck needs a remove or fix.
- Take more elites when your deck is ahead of the curve.
- Keep your options branching as long as possible.
Rest-site math
- Upgrade when full HP and the boss is a damage check.
- Heal when you can't safely reach the next fight otherwise.
- An upgrade on your key card often beats a small heal.
- Value rests more as difficulty and enemy damage climb.
# Potion economy
Potions are the most over-hoarded resource in the game. They expire worthless if you die holding them. The expert default is to use potions proactively to keep HP high and to convert close fights into clean ones — not to save them for an emergency that never comes.
- Hold a slot open before elites and bosses so you can pick up and use a clutch potion.
- Spend a potion to take an elite at full HP rather than limping in and dying.
- Use damage potions to kill scaling/summoning enemies a turn earlier.
- Don't save a potion for a hypothetical worse fight — the fight in front of you is the one that's real.
- Sell or use surplus potions before they sit dead through the act boss.
# Pivoting archetypes mid-run
Rigid players force the plan they wanted; experts draft the deck the run is offering. If your intended win condition never shows up but a different cluster keeps appearing, pivot. The cost of pivoting is the cards you've already drafted — but a coherent off-plan deck beats a half-built dream deck every time.
Signals to pivot
Your key payoff hasn't appeared by mid Act 1, a relic supercharges a different plan, or the shop is offering an obviously better engine than the one you're forcing.
How to pivot cleanly
Keep generically strong cards (block, removal, scaling) that serve any plan, and let go of dead niche pieces. Don't throw away the run chasing a third plan.
# Fight-by-fight target prioritization
In multi-enemy fights, the order you kill things in decides whether you take 5 damage or 35. Prioritize by the threat each enemy poses over the whole fight, not just this turn.
Kill first
Buffers and summoners that snowball, and low-HP enemies you can fully remove to shrink incoming damage.
Mitigate
Big single attackers you can't kill yet — drop Weak on them and block the rest.
Ignore
High-HP stallers that aren't the real threat. Don't waste burst chipping them while a buffer ramps.
# Habits of top climbers
- They plan the whole act before the first node, then adjust as info arrives.
- They use potions and HP as resources, not safety blankets.
- They skip far more cards than newer players — the deck stays tight.
- They re-read intent and recount lethal every single turn.
- They judge their own play by decision quality, then review losses honestly.
- They track patches: a single balance change can flip a card's evaluation, so they verify current values in-game rather than trusting old habits.
Take it higher
The Ascension guide covers how rising difficulty reshapes every decision above.