Machine Learning Power

Is Machine Learning good in Slay the Spire 2?

Machine Learning is a rare Defect power in Slay the Spire 2. In combat it resolves as: At the start of your turn, draw additional cards. It’s a dependable, middle-of-the-pack option, landing at C tier in our rankings.

Across 448 recorded runs, decks running it win 30.3% of the time, comfortably clear of the 14.3% average, by 16.0%. It’s a more selective pick, turning up in just 1.7% of decks that could run it. Offer data backs the pick — taken 42.2% of the time it is shown, with 33.6% for runs that took it against 25.5% for runs that did not, a +7.9% pick lift across 357 decided runs. Silent players keep it in 0.2% of their runs, so it’s a familiar sight in those decks. On average it shows up 1.2 times in the decks that take it.

It performs best next to Mad Science — together they post a +30.3% swing in win rate. It also has strong chemistry with the relic White Star, which moves its win rate by +29.2%. At 1 energy, it’s upgraded in 28.4% of runs and usually joins the deck around floor 17.

That’s 134 wins, 309 deaths, and 5 abandoned runs with it in the deck. The encounter that most often ends a run carrying it is Kaiser Crab Boss, which closes out 6.7% of its appearances. Read on for Machine Learning’s complete synergy, class, and threat breakdown to decide whether it earns a slot in your Defect run.

Pick or Skip: Machine Learning

Out of 918 recorded offers Machine Learning is taken 42.2% of the time, and taking it is worth +7.9% in pick lift — 33.6% for runs that picked it against 25.5% for runs that skipped, across 357 decided runs.

Taken 387 times out of 918 offers.

Pick rate when offered counts offers, not runs: out of every time the game put this in front of a player, how often it was taken. It is not the same number as the share of runs that end up containing it.

Pick lift is the sample-adjusted gap between runs that took the option and runs that passed on it. It is the fair way to rank choices, because raw win rates put the smallest samples on top.

When to Pick Up Machine Learning

Machine Learning is strongest when it joins the deck in Act 3, where runs win 61.9% across 21 runs — one of 3 recorded pickup windows.

Win rate of runs by the act Machine Learning entered the deck. Acts that have not reached the sample floor show their run count only.

Best Cards to Pair with Machine Learning

Ranked by win-rate lift, Machine Learning performs best alongside Mad Science (+30.3%), Energy Surge (+29.8%), and Supercritical (+24.5%).

Mad Science+30.3%28 runsEnergy Surge+29.8%31 runsSupercritical+24.5%54 runsModded+22.1%55 runsCreative AI+21.5%49 runsMulti-Cast+19.8%18 runs

Based on small co-occurring samples — treat these as early signals, not final verdicts.

Machine Learning Win Rate by Class

Machine Learning is played across 2 classes, led by Silent (72.7%) and Defect (28.9%) by win rate.

The Silent72.7%11 runsThe Defect28.9%427 runs

Best Relics for Machine Learning

Machine Learning pairs best with the relics White Star (+29.2%), Gold-Plated Cables (+27.5%), and Game Piece (+25.7%), ranked by win-rate lift.

White Star+29.2%23 runsGold-Plated Cables+27.5%16 runsGame Piece+25.7%24 runsRed Mask+25.5%78 runsCandelabra+25.2%55 runsUnceasing Top+24.7%22 runs

Based on small co-occurring samples — treat these as early signals, not final verdicts.

Machine Learning Card Details

Machine Learning is a rare power costing 1 energy; it’s upgraded in 28.4% of runs and usually picked up by floor 17.

At the start of your turn, draw additional cards.

What Ends Runs with Machine Learning

Machine Learning’s most common run-enders are Kaiser Crab Boss (6.7%), Test Subject Boss (6.3%), and Aeonglass Boss (5.6%).

• Kaiser Crab Boss6.7% of runs
• Test Subject Boss6.3% of runs
• Aeonglass Boss5.6% of runs
• Queen Boss5.1% of runs
• Infested Prisms Elite4.2% of runs

See Also

Compare Machine Learning with the rest of the best Slay the Spire 2 cards, relics, and classes below.