Bidding Strategy

Bidding is how you stake NCT on your assertion. A good strategy aligns bid size with confidence so you can earn rewards without taking unnecessary risk.

What Bidding Controls?

• Higher bids can increase potential upside when you are correct
• Higher bids increase losses when you are wrong
• Bidding requires sufficient wallet balance to sustain the arbitration window

Core principles

Bid 0 on unsupported or low-confidence cases

If you return UNKNOWN, your bid should be 0.

If your signal is weak, consider:

• return UNKNOWN, bid 0
• or return a verdict with a conservative bid (only if your policy supports this)

Bid rules by verdict (required)

Each bounty includes a bid range in the webhook payload (min_allowed_bid and max_allowed_bid). Your bid must follow these rules:

• If you return MALICIOUS or BENIGN, you must place a bid within the allowed range (min_allowed_bid to max_allowed_bid).
  You cannot bid 0 for these verdicts.
• If you return UNKNOWN, your bid must be 0.

If you cannot justify a bid within the allowed range, return UNKNOWN instead.

Start conservative

When your Engine is new or recently changed:

• keep bids low until reliability and accuracy are stable
• increase only after you are confident in your signal quality

Map bid to confidence

A simple and safe mapping is:

• confidence low → small bid
• confidence medium → moderate bid
• confidence high → closer to max allowed bid

Keep the mapping consistent and explainable.

Respect bounty constraints

Bounties may define minimum and maximum bid rules. Your strategy should:

• never exceed max allowed bid
• avoid bidding below minimum if you want to participate
• adapt when constraints change

A Simple Starter Strategy (example)

• If unsupported type: verdict UNKNOWN, bid 0
• If scan failed or timed out: verdict UNKNOWN, bid 0
• If strong malicious signal: verdict MALICIOUS, bid near max allowed
• If strong benign signal: verdict BENIGN, bid low to moderate
• If ambiguous: verdict UNKNOWN, bid 0

This reduces risk while you build confidence in the Engine.

Tuning over time

As you learn:

• increase bids only in areas where you are consistently correct
• decrease bids in areas where you see mistakes or unstable behavior
• consider separate strategies per artifact type (file vs url)