Reliability Score

The maker reliability score is a metric that tracks how dependably a maker follows through on submitted quotes. Makers who consistently honor their quotes receive a bonus to their league ranking, while makers who frequently cancel are penalized. This page explains the formula, what counts as a cancellation, how the score affects RFQ routing, the tier system, and how to maintain a high score.

What the Reliability Score Measures

The reliability score captures three aspects of maker behavior:

• Response rate — How consistently the maker quotes on RFQs they receive.
• Fill rate — How often submitted quotes result in successful fills (i.e., the maker does not cancel before the taker executes).
• Quote quality — Indirectly, through the cancel rate. Makers who quote speculatively and then cancel have poor quality.

The primary input to the score is the cancel rate: the ratio of cancelled quotes to total quotes submitted.

How It’s Calculated

The reliability factor is computed using a weighted formula:

reliabilityFactor = clamp(1.1 - cancelRate * 1.5, 0.5, 1.1)

Where:

• cancelRate = number of cancelled quotes / total quotes submitted (a decimal between 0 and 1).
• clamp(x, min, max) = constrain x to the range [min, max].
• The coefficient 1.5 on cancelRate means even modest cancel rates have a noticeable impact.

Worked Examples

Key thresholds:

• 7% cancel rate drops the factor to approximately 1.0 — the neutral level where there is no bonus and no penalty.
• 20% cancel rate drops the factor to 0.8 — a meaningful 20% penalty.
• 40%+ cancel rate clamps at the floor of 0.5 — your effective score is halved.

What Counts as a Cancellation

A quote is considered “cancelled” under any of these conditions:

1. On-chain cancellation — The maker calls OptionsEngine.cancelQuote(quote), which marks the quote’s EIP-712 digest as used and prevents execution.
2. Bulk nonce invalidation — The maker calls OptionsEngine.incrementNonce(), which invalidates all outstanding quotes with nonces below the new on-chain nonce. Each invalidated quote counts individually.
3. Relay-side withdrawal — The maker sends a withdrawal message to the relay to retract a quote before the taker can execute it.

The Rolling Cancel Rate

The cancel rate is computed as a rolling metric over recent activity:

cancelRate = cancelledQuotes / totalQuotesSubmitted

Each cancellation increases the numerator while each submitted quote increases the denominator. Over time, consistent quoting without cancellations drives the cancel rate toward zero.

How It Affects RFQ Routing Priority

The reliability factor serves as a multiplier in the maker league ranking system:

effectiveScore = baseScore * reliabilityFactor

This means:

• A maker with a perfect reliability factor (1.10) gets a 10% bonus to their league score.
• A maker with a 0.50 reliability factor has their score halved.
• Two makers with identical base scores will be ranked differently based on their reliability.

Routing Implications

While all connected makers receive public RFQ_BROADCAST messages regardless of score, the reliability factor influences:

• League ranking — Higher-ranked makers are more visible to takers reviewing the leaderboard.
• Taker preferences — Informed takers reviewing the leaderboard may preferentially route private RFQs to makers with high reliability scores.
• Future routing features — The reliability score provides infrastructure for potential future features like priority routing or tiered broadcast ordering.

Score Tiers

Makers are grouped into tiers based on their reliability factor:

How to Check Your Score

Local Tracking

Track your cancel rate locally within your maker bot:

let totalSubmitted = 0;
let totalCancelled = 0;
 
function onQuoteSubmitted() {
  totalSubmitted++;
  logReliability();
}
 
function onQuoteCancelled() {
  totalCancelled++;
  logReliability();
}
 
function getReliabilityFactor(): number {
  if (totalSubmitted === 0) return 1.1; // no history, assume perfect
  const cancelRate = totalCancelled / totalSubmitted;
  return Math.max(0.5, Math.min(1.1, 1.1 - cancelRate * 1.5));
}
 
function getTier(): string {
  const factor = getReliabilityFactor();
  if (factor >= 1.05) return "Gold";
  if (factor >= 0.95) return "Silver";
  if (factor >= 0.75) return "Bronze";
  return "At Risk";
}
 
function logReliability() {
  const factor = getReliabilityFactor();
  const tier = getTier();
  const rate = totalSubmitted > 0 ? (totalCancelled / totalSubmitted * 100).toFixed(1) : "0.0";
  console.log(
    `[RELIABILITY] ${tier} | factor=${factor.toFixed(3)} | `
    + `cancel_rate=${rate}% | submitted=${totalSubmitted} cancelled=${totalCancelled}`
  );
}

Relay Health Endpoint

Monitor your maker’s overall activity via the relay REST endpoint:

curl http://127.0.0.1:8080/health

The response includes totalQuotes and connectedClients, which can be used alongside your local tracking for cross-verification.

Tips for Maintaining a High Score

// Conservative: 60-second deadline
const deadline = BigInt(Math.floor(Date.now() / 1000)) + 60n;
 
// Standard: 120-second deadline
const deadline = BigInt(Math.floor(Date.now() / 1000)) + 120n;

Score Boundaries

Recovery

If your score drops due to a spike in cancellations (e.g., an accidental incrementNonce() call), recovery requires consistent quoting without further cancellations. Since the metric is a ratio, submitting many new quotes without cancelling gradually dilutes the historical cancel rate.

For example, if you have 100 total quotes with 20 cancellations (20% cancel rate, factor = 0.80):

• After 100 additional quotes with 0 cancellations: 200 total, 20 cancelled = 10% rate, factor = 0.95.
• After 200 additional quotes with 0 cancellations: 300 total, 20 cancelled = 6.7% rate, factor = 1.00.
• After 400 additional quotes with 0 cancellations: 500 total, 20 cancelled = 4% rate, factor = 1.04.

Next Steps

• Maker Overview — How reliability fits into the broader maker framework.
• Submitting Quotes — Best practices for quote construction and deadline management.
• Nonce Management — How incrementNonce() affects your reliability score.
• Private RFQ Routing — How reliability influences taker preferences for exclusive routing.