How to playtest a Commander deck before game night

A Commander deck can look perfect in a deck builder and still stumble the first time you shuffle it.

That is why you should playtest a Commander deck before game night. Not because goldfishing tells you everything. It does not. It will not show you who saves removal, who makes a deal, or who panics when your commander resolves.

But it will show you whether your deck can keep hands, cast spells, find mana, draw cards, and move toward a win before turn seven. If it cannot do those things alone, it will not magically do them at a real table with three opponents trying to stop you.

This article is for Commander players who already have a rough 100-card list and want a practical way to test it before buying missing cards, sleeving the deck, or blaming bad luck.

Key takeaways

• Goldfishing is solo playtesting: you play your first several turns with no opponents to test consistency.
• Run at least 10 opening-hand tests before making major edits.
• Track specific signals: lands, ramp, first play, commander timing, card draw, and a path to winning.
• Do not judge the deck by one perfect hand or one disaster hand.
• Solo testing catches structural problems; real games test interaction, politics, and resilience.

Start with a 10-hand goldfish test

The first test is simple: draw seven cards, decide whether you would keep the hand, then play the first five turns as honestly as possible.

Do that 10 times.

Ten hands is not a giant sample size, but it is enough to catch obvious problems. If seven of 10 hands miss early mana, you have a land or ramp issue. If six of 10 hands cast nothing meaningful before turn four, your curve is probably too high. If every hand needs the commander to survive for the deck to function, the list is fragile.

Use the same mulligan standard you would use in a real game. Do not keep a bad seven just because you are testing. Do not give yourself free redraws because the hand feels boring. The point is to see what the deck actually gives you.

For each hand, write down:

• starting land count
• whether you kept or mulliganed
• first meaningful play
• ramp available by turn three
• card draw or card selection by turn five
• whether you could cast your commander on curve
• whether the hand had a believable plan to win

That last point matters. A hand can technically cast spells and still do nothing.

Test the first five turns, not the whole game

Most solo Commander testing gets weird because players try to play a full game against nobody.

Do not do that at first.

The first five turns tell you most of what you need to know about structure. By then, a functional deck should have developed mana, started its engine, interacted with a hypothetical threat, or positioned itself to do something powerful next.

Ask these questions after turn five:

• Did I make my land drops?
• Did I spend mana efficiently?
• Did I have anything useful to do before turn three?
• Did my commander matter, or was I stuck waiting for it?
• Did I draw extra cards or run out of material?
• If an opponent presented a must-answer threat, did I have an answer?

You are not trying to prove the deck wins a goldfish race. You are checking whether the deck enters the game on time.

A lot of casual Commander decks fail here. They have exciting six-drops, splashy enchantments, and clever synergy pieces, but the first three turns are just land, tapped land, pass. That might be fine for one battlecruiser pod. It is not fine as a default plan.

Separate bad luck from bad structure

One bad hand does not mean the deck is broken.

Three similar bad hands probably mean something.

That distinction is where playtesting helps. Commander has variance built in: 100-card singleton, multiplayer pacing, uneven threat assessment, and opening hands that sometimes look like a prank. You cannot remove variance completely, and you should not try.

What you can do is look for repeated patterns.

Pattern 1: two lands and no ramp

If this keeps happening, count your real mana sources. Most Commander decks want roughly 35 to 38 lands, plus enough ramp to reach their important turns. A lower land count can work, but only if the curve is low and the deck has cheap card selection or ramp.

Be honest about tapped lands too. A hand with three lands is very different when all three enter tapped and your first spell costs three mana.

Pattern 2: all setup, no payoff

This hand ramps, filters, and maybe draws cards, but it never threatens anything.

That usually means the deck has too many enablers and not enough cards that actually advance the win condition. Cultivate is useful. Sol Ring is absurd. But mana is only good if it turns into pressure, engines, or answers.

Pattern 3: all payoff, no setup

This is the classic Commander trap: five cool cards you cannot cast yet.

If your opening hands keep showing expensive haymakers, trim the top end. A deck does not need 18 cards that cost six or more unless it is built to cheat them out or ramp very hard. For most decks, 6 to 10 expensive finishers is already plenty.

Pattern 4: the commander has to fix everything

If every good hand starts with "once my commander resolves," you have a resilience problem.

Commanders get removed. They get countered. They get taxed out of reach. Your deck can care about its commander without becoming useless when that card costs two more mana.

Run the "one disruption" test

After the basic 10-hand test, add one imaginary interruption.

Pick the turn where your deck starts to matter, then pretend an opponent removes the key piece. Maybe your commander dies. Maybe your first engine gets hit by Swords to Plowshares. Maybe someone destroys your mana rock with Beast Within.

Then keep playing for two turns.

This test is uncomfortable because it reveals the fake hands. The ones that look amazing only if nobody touches anything. Commander is not solitaire, even when you are testing alone.

Ask:

• Can I rebuild after one removal spell?
• Do I have another engine, draw spell, or threat?
• Did I keep too many cards that only work with one permanent?
• Does my deck fold if the commander costs two more?

You do not need every hand to pass. You do need the deck to show signs of recovery.

Test your mulligan rules separately

Goldfishing also teaches you which hands your deck should keep.

Some decks can keep two lands and a two-mana rock. Some cannot. Some decks need the commander early. Others can ignore it for five turns. A graveyard deck might keep Faithless Looting and two lands; a creature-combo deck might need a mana creature and protection.

Build a short keep/ship rule for your list.

For example:

• keep three lands plus ramp or card draw
• keep two lands only with a turn-two ramp spell and a play after that
• ship any hand with no action before turn four
• ship any hand with only payoff cards
• ship hands that cannot cast the commander or function without it

This is where solo testing connects to real games. You are training yourself not to talk yourself into bad sevens because the cards are technically powerful.

If you want a deeper hand-evaluation framework, read How to Mulligan in Commander after this. Mulligan rules and playtest notes work best together.

Check the curve after the hands tell you where to look

Do not start by staring at the mana curve chart for 20 minutes.

Start with hands. Then use the curve to explain what happened.

If you had no early plays in six of 10 tests, check how many real 1-3 mana cards you run. If you kept drawing too many expensive cards, look at the 5+ mana pile. If you missed colors, inspect the mana base instead of blaming variance.

A deck builder is useful here because it turns your notes into visible problems. In GrimDeck's deck builder, you can review the list, sort by mana value, check color requirements, and see whether the cards you keep drawing are clustered in the same awkward part of the curve.

Do not cut cards just because a chart looks uneven. Cut cards because the games showed you a pattern and the chart helped confirm it.

Playtest the closing plan

Once the first five turns are functional, test how the deck actually wins.

Start from a reasonable turn-six or turn-seven board. Give yourself a commander, a few lands, one engine piece, and three cards in hand. Then ask what happens next.

Can the deck close?

That sounds obvious, but a surprising number of Commander decks are built to do cool things forever without ending the game. They draw cards, make tokens, recur value, and then politely pass the turn until someone else wins.

Look for a closing plan that is specific enough to name:

• commander damage with protection and evasion
• a token board backed by an overrun effect
• a drain engine that scales with sacrifice loops
• a combo finish that the deck can find without playing the same game every time
• repeated value that eventually becomes lethal pressure

If your answer is "I guess I overwhelm the table," keep testing. That might be true, but vague win conditions are often just unfinished deckbuilding.

What not to overreact to

Playtesting is useful, but it can make you twitchy.

Do not cut a card because it was bad once. Do not add a land because one hand missed its third land drop. Do not rebuild the deck because a single goldfish line looked slow.

Avoid these overreactions:

• cutting all expensive cards after one clunky draw
• adding too many tutors because one game did not find the payoff
• removing interaction because no opponent existed during solo testing
• lowering the curve so much that the deck no longer has a way to win
• treating a perfect hand as the deck's normal speed

The goal is not to make the deck goldfish beautifully. The goal is to make it play real games better.

A simple Commander playtest checklist

Use this before game night:

1. Run 10 opening hands.
2. Track keep/mulligan decisions.
3. Play each kept hand through turn five.
4. Mark first meaningful play, ramp by turn three, and draw by turn five.
5. Note whether the commander was cast on time.
6. Pretend one key card gets removed and play two more turns.
7. Review repeated problems, not one-off weird draws.
8. Make 3 to 5 changes at most.
9. Play real games.
10. Revisit the notes after the deck has faced actual opponents.

That last limit is important. If you change 20 cards at once, you will not know what fixed the deck or what made it worse.

Final answer

To playtest a Commander deck before game night, goldfish at least 10 opening hands, play the first five turns honestly, track specific patterns, and make small changes based on repeated problems. Solo testing will not replace real pods, but it will catch the boring failures: bad mana, dead hands, no early plays, fragile commander dependence, and win conditions that never actually close.

If you want to make that process less fuzzy, build the list in GrimDeck's deck builder, run your test hands, and use the notes to trim the cards that keep disappointing you. The deck will still surprise you at the table. It just should not surprise you by failing to cast spells.