Torrent Gear

Balance of chance and progress: to be "honest"

1) What the player calls "fair"

"Honesty" in gamification is a combination of three things:

Predictability of effort: I understand how much action there is before the award.
Equality of chances: my chance is no worse than that of others on equal terms.
Rule transparency: I know what affects chance/progress and why.

The player suffers "bad luck" if he sees insurance (guarantees) and alternatives (paths through progress).

2) Axis "Chance ↔ Progress"

Chance (RNG drops): high emotional amplitude, fast peaks of pleasure, but the risk of frustration with long "dry" series.

Progress (deterministic milestones): low amplitude, but stable motivation "one more step."

Conclusion: use a hybrid: chance for a spark of emotion, progress to guarantee the result.

3) Basic balance box (HAR: Hybrid, Assured, Rate-limited)

1. Hybrid: Each valuable award has both paths: a random drop or cumulative progress.

2. Assured: after N attempts without success, insurance is included: pity timer/bad-luck protection.

3. Rate-limited: There are tempo/limits so that the elite do not "suck out" the entire pool and break the economy.

4) Insurance against failures (pity timers)

Options:

Hard pity: guaranteed reward on the Nth attempt (for example, on the 50th - frame).
Soft pity: the chance grows after each failure (for example, + 2% to the base chance for each attempt without a drop).
Token pity: Tokens/shards are issued for each failure; they can be exchanged for a reward.

Configuration rules:

Base chance p₀: 1-5% for epics, 0. 2-1% for legends.
Growth cap: so that soft pity does not go> 40-50% without a determinant.
N hard pity: target 95th percentile of time/attempts (example: 95% of players will receive an award ≤ 21 days).

5) Progress milestones

Linear: collect 100 points - get a reward.

Ladder: small prizes every 10 points, large - by 100.

Sets: close 5 thematic missions → metanagrade.

Seasonal tracks: free branch + premium branch of cosmetics (without affecting the matwedding of bets).

Anti-crowding: enter task variability (different providers/modes), daily mouthguards and boosts on "scarce days."

6) Transparency and control of perception

Show the chance growth formula in simple terms: "Each failure + 2% to the chance, no more than 1 times 10 minutes."

Separate screen "How it works": basic chance, limits, pity logic, deadlines.

Player logs: "You scored 14 tries, the current chance is 29%, guaranteed on the 35th."

7) Emission economics (so as not to "oversupply")

Targeted rarity coverage:

Rare: 25-35% active per season
Epic: 8–12%
Legendary: 2–4%
Issue budget: Plan for the number of "instances" or equivalent costs of progress.
Rotation: after the season - to the archive; rare "retro comebacks" on special occasions.

8) Technical "honesty" patterns

Trigger idempotency: event → one accrual, retrays are safe.

Deferred validation: check complex conditions asynchronously, but atomically fix the reward.

The server decides RNG, the client only visualizes (no client sides).

9) Anti-abuse

Pace cap: no more than X tries in Y minutes, cold window after fast pharma.

Variety of actions: part of the progress is available only when changing the mode/provider/bet.

Batch deduplication - Repetitive micro sessions at precise intervals are not making progress.

10) Psychology of "honesty" in the interface

Fidbeck on failure: "+ 1 shard to guarantee"; progressive bar pity to guarantee.

Fidbeck for success: low-noise, brief VFX; the "show off" button is optional.

Tonality: Avoid marketing ambiguity, don't promise "almost guaranteed."

11) Control metrics (not just retention)

Gini by award: The lower, the more evenly distributed with similar efforts.

P95 time to reward: keep within target (e.g. ≤ 21 days for epic).

Bounce after dry streak: The proportion of players who have left after a long losing streak is a major red flag.

Equip rate/Equip duration: whether the award has been received and worn.

Token burn rate: whether shards/tokens are spent or accumulated (a sign of fear of "spending unprofitable").

Report/complaint rate on "injustice."

12) A/B that makes sense to run

1. Hard vs Soft pity at fixed p₀.

2. Distance between milestones: 8-10-12 steps.

3. Token insurance option: 1 shard per attempt vs shards only after X attempts.

4. Pace cap: limits by minute vs by day.

13) Example of settings (pseudo)

Base chance and soft pity:


p0 = 0. 02 # 2% base chance delta = 0. 015  # +1. 5% for failure p_cap = 0. 45 # maximum 45%
cooldown = 10m # attempts not more than 1 in 10 minutes

Hard pity:


if attempts_without_drop >= 30:
grant_reward()

Token pity:


shard_per_fail = 1 cost_to_craft = 25 # guarantor for 25 failures

14) 'Honesty' checklist ahead of release

There are both paths: random drop and deterministic milestones.
Soft/hard pity and chance caps are configured.
Transparent screen "How it works" + log attempts.
Tempo limits and anti-abuse.
Time-to-award test P95 synthetic sessions.
Rotation/issue fits into the seasonal budget.
UX to failure/success quiet, respectful, no manipulation.

15) Ethical frame

Rewards offer no mathematical advantage in games.

Honest probabilities without "hidden weight."

Easy ability to turn off effects and hide the showcase.

Understandable exit paths: pauses, limits, links to Responsible Gaming.

Fairness is the control of the variance of impressions. Add emotion through chance, but record the result through progress and insurance. Keep the issue in the budget, make the rules visible and check the P95 of time before the award - then the system is perceived as "honest," motivates to continue and does not break the economy of the season.