CargoX Relayer — Admin Analytics

TX Analyzer — minimum gas to win

Paste a relay TX hash to inspect competitors in ±2 surrounding blocks and see the minimum gas price that would still have won (or what was needed to beat the winner).

Market Gas — proactive scan

Periodic scan of recent contested winner TXs from the Document Minter. Aggressiveness picks the percentile, Safety margin scales it, and the cap pauses bidding when the market is too hot. Lookback window and safety margin are admin-configurable below.

Aggressiveness

Cap (gwei) — skip bidding above

Min winners/min

Safety margin ×

Lookback blocks (~? min)

Scan every (sec) (~? min)

Min gwei (floor)

Max gwei (ceiling)

Outliers

Mode

Contested detection

Mode

Volume regimes (docs/min thresholds)

Low threshold /min

Strategy

Override aggr

Safety ×

Min gwei

Max gwei

Lookback (~? min)

High threshold /min

Strategy

Override aggr

Safety ×

Min gwei

Max gwei

Lookback (~? min)

Override fields: blank = inherit global. Cap (gwei) and Min winners/min stay global — Cap is a market-broken safety floor, Min winners/min is superseded by Strategy=Pause.

Stale fallback

When market data missing

Regime preview — what each zone would bid right now

—

Bid formula per regime: percentile × safety margin, then clamped to [min gwei, max gwei] if set. The active regime is highlighted in blue.

Recommended bid

—

Cap status

—

Samples in window

—

Max of day

—

24h rolling

Percentile breakdown (contested winners only)

min	p25	median	p75	p90	p95	max	avg
Loading…

Scan refreshes every 2 minutes. The "Contested" filter excludes low-gas reserved-doc winners using a dynamic threshold (gas > median × 1.3).

Recent scans (live, last 30)

Time	Total	/min	Cont.	Outl	min	p25	p50	p75	p90	max	Blocks
Loading…

Dynamic Boost Monitor

Live state of the globally-learned contested gas price. Auto-refresh every 5 s.

Current price

—

Freshness

—

Queue

—

pending analyses

Fallback rate (1h)

—

Recent updates (last 20)

Timestamp	Outcome	Your gas	Reference	Sample	EMA value	Fast gas	TX / Reason
Loading…

Competitor Ladder

Who beats us most often in contested races, and by how much. Click a winner's reward address to open its relayer detail page; click a TX hash to open the TX Analyzer. Multi-relayer fleets (≥ 2 reward addrs sharing a signing wallet) get a 🚛 badge.

Window: Our relayer filter:

Top competitors

#	Reward addr	Label	Wins	Avg margin	p95 margin	Fleet	Method	Last seen
Loading…

Boost Tuning Advisor

Per-relayer overpay analysis — same-block (gas-lost) races only. Latency losses (winner mined first) are reported separately and excluded from the margin math because gas changes can't fix them. Positive margin = competitor paid more than us → consider boost UP. Negative margin = we paid more but lost the gas race → consider boost DOWN.

Relayer	Gas races	🐌 Latency	Wins	Win rate	p50 margin	p95 margin	Current boost	Suggested
Loading…

Recent matches (last 50; click a TX to inspect)

When	Our relayer	Our TX	Our gas	Winner	Winner TX	Winner gas	Block	Method
Loading…

Scout & API Probe — observe one CargoX queue idle

Polls getSignatures(rewardRecipient=ADDR) for any address at a chosen interval, records every doc seen along with times_shown transitions and queue dropouts. Read-only — never broadcasts a TX. Use to measure reservation TTLs without registering a scout wallet of your own.

Address (rewardRecipient)

Label

Interval (ms)

Quick-pick from our relayers: loading…

Queue position

—

CargoX-assigned rank

Live queue

—

docs in last poll

Samples

—

docs with TTL captured

Median TTL

—

reserved → contested / dropped

p95 TTL

—

longest 5% of holds

Live queue (last poll) — what CargoX shows for this address

Doc ID	times_shown	Reward (CXO)	From	Recipient	Nonce	Doc created	First seen	Last seen
No active probe.

Lifecycle log (recent observations)

Doc ID	First → Last `times_shown`	Reward (CXO)	First seen	Transitioned	Dropped	TTL
Start a probe to see observations.

Path (starts with /)

Query params (JSON object, optional)

Quick endpoints:

No probe sent yet.

How it works: burst-polls the chosen address at a fast interval (default 250 ms), hashes each response, and records the time between cache invalidation moments. After the run, the histogram shows the distribution of TTL windows and the simulation table predicts the fresh-data hit rate at common polling intervals. Read-only — never broadcasts a TX.

⚠ Address choice matters: we measure the time between observable response changes. If the queue doesn't churn (no docs arriving / times_shown not changing), we measure DOC ARRIVAL FREQUENCY in that queue, not cache TTL. Pick one of your own active relayer addresses (queue cycling frequently). Slow-reserveds or external addresses with rare activity give misleading TTL values.

Address (rewardRecipient)

Probe interval (ms)

Duration (s)

Label

Strategy

Shift (ms)

Jitter (± ms)

fixed: baseline measurement. Use this first to see raw cache TTL distribution.

Status

idle

Polls done

0

Strategy efficiency

—

% fresh / total

Median TTL

—

Current interval

—

live (mutable in adaptive)

Errors

0

TTL distribution histogram

TTL window	Count	Bar
No data yet — start a probe to collect TTL samples.

Polling simulation

Given the observed TTL distribution, what fraction of polls at each cadence would hit fresh data?

Interval	Req/s	Fresh rate	Fresh/s
Run a probe to populate.

Live event stream

Cache hits collapsed · click fresh rows to see full doc list

No events yet — start a probe to see live response stream.

Reserved Watch — cockpit idle

Probes every active relayer in parallel. For each reserved doc that transitions to contested or drops from the queue, the attribution scanner checks the relayer's signing wallet on PolygonScan for a TX to the Document Minter within the observation window. Per-relayer relay rate tells you who reliably fulfils their reservations.

Label

Interval per target (ms)

Top-N polling (optional)

+ discover every ticks

Targets — network candidate pool

Loading candidate pool…

Targets

—

relayers being polled

Polls done

—

total since start

Docs observed

—

unique reserved docs seen

Avg relay rate

—

% of reserved docs relayed by holder

Per-relayer cockpit

Relayer	Signing wallet	Queue #	Queue size	Polls	Reserved seen	Transitions	Relay rate	Samples	Last poll
No campaign running yet. Click Start to begin.

Recent observations (all relayers)

Watched relayer	Doc ID	`times_shown`	Reward (CXO)	First seen	Transitioned	Dropped	Holder relayed?
No observations yet.

Pre-warm latency analysis ● OFF — shadow poll at predicted broadcast moment

Enable analysis (costs +1 API call per scout-detected reserved)

Probe relayer

Reservation TTL (ms)

Lead time (ms)

Analysis disabled. Tick the checkbox to start scheduling probes.

Probes fired (24h)

—

Avg times_shown when found

—

closer to 1.0 = TTL well-tuned

Would have won

—

Pending probes

—

scheduled, not yet fired

Win rate by times_shown at probe (last 24h)

times_shown	Samples	Would-have-won	Win %
No data yet.

Recent probes

Doc ID	Scout	Reserved	Probe fired	Found?	times_shown	Winner	Would-have-won?
No probes yet.

Access Denied

Admin Analytics — Deep Dive

TX Analyzer — minimum gas to win

Market Gas — proactive scan

Percentile breakdown (contested winners only)

Recent scans (live, last 30)

Dynamic Boost Monitor

Recent updates (last 20)

Competitor Ladder

Top competitors

Boost Tuning Advisor

Recent matches (last 50; click a TX to inspect)

Scout & API Probe — observe one CargoX queue idle

Live queue (last poll) — what CargoX shows for this address

Lifecycle log (recent observations)

TTL distribution histogram

Polling simulation

Live event stream

Reserved Watch — cockpit idle

Per-relayer cockpit

Recent observations (all relayers)

Pre-warm latency analysis ● OFF — shadow poll at predicted broadcast moment

Win rate by times_shown at probe (last 24h)

Recent probes