CME proof ledger
Do the signals work yet?
Track record immatureThis page is where Convexly proves, over time, whether its Coherence Signals turned into useful settled outcomes. Today the honest answer is simple: the signal ledger is live, but the realized track record is still immature until enough markets settle.
What is live
Emitted ledger
Every signal is versioned and hash-linked.
What is not mature
Realized PnL
Performance claims wait for settled rows.
Next proof gate
Walk-forward
V0.2 is tracked under Manifest #288046.
Track-record caveat
Related-market consistency checks stay experimental until markets resolve. Positive realized-net rate, realized PnL, and matched controls govern any performance claim.
Evidence mechanics
Aggregate positive realized-net rate and cumulative gross + net PnL are public. Per-signal evidence, including resolved date, settlement direction, Z-score, and realized net PnL per row, is gated to Researcher.
Manifest #288046 tracks the V0.2 methodology. Each emitted row keeps its feed version, and /research/verify walks the SHA-256 chain that locks every row.
The resolution tracker runs daily at 12:00 UTC; this page refreshes within five minutes after new rows land.
Realized track record · last 90d
54.0% profitable(87 of 161 decided)
generated 2026-06-19
Emitted
4,075
signals
Resolved
161
4.0% of emitted
Pending
3,914
58 matured; awaiting settlement
Cumulative net PnL
-$228.8K
modeled, post-fees
Median time-to-resolution
—
across resolved signals
1. Emitted
ready4,075
Signals are written to the audit ledger with the methodology version and expected-resolution date.
2. Due soon
watch450
Inside the 14-day validation cohort or already past expected resolution.
3. Source settled
ready161
Only source-confirmed settlements move into realized PnL. Pending source markets do not count as wins or losses.
4. Claim permission
ready54.0%
Performance language stays gated until resolved rows create a denominator.
Lifecycle ledger
Emitted
4,075
written to signals_emitted
Pending
3,914
no source-confirmed outcome yet
Checkpointed
450
short-horizon validation cohort
Matured
58
past expected resolution
Resolved
161
source-confirmed rows
Evaluable
161
profit/loss denominator
Invalid/voided
0
excluded from profitability
Disputed
not wired
not separately wired yet
Stale
not wired
not separately wired yet
How it adds up: profitability rate counts resolved rows with positive net PnL; voided + ambiguous resolutions are excluded. A signal with zero PnL is treated as not-profitable (the trade was take + close at the same price). Net PnL is post-fee, post- slippage as modeled at signal-emit time.
Emitted
4,102
cumulative ledger
Resolvable
3,956
146 excluded (partial)
Resolved
161
4% of resolvable
Positive rate
54.0%
87 positive net rows
Resolved net PnL
-$228.8k
gross -$227.6k · resolvable cohort
Resolution-quality denominator
3,956 of 4,102
146 partial-provenance rows excluded
Proof ladder
What has to happen before CME can make stronger claims.
This is the plain-English lifecycle from structural signal to performance evidence. If a step is blocked, the public claim stays conservative.
1. Emit
readySignals are written to the ledger
4,102 emitted
This proves the pipeline is producing versioned candidates. It does not prove those candidates made money.
2. Mature
collectingMarkets reach settlement windows
3,956 resolvable
Only full-provenance rows with enough settlement linkage can enter realized-performance math.
3. Resolve
readySource-confirmed outcomes land
161 resolved
This is the current proof bottleneck. Until rows resolve, modeled EV remains a review signal rather than a performance claim.
4. Compare
pendingMatched controls and failures publish
1/30 primary pairs
Convexly needs primary K=20 measured matched-control pairs, visible failures, and formal paired analysis before stronger public language is allowed.
5. Upgrade
blockedClaim language can improve only after evidence
4.1% resolved
The product graduates by published outcome and matched-control evidence, not by shipping a prettier dashboard or a higher modeled-EV number.
146 signals have partial provenance and are excluded from positive realized-net rate, resolved PnL, and any other resolution-quality metric on this page.
These rows were reconstructed from a sanitized public archive after the original private archive went uncommitted; their condition_id fields are NULL so the resolution-tracker cannot match them to Polymarket settlements. They count toward Emitted only. See the audit chain for the full provenance breakdown per row.
Per-signal evidence — Researcher tier
161 resolved signals on the ledger.
The aggregate view above is open: how many signals have fired, the positive realized-net rate, and cumulative net PnL establish the credibility loop. The per-row methodology evidence — resolved date, settlement direction, Z-score, net EV, realized net PnL — is exclusive to the Researcher tier so paying customers retain the methodology disclosure they pay for.
How this ledger is locked
- Append-only: the signal-emission ledger is service-role-only and append-only. Each row carries a SHA-256 input hash bound to the canonical-JSON serialization of the signal at emit time.
- Resolution-locked: signals_resolved is FK + PK against signals_emitted. The resolution-tracker cron writes one row per resolved signal with a SHA-256 resolution hash chaining the outcome to the original input hash.
- Methodology status: Each emitted row carries its own feed version. The V0.2 threshold + sizing + cost model + metrics were frozen ex-ante at git SHA 8616a63 per Manifest #288046.
- Verifiable: walk the chain in your browser at /research/verify.