Economy in Numbers

Puts concrete numbers on what 08, 09, 14, 15 and 30 define as rules. All values are starting points, generated or checked by modelos/economia.py (output in modelos/economia-saida.md) and to be recalibrated with the alpha. Gold is the game currency (NPC shops, player shops and auction); it never becomes a token (14 §1). Nothing here describes real-money earnings; what the player receives from the token is a small, variable extra (14 §6). Casinos are on hold and do not appear. Reference scale: levels 1 to 30. Note: the Chapter 1 level cap is 20 (26 §4); the "21–30" columns and the "level 30" rows apply to the start of Chapter 2 (level 20–35) and stay here only as a reference. modelos/economia.py was recalibrated for Chapter 1: only the 1–10 and 11–20 bands exist (11–20 is the top), with the same per-hour values as this table, and the target band in §10 uses the level 11–20 regular player as its reference.

1. Gold per hour, by activity and level band

Realized value per hour of play (new NPC gold plus gold that only changes hands between players), before the sinks. In the last column, the share that is new gold (enters the economy); the rest is transfer between players and creates no gold.

Notes:

• Quests fall at the cap (from 300 to 150): the quest ends and gold comes instead from dungeons, events and crafts, which are repeatable.
• Time-weighted average (model, Chapter 1): 46 / 116 gold per hour realized in the 1–10 / 11–20 bands, of which 40 / 101 is new gold. (For the start of Chapter 2, outside the current model: 236 / 262 realized, 203 / 213 new gold in the 21–30 band and the cap.)
• Gold is fuel, not power. It serves for repair, food, travel and training. Power has a cap per tier (08); gold buys convenience.
• Daily cap of 5 h (20 §2.5): above it, monster gold and NPC sales yield 50%. It does not cut off those who play a lot, it only removes the incentive to sit 12 h at one spot.
• Model profiles: casual 20 h per month (50% of players), regular 60 h (35%), intense 150 h (15%). The reference earning week of a regular at level 11–20 (top of Chapter 1) is ~1,605 gold (60 h ÷ 4.345 weeks × 116 gold/h). Old reference, level 21–30: ~3,260 (Chapter 2).

2. NPC base prices

NPC prices are fixed (they neither rise nor fall with supply); only player shops fluctuate. Raw material sold by NPCs is bound to the account (08).

Lump-sum prices:

Guide ratio: food plus potion cost 9 gold per hour in the 1–10 band (≈ 22% of the hour's new gold) and 54 in 21–30 (≈ 27%). It is a felt sink without becoming a punishment.

3. Repair and durability loss

Based on 30 (weapon 100, shield 120, chest 150, legs 120, head 90, hands 70, boots 80, lantern 80; −1 per elite or boss strike, −0.25 per horde, −10 per piece on death).

• Formula: NPC price per point = 0.05 × item level (1.5 at level 30, as in 30). Player smith: 1 gold per point at tier 1, scaling with it.
• Full set from zero, level 30: 810 points × 1.5 = 1,215 gold at the NPC. Repair reduces max durability by 3% (NPC) or 1% (player smith); with a max below 40% the item becomes worn out (30). A well-kept piece lasts 20 to 60 repairs.
• Split in the model: 60% of repair at the NPC (pure sink: 9 / 36 / 68 / 81 gold per hour) and 40% with a player smith (transfer, pays the market fee). The smith is cheaper and loses less max, which gives a reason to look for a craft; the NPC is instant.
• Death in PvE: −10 per piece (80 points on a full set, ≈ 120 gold at level 30). Death in PvP: +50% wear (13).

4. Market fee (proposal and rationale)

Rationale:

1. Gold: burn only, nothing to the studio. Gold does not leave the game or become a token; splitting the fee with the studio would suggest external value for gold. Here the fee is a balancing sink.
2. Shop 3% < auction 5%: the shop requires setting up and stocking a storefront and gives life to the village; the auction is pure convenience. The difference pays for the living village.
3. The 1% deposit discourages listing everything "just to try" and clogging the base, without punishing those who really sell.
4. The fee is not the main sink. In the model it is ~1% of the sinks (§10): repair, food and travel weigh more. This is on purpose: a high fee punishes those who trade, and what we want is circulation.
5. Token 5%: keeps 14. With the studio's 5% royalty on resales of its NFTs, friction reaches 10%, high by NFT market standards. If the alpha shows a drop in volume, the (studio's) royalty is the first to go down, not the part that feeds the fund.
6. No tax on direct item trades, only the gold transfer fee (§6).

5. Land taxes

In token (15), except the instanced house, which is not an NFT and is paid in gold. Monthly percentage of the declared value (Harberger):

• The ranges stay within the 0.5% to 1% proposed in 15.
• Inactivity: 60 days without use (transactions with distinct players) doubles the tax; 120 days goes to auction (15). "Pasture" mode suspends it.
• Destination: 50% burned, 50% buyback (14). No appreciation language.
• Limits: per KYC, 3 urban plots, 3 rural and 1 concession; per clan, 2 concessions; per alliance, 3 (19).
• Example: on a plot declared at 100,000 tokens at 1%, the tax is 1,000 tokens per month. The owner declares the value at which they would accept selling; the tax should weigh on idle land, not on those who use it.

6. Gold transfer: cap and fee

The cap applies to gold given; buying and selling have an item in the middle and pay the market fee. The goal is to prevent a bot account from "unloading" millions of gold onto human accounts by direct transfer, the cheapest path. In the recalibrated model (Chapter 1, 200 thousand players, 15% bots), the 500-per-day cap does not even limit: each bot's stock in the 11–20 band stays below the cap, and with or without the cap the result is the same (8.19 weeks with lever x1.00). In the previous version (21–30 band) the difference was 3.96 to 6.09 weeks. The cap remains as protection against more productive bots; calibrate in the alpha.

7. Weekly production limits

Accumulable for up to 2 weeks (those who go days without playing do not lose their quota; 14 §8). Per character; what exceeds the limit yields ½ until the end of the week.

The limit does not stop play: above it the node still gives material, at ½ yield. What the cap does is prevent a 300 h-per-month bot from producing 10 times a human's output. Valuable resources respawn in real time with a cap (08); a concession produces by time, not by click.

8. Monthly withdrawal cap per account

Context: the placeholder in 08 and 14 is US$ 100 per month per verified account. The fund's average is ~14 tokens per active player per month, ≈ US$ 0.27 (tokenomics-saida.md). Someone who sells an NFT or a deed may receive much more.

Proposal: confirm US$ 100 per month, with a trust ladder.

Rationale:

1. US$ 100 is ~370 times the fund's average: it does not limit the ordinary player; it limits those who try to become a reward farm.
2. Keeps the token as an extra, not a salary (00, 14 §6).
3. Reduces the damage of laundering and fraud, which depend on volume per account (08).
4. The US$ 25 ladder for 90 days limits the damage of a fake account that passed KYC.
5. Open point (legal): the studio does not operate withdrawals (00 L1; conversion only through a licensed partner). The cap applies to what the studio controls: redemption from the season fund and the release of the on/off-ramp by the partner. Sales between players to their own wallet cannot be technically limited by the studio. This needs to be in the terms and the legal opinion (16).
6. Published in the terms of use, reviewed at most once per quarter.

9. Country list (approach)

Allowlist by function, not by game. The game (gold, in-game cosmetics, PvP, social) applies wherever the launcher is distributed; what changes is the real-economy package (wallet, token, market, NFT, withdrawal).

• Proposed start: Brazil as the only level A country in the closed beta, with verified 18+ testers (14 §9). Expansion country by country, by legal opinion; US and European Union last and only with advisory (SEC, MiCA; 16).
• Country by KYC document and declared residence, not just by IP; IP and VPN serve as a risk signal (a mismatch locks level A functions until review).
• State rules (US) or municipality rules may differ: review with legal.
• Living, versioned table, with date and source of the opinion; a change requires two signatures (legal and owner). Review every six months or on a change in law.
• A minor under 18 is level B in any country (08).

10. Gold simulation results

modelos/economia.py (standard library only; python3 modelos/economia.py > modelos/economia-saida.md). Monthly step, 12 months, 6 player cells (3 profiles × 2 bands, 1–10 and 11–20) and a block of bots. Recalibrated for Chapter 1 (max level 20); the per-hour values are those in §1 and §2.

Target band: gold in hand per active player between 1.0 and 3.0 weeks of earnings of a level 11–20 regular, that is, 1,605 to 4,814 gold, from month 4, and growth ≤ 2% per month from month 9 (rule in 14 §5). (The old band, 3,260 to 9,790, was for level 21–30.)

Base sink lever: new model parameter, a launch multiplier on repair, travel, training, mount, house and vanity (consumables stay at the table price). With the prices in §2 and §3 (x1.00) all three scenarios leave the band from above; with x1.25 all three enter. The numbers below are with x1.25 (5% bots, controller on); in parentheses, the result with x1.00.

The scenarios use a falling population, the hardest case: fewer people sharing the stock.

Findings:

• With x1.00 all three scenarios are outside the band: from 2.44 to 3.16 (20 thousand), 2.76 to 3.85 (200 thousand) and 3.07 to 4.87 weeks (1 million), from month 4 to 12. Cause: the 11–20 band yields 116 gold per hour (versus 236 in 21–30), so the reference week falls to almost half, but the stock per player falls less, because the lump-sum sinks of Chapter 1 are only the 1,200 training and the 1,500 mount (the 8,000 mount and the 3,000 training fall outside).
• Proposed lever: x1.25 (≈ +25% on repair, travel, training, mount, house and vanity). With it, from month 4 to 12: 1.51–1.76 (20 thousand), 1.70–2.20 (200 thousand) and 1.88–2.47 weeks (1 million), inside the band. The largest monthly growth from month 9 to 12 is −0.2%, 1.0% and −0.2%, within the 2%. The lever range that works in all three scenarios is narrow (x1.20 to x1.25; with x1.30 the 20 thousand and 1 million scenarios exceed 2% per month and with x1.40 the 200 thousand one does too): the base lever is a starting point, the controller does the rest.
• The 1 million scenario with x1.00 does not break the 2%-per-month rule (largest growth 1.7%; the 3.1% of the previous version came from the 21–30 band), but it breaks the band (4.87 weeks): this is what the x1.25 lever fixes. Months 1 to 8 are the initial filling (everyone is born with 50 gold); the growth of 140% to 160% in month 2 and of 28% to 57% in months 3 and 4 is that, not inflation.
• The per-hour balance is positive in both bands (+16 and +27 gold per hour between new gold and hourly sinks, with x1.00); balance comes from the lump-sum sinks (training, mount, house, vanity, 10% to 11% of the sinks) and from what the player stops spending. Without them, the stock only rises.
• The sink controller matters: with x1.00 and 0% bots, without the controller gold per player goes to 3.50 weeks and grows 6.5% per month in months 9 to 12; with the controller, 2.12 weeks and −0.8%.
• Bots remain the real risk. With controller, 200 thousand scenario, month 12, x1.25: 0% → 0.75 week; 5% → 2.20; 15% → 6.07; 30% → 14.59. With x1.00: 2.12 / 3.83 / 8.19 / 17.29. Without controller (x1.00): 3.50 / 5.52 / 10.28 / 19.97. With x1.25, the band tolerates ~5% bots; without the lever, not even that. Operational goal: bots below 5% of active accounts, with fast banning (the model bans 30% of the bot stock per month). The bots also stay in the 11–20 band (the best in Chapter 1), which reduces new gold per bot; their share of the faucet is 18% (5%), 42% (15%) and 63% (30%).
• Transfer cap for a new account (500/day) has no effect in the recalibrated model: with or without the cap the result is equal (see §6).
• Sink lever (controller): if gold per player grows more than 2% per month, hourly and lump-sum sinks rise 5% per month (up to ×1.6); if it falls more than 2%, they go down. Gold is never confiscated (14 §5).
• Sink composition at month 12 (200 thousand, x1.25): repair 46%, consumables 21%, travel 19%, progression 3%, banned bots 3% (stock destroyed), vanity 4%, market fee 1%, house 3%.

Model limits: fixed NPC prices (no price inflation between players), no return of dormant players (their stock, from 92 million to 3.2 billion gold at the end with x1.25, stays idle and does not enter the per-player figure), bots in a single block, laundering through auction zeroed (assumes the anti-fraud of 08), and all hours and rates per band are prototype assumptions. It serves to see the order of magnitude and what weighs, not to predict. The Chapter 2 sinks (8,000 mount, 3,000 training, 21–30 band) are not yet in the model.

11. What is left for the alpha

1. Measure real new gold per hour by activity and band and compare with §1.
2. Measure gold per active player against the band (1,605 to 4,814, level 11–20); above 2% per month, trigger the lever (repair and travel first). Launch with base lever x1.25 (or equivalent prices).
3. Test the transfer caps (500 and 2,000 per day) against real bot patterns.
4. Calibrate the withdrawal cap and the US$ 25 / US$ 100 ladder only after the legal opinion.
5. Record the country list in a versioned table, with the source of the opinion.
6. Redo the target band and the reference earning week (§1, §10) with level 20 as the top of Chapter 1, and run modelos/economia.py with the 1–10 and 11–20 bands as the only launch ones. Done: band of 1,605 to 4,814 gold, reference week of 1,605, model with both bands and base lever x1.25. The exercise remains to be repeated for Chapter 2 (21–35 band) when it enters scope.