TLDR
Goldfishing tests whether your Commander deck can execute its plan without opposition. It is useful for spotting bad opening hands, missed land drops, color problems, awkward sequencing, slow engines, unreliable commander timing, and cards that repeatedly do nothing. Real games test everything created by opponents: interaction, pressure, target selection, politics, threat perception, recovery, and whether your deck fits the pod.
Use goldfishing to diagnose the machine. Use multiplayer games to learn whether that machine survives contact with three other players. Neither test replaces the other.
What Commander goldfishing can tell you
A goldfish is a controlled solo run. Draw an opening hand, make mulligan decisions, play turns, and record what happens as though the deck were allowed to develop without meaningful opposition. EDHREC similarly describes goldfishing as a way to check whether a deck runs smoothly and rehearse its lines, while emphasizing that it leaves out real interaction. EDHREC’s discussion of solo playtesting
That controlled environment makes goldfishing good at answering mechanical questions:
- How often does the deck keep a functional opening hand?
- Can it make its early land drops and produce the colors its spells require?
- When does its ramp actually accelerate the plan?
- How reliably can it cast the commander?
- Does card draw arrive before the hand empties?
- How many turns does an engine or win setup require without disruption?
- Which cards repeatedly sit in hand without a useful target?
- Does the sequencing force uncomfortable choices between development and holding up interaction?
These observations are especially useful after changing the mana base, lowering the curve, adding a new engine, or replacing a large package of cards. A repeated problem across several runs gives you a concrete lead. If five hands struggle to produce a second color, investigate the mana base. If the commander consistently arrives after the deck needs its effect, examine ramp, curve, and alternative plays.
For a repeatable process, use the 10-game Commander goldfish tracking method rather than judging the deck from one spectacular opening hand. The goal is not to prove that the deck is powerful. It is to find patterns you can act on.
What goldfishing cannot prove
Commander is normally a multiplayer format played with three or more players, so its usual game environment includes multiple opponents, targeting choices, attacks across the table, and competing plans. Wizards’ Commander format overview describes that multiplayer structure and the 40-life starting total. A solo run cannot reproduce those decisions.
An uninterrupted turn-six win setup therefore means only that the deck can assemble that setup by turn six under the tested draw. It does not establish an expected win rate, matchup quality, resilience, or appropriate power level. It also says nothing about whether opponents would recognize the setup on turn four and dismantle it.
Goldfishing is particularly weak at evaluating reactive cards. A counterspell, protection spell, board wipe, graveyard-hate piece, or political removal spell may look dead when there is no opponent creating a target. Cutting those cards because they underperform in solitaire can make the deck smoother in testing and much worse in actual Commander.
Solo combat is equally deceptive. You can calculate theoretical damage, including commander-damage lines, but you cannot forecast blockers, crack-back risk, removal, negotiated attacks, or the way attacking one player changes the other players’ decisions. Commander’s 21-damage loss condition tracks combat damage from the same commander, but knowing the arithmetic is different from successfully navigating a multiplayer combat step.
What real Commander games reveal
Real games test whether your deck can make good decisions while other decks resist it. They reveal whether your removal has useful targets, whether you can recover after a board wipe, and whether your win condition is still practical after an important engine disappears.
| Testing question | Goldfishing | Real games |
|---|---|---|
| Opening hands and mana | Strong diagnostic test | Adds matchup and turn-order context |
| Curve and sequencing | Shows internal friction clearly | Tests sequencing under pressure |
| Interaction | Checks whether mana can be held open | Shows whether targets and timing are useful |
| Resilience | Can simulate a chosen setback | Shows which setbacks opponents actually create |
| Combat | Calculates theoretical clocks | Adds blockers, retaliation, politics, and target choice |
| Power and table fit | Cannot establish either | Provides evidence when tested across suitable pods |
| Win conditions | Shows assembly speed | Tests concealment, protection, disruption, and execution |
Multiplayer games also expose your threat profile. A value engine may be mechanically efficient yet attract removal before it pays back its cost. A commander may make the deck so legible that opponents save interaction for it. Conversely, a slower-looking deck may receive enough time to assemble a strong position because the table identifies another player as the immediate threat.
These are not merely social complications pasted onto deckbuilding. They affect card choices. A commander that gets removed repeatedly may need cheaper protection, more independent engines, or a plan that functions without recasting it. Commander tax adds two generic mana for each previous cast from the command zone, so repeated removal can fundamentally alter the deck’s curve.
Do not mistake one bad game for a bad deck
Real games contain more useful information than goldfishing, but they also contain more noise. One game can be shaped by a poor matchup, an unusual draw, early attention from the table, or a mismatch in expectations. The Commander Rules Committee’s discussion of the format emphasizes its social contract and communication about the desired experience. Commander Rules Committee FAQ
That means a crushing loss does not automatically demand ten card changes. Record the cause. “Lost on turn seven” is less useful than “kept a two-land hand without selection, missed two land drops, and never cast the five-mana commander.” Likewise, “the deck was too strong” should become a specific observation about speed, tutors, resource acceleration, interaction, or how consistently the win condition appeared.
If the problem is expectation mismatch rather than mechanical failure, use a power-level checklist based on real-game behavior before weakening a coherent deck at random.
A practical two-stage testing process
Start with solo runs when the list is new or has received substantial changes. Track at least the opening hand, mulligan decision, first missed land drop, color access, commander timing, first meaningful draw effect, interaction availability, and the turn when the deck presents a credible win setup.
Then move into real games without immediately changing cards after every result. For each game, record one sentence in four categories:
- Development: Did mana, colors, draw, and sequencing work?
- Resistance: Which opposing play disrupted you, and could the deck reasonably answer or recover?
- Conversion: When you became established, could you turn that position into a win?
- Fit: Did the deck’s speed and play pattern match what the pod expected?
After several games, separate recurring deck problems from isolated events. Repeated mana failure deserves structural attention. Losing once to a narrow hate card may not. Building large boards without ending games suggests a conversion problem; the next step is reviewing how the deck plans to close Commander games, not adding another generic value engine.
Use each test for the question it can answer
The cleanest tuning rule is simple: trust goldfishing when it exposes repeated internal friction, and trust real games when they expose interaction, resilience, threat perception, combat, or table-fit problems. When the tests disagree, investigate why.
A deck that looks smooth alone but fails at the table may be too dependent on one permanent, too slow to hold up interaction, or too obvious when preparing to win. A deck that looks clumsy alone but performs well in games may contain reactive cards whose value only appears against opponents. Preserve those distinctions, change a small package at a time, and test again. That cycle produces better Commander decks than treating either solitaire speed or a handful of wins as the final verdict.