Worked examples
Calculate collateral, reward allocation, and a scoring result with explicit assumptions.
Collateral for a $20M portfolio
Assume one of your roles carries $20M of outstanding notional, the face value you are responsible for. Apply the marginal tiers. Each band uses its own rate, like income-tax brackets:
| Band | Calculation | Required collateral |
|---|---|---|
| First $1M | $1M × 5% | $50,000 |
| Next $9M | $9M × 3.5% | $315,000 |
| Remaining $10M | $10M × 2.5% | $250,000 |
| Total | Sum of the three bands | $615,000 |
Assume an illustrative oracle (on-chain price source) price of $0.25 per ASSET. The requirement is then 2,460,000 ASSET. This price is an assumption, not a current quote. If you post exactly that amount, you start at 100% collateral health, before reservations or other restrictions.
If the price falls to $0.20, the same stake is worth $492,000, or 80% of the requirement. Under the ordinary health thresholds, the account becomes Degraded. To restore 100%, you need another $123,000, equal to 615,000 ASSET at that price. For an SV, top-up grace can change the reported state, but raw collateral coverage stays at 80%.
One reward epoch
Make these assumptions:
- An illustrative, fully funded epoch (reward period) slice of 9,000 ASSET.
- No carryover or rounding residue.
- All three role floors are active.
- No eligibility reductions.
- Each role has at least three qualifying validators and $5M notional.
This is an allocation example. It is not the promised slice for the current three-hour epoch.
Each role first receives its 20% floor: 1,800 ASSET. That leaves 3,600 ASSET to divide by effective weight. Suppose total effective weights are 180,000 for TV, 90,000 for SV, and 90,000 for IV:
| Role | Floor | Share of remaining pool | Total |
|---|---|---|---|
| TV | 1,800 | 1,800 (50%) | 3,600 |
| SV | 1,800 | 900 (25%) | 2,700 |
| IV | 1,800 | 900 (25%) | 2,700 |
Within TV, take an operator with these values: $100,000 posted stake value, weighted stake age A = 12 months, N = 50 distinct assets served, and n = 1 counted offence in the last 12 months:
- Age multiplier:
A / (A + 12) = 0.5. - Experience multiplier:
1 + 2N / (N + 50) = 2. - Reputation multiplier:
max(0.5, 1 − 0.1n) = 0.9. - Weight before eligibility gates:
100,000 × 0.5 × 2 × 0.9 = 90,000.
Assume this operator is eligible and the two other TV weights are 45,000 each. Its share is then 50% × 3,600 = 1,800 ASSET. If the treasury backs part of the account, the account's reward is split further. The real calculation uses fixed-point arithmetic, snapshots, eligibility gates, and funding checks. Read Rewards.
One Brier evaluation
Assume 100 eligible assets, all predicted at 10% probability of default, and 30 observed defaults within the defined outcome horizon. Each asset has a scheduled payment inside the horizon, and the evaluation time has arrived.
- Observed score:
(30 × 0.9² + 70 × 0.1²) / 100 = 0.25. - Expected score:
0.1 × 0.9 = 0.09. - Implemented standard error:
sqrt(0.09 × (1 − 0.09) / 100) ≈ 0.028618. - With multiplier 2, the threshold is
0.09 + 2 × 0.028618 ≈ 0.147236.
Because 0.25 exceeds 0.147236, the evaluation reports a breach, and the provisional penalty process starts. One breach does not always take collateral: it counts on the penalty ladder, where early steps only jail and the debit applies at the slash step. This example uses the standard-error formula that the protocol implements. Other statistical calibration methods can use a different formula.
If the sample is below the configured minimum of 30, the protocol skips it. It does not count a skipped sample as a successful calibration. Assets without a scheduled payment inside the horizon are excluded. See Scoring Validator for outcome timing and Penalties and disputes for what follows a breach.