Twenty hours into your RPG, a tester has 4.2 million gold and nothing worth buying. Every merchant is sold out of things they care about, the best sword came from a boss drop anyway, and the number in the corner of the screen has stopped meaning anything. Nobody broke your game. The economy just never had a drain.
This is the most common balance failure in games with currency, and it is also the most preventable. You don't need a simulation engine or a data scientist to catch it. You need a list of every way currency enters the game, every way it leaves, and a spreadsheet that converts both into the same unit: value per hour of play.
Faucets and sinks
The vocabulary comes from MMO economics but it applies to any game with a resource the player accumulates — gold, scrap, mana crystals, reputation, even time.
A faucet is anything that creates currency: quest rewards, enemy drops, selling loot, daily login bonuses, resource nodes. A sink is anything that destroys it: shop purchases, repair costs, crafting fees, fast travel tolls, respec charges, consumables.
The critical word is destroys. Trading gold to another player is not a sink — the gold still exists. Buying a sword from an NPC vendor is a sink, because that gold leaves the world permanently. Selling a sword to that same NPC is a faucet. If you get this distinction wrong on one line of your sheet, the whole model lies to you.
Two rules follow from it:
- Player-to-player trade is neutral. It moves currency around, it never reduces the total.
- Any faucet without a matching sink is inflation. Slowly at first, then all at once, because faucets usually scale with player power while sinks are often fixed prices set during the tutorial.
Step 1: inventory everything, in one table
Before you tune anything, write it all down. Most economy bugs are discovered here, not in the maths — you find the three faucets you forgot you shipped.
| Source | Type | Amount | Frequency | Gold/hour |
|---|---|---|---|---|
| Common enemy drop | Faucet | 12 | ~40/hr | 480 |
| Side quest turn-in | Faucet | 350 | ~2/hr | 700 |
| Selling junk loot | Faucet | 90 | ~6/hr | 540 |
| Chest in world | Faucet | 200 | ~1.5/hr | 300 |
| Total in | 2,020 | |||
| Potions & ammo | Sink | 45 | ~14/hr | 630 |
| Gear repair | Sink | 180 | ~2/hr | 360 |
| Fast travel | Sink | 25 | ~8/hr | 200 |
| Total out | 1,190 |
Net flow: +830 gold/hour. That is your real balance number, and it is the only one worth arguing about. Notice how ordinary each line looks in isolation. Twelve gold from a bandit is not a problem. Twelve gold from a bandit, forty times an hour, for sixty hours, is 28,800 gold that has to go somewhere.
Frequency estimates are where designers stall. Don't. A rough figure from watching one playtest is fine — you are looking for order-of-magnitude errors, not three decimal places. Mark the guesses so you know which numbers to firm up later. If you keep your balance and economy sheets in GameDesignerX, they live next to the items and mechanics they reference, so when a designer changes a potion's price the sheet is one click from the item entry rather than in a spreadsheet nobody remembers to update.
Step 2: decide what net flow you actually want
Positive net flow is not automatically wrong. It is wrong when it is unintentional. Pick a target and defend it:
- Net positive, aggressively (+40% or more per hour): power-fantasy games and short campaigns. The player should feel rich. Works when the game ends before inflation matters.
- Net positive, gently (+5% to +15%): most single-player progression games. Savings accumulate toward a meaningful purchase every few hours.
- Roughly neutral (±5%): survival and management games where currency is a tension mechanic, not a reward.
- Net negative in stretches: hardcore survival, or the late game of an extraction shooter, where losing ground is the whole point.
Write your target into the document as a design intent, not just a number. "Players should be able to afford one tier upgrade every 90 minutes of normal play" is testable. "The economy should feel fair" is not.
Step 3: run the twenty-hour inflation test
Take your net flow and project it. With +830/hour and no scaling, a player who plays 20 hours ends with roughly 16,600 gold banked beyond what they spent. Now compare that against your most expensive purchasable item.
If your best shop item costs 8,000, you have a problem: the player can afford everything by hour ten and the economy is decorative for the rest of the game. If it costs 200,000, you have a different problem — a grind wall that no reasonable player will clear.
Then add the part everyone forgets: faucets scale. Late-game enemies drop more. If your hour-40 enemy drops 60 gold instead of 12, your faucet quintupled while fast travel still costs 25. Project the curve at three points — early, mid, late — not just at launch:
| Hour | Gold in/hr | Gold out/hr | Net | Cumulative |
|---|---|---|---|---|
| 1–10 | 2,020 | 1,190 | +830 | 8,300 |
| 11–25 | 4,500 | 1,800 | +2,700 | 48,800 |
| 26–40 | 9,000 | 2,400 | +6,600 | 147,800 |
That third row is where the "4.2 million gold" screenshot comes from. The fix is almost never nerfing drops — players hate that and it makes combat feel worse. The fix is a sink that scales with the faucet.
Step 4: build sinks players choose, not sinks they resent
There are three shapes of sink, and only two of them feel good.
Percentage sinks scale automatically. An auction house that takes a 10% cut, a repair cost proportional to item value, a death penalty of 5% of carried gold. These stay balanced through the whole curve without any retuning, which is why they're the workhorse of long-lived economies.
Aspirational sinks are optional and expensive: cosmetics, housing, mounts, a legendary craft that costs a fortune. Players spend enthusiastically because the purchase is a choice and a flex. Aspirational sinks are the reason a game can survive a generous faucet.
Friction sinks — tolls, repair bills, ammo costs — are necessary in small doses and poisonous in large ones. They tax the act of playing. Keep them under about a third of total outflow, and never let a friction sink be the thing that stops a player from doing the fun activity.
A practical target for a long game: roughly 30% friction, 30% consumables and upgrades, 40% aspirational. Then check that the aspirational tier keeps producing new items as the player's income climbs.
A checklist before you call it balanced
- Every faucet and sink is written down, including ones outside combat (login rewards, refunds, quest-fail compensation)
- Player-to-player trade is correctly marked as neutral, not as a sink
- Net flow is stated as an intent, with a number and a sentence of reasoning
- The model is projected at three points on the progression curve, not one
- At least one percentage-based sink scales with income
- Most expensive purchasable item costs roughly 3–6 hours of net income at the tier it unlocks in
- No friction sink can block a player from the core activity
- Estimated frequencies are flagged as estimates, with a plan to verify them in playtest
Verify against real sessions
Spreadsheet balance is a hypothesis. The cheapest way to test it: log currency totals at fixed intervals during your next playtest — every 30 minutes, or at each checkpoint — and plot the actual curve against your projection. Divergence tells you which frequency estimate was wrong, usually in the first ten minutes of looking.
Record that comparison somewhere durable and linked to the build it came from, so the next tuning pass starts from evidence instead of memory. Whether that's a playtesting module, a balance sheet, or a shared doc matters less than the discipline: change one number, note why, re-measure.
The four-million-gold screenshot is not a mysterious late-game failure. It is a plus-830 that nobody wrote down.