paper trader - bounded tickets - no execution

Accuracy is not enough. Tradability is the test.

Eivra now converts agent probability edges into paper-only prediction-market tickets. Each ticket has a fixed maximum loss, a market-entry price, and a resolved P&L when the market closes.

source: liveselected bankroll: $5000edge gate: 5%max ticket loss: $10080 live

Would trade today

Live-only paper watchlist ranked by current strategy rules.

json feedBlocked

Unique live

Tradable

Global review

Tradable risk

$1700

Tradable EV

$686

9 global live paper markets need resolution review before any would-trade watchlist is trusted.execution: paper watch onlyreal money: disabledwatchlist review: 6selected strategy: selected-query

Strategy	Proof	Tradable	Review	Risk	EV
1. Mirror live, edge >= 10pp edge 10% / avg live edge 25.5%	Collecting	3	1	$300	$275
2. Sage live, edge >= 5pp edge 5% / avg live edge 27.5%	Collecting	4	0	$400	$194
3. Mirror live, edge >= 5pp edge 5% / avg live edge 15.8%	Collecting	4	0	$400	$192
4. Sage live, edge >= 10pp edge 10% / avg live edge 34.0%	Collecting	3	1	$300	$184
5. All agents live, edge >= 10pp edge 10% / avg live edge 26.0%	Collecting	3	1	$300	$154
6. All agents live, edge >= 10pp edge 10% / avg live edge 26.0%	Collecting	3	1	$300	$154

Market exposure digest

Groups open live paper tickets so repeated agent exposure is visible.

json feedObserving

Markets

Signals

Multi-agent

Top share

10.5%

Open risk

$3800

Open EV

$1693

Open live paper exposure is grouped by market so repeated agent bets are visible before review.threshold 35%review onlyreal money: disabled

Market	Signals	Agents	Risk	Share	EV
Will Anthropic make Claude Mythos broadly available before July 1, 2026? manifold / ai-tech / NO	4	4	$400	10.5%	$377
Will Andy Burnham lose a by-election in 2026? manifold / politics / NO	4	4	$400	10.5%	$237
Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers? manifold / other / NO	4	4	$400	10.5%	$164
MicroStrategy sells any Bitcoin by May 31, 2026? polymarket / crypto / NO	4	4	$400	10.5%	$45
Will Claude replace Grok on X in 2026? manifold / ai-tech / NO	4	4	$400	10.5%	$34
China arranges for permanent Iran peace deal in exchange for Nvidia chips manifold / ai-tech / NO	4	4	$400	10.5%	$33
Will Anthropic restore access to Fable 5 for US customers by the end of June? manifold / ai-tech / NO	3	3	$300	7.9%	$382
Will Abelardo de la Espriella be the next president in Colombia? manifold / politics / NO	2	2	$200	5.3%	$148

Configurable paper strategy

analytics only

Selected strategy

All agents live, edge >= 10pp

Resolved

P&L

-$73

Open EV

$181

Open loss

-$400

Worst bankroll

$4527

Open exposure is 8.0% of selected bankroll; remaining paper capacity is $100. This is analytics only; real-money execution remains disabled.

Break-even open wins

3 / 4

Expected after open

$108

Best after open

$256

30-day lab status

All agents live, edge >= 10pp

lab jsonDegraded

Proof days

30/30

Accepted P&L

-$73

Selected missed

$1164

Leaky rules

Proven rules

Positive thin

Resolver

resolver lag

Needs attentionday 30/30

primary action

Investigate overdue or unknown-close live paper markets before trusting open EV.

The evidence SLA has a warning, so repair or account for that before treating the proof window as clean.

watch Evidence SLA: Investigate overdue or unknown-close live paper markets before trusting open EV. Resolution catch-up: Run or restore the Eivra resolver, then publish a fresh paper snapshot.

Paper-only lock

Pass

execution disabled

Evidence SLA

Warning

Degraded

30-day capture window

Pass

30/30 complete days

Write mode

Collecting

Artifact only

Resolution catch-up

Warning

Resolver lag

Strategy registry

Pass

Synced

Agent-edge profitability

Collecting

Selected strategy tradability

Warning

48 skipped resolved / $1164.03 missed

Selected proof source lag

Pass

no direct canonical row

30-day proof gate

audit jsonaudit BlockedCollecting

Live days

26/30

Resolved

4/30

Gate P&L

-$73

Gate ROI

-18.3%

Max DD

$200

A strategy cannot become a candidate until it has 30 live calendar days, at least 30 resolved live paper tickets, positive resolved P&L, positive ROI, and drawdown under $500. This is a proof label only, not an execution signal.

Blockers

4 more live calendar days needed.26 more resolved live trades needed.

Audit blockers

8 checks still block capital review.

Blocked

Agent-edge audit

Collecting

Supabase

Proven rules

0/12

0 minimum sample

Positive thin

cannot count as proof yet

Missed P&L

$2491

not counted as proof

Resolution hygiene

blocked

9 needs review (9 overdue, 0 unknown close) / 38 open

Open EV is not realized profit; overdue or unknown-close markets must be resolved or investigated.

Resolved live trades

collecting

9/30

Best live rollup: Magpie live, edge >= 10pp.

Agent-edge profitability

collecting

0/12 proven rules (6 positive thin, 2 unresolved)

Keep collecting until each positive rule also clears the 30-day and 30-resolved-ticket gates.

Positive window P&L

collecting

$151.21

Judged only after enough captured days and resolved live trades exist.

Positive window ROI

collecting

16.8%

Judged only after the proof window is complete.

Drawdown limit

collecting

tracked in rollups

Drawdown blockers come from the durable proof gate.

Persisted proof log

audit json artifact json write readiness evidence sla lab status agent proof trade ledger published proof json feed

Capital review

Blocked

Evidence SLA

Degraded

Registry sync

Synced

Write mode

Artifact only

Proof evidence sources

Supabase persistence and GitHub artifacts are tracked separately so missing DB writes cannot hide the paper-only proof trail.

Blocked

Source	Status	Current	Target
Supabase proof log Dashboard rollups are reading append-only proof rows from Supabase.	Available	1000 rows / 30 complete days	30 complete days, current strategy registry captured
GitHub proof artifacts The scheduled workflow uploads a live proof bundle even when Supabase writes are disabled.	Active	05:22 UTC / 30d retention	self-contained proof bundle per workflow run
Resolution hygiene Open EV is split until overdue or unknown-close markets are reviewed.	Blocked	29 tradable / 9 review / 38 open	0 review-required live paper markets
Capital review gate Investigate 9 live paper markets needing review before trusting open EV.	Blocked	Blocked	operator review only, execution disabled

Artifact workflow

Artifact available

Latest usable proof artifact is paper-trading-proof-28227029792.

Latest run

28227029792

success / 19h ago

Latest artifact

Available

expires in 29d

Published proof

Published proof available

2026-05-27 / 15 live rows / 12 rules

readiness collecting

Artifact history

Collecting

1/30 days / 24 retained

23 same-day reruns ignored

Artifact action

Audit the latest downloaded artifact bundle before treating artifact-only proof as durable evidence.

.github/workflows/paper-trading-snapshot.ymlartifact paper-trading-proof-<run_id>files 5retention 30dartifact history 1/30 daysretained 24 / selected 1 / deduped 23audit npm run paper:artifact-audit -- ./paper-artifacts --json

Resolution review queue

Resolution hygiene is blocking the proof lab until the listed live paper markets are reviewed.

Review required

Current

Published

Unique

Provider

Action

review JSON

Market	Evidence	Agent	EV	Close
This Market will Resolve at number of Traders% manifold / open / overdue source market Source ambiguous / 28d ago	Current snapshot live	Hawk NO	$18.31	29d ago
Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers? manifold / open / overdue source market Source ambiguous / 26d ago	Current snapshot live	Magpie NO	$63.63	26d ago
Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers? manifold / open / overdue source market Source ambiguous / 26d ago	Current snapshot live	Mirror NO	$54.54	26d ago
Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers? manifold / open / overdue source market Source ambiguous / 26d ago	Current snapshot live	Hawk NO	$31.81	26d ago
Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers? manifold / open / overdue source market Source ambiguous / 26d ago	Current snapshot live	Sage NO	$13.63	26d ago
Will Abelardo de la Espriella be the next president in Colombia? manifold / open / overdue source market Source open	Current snapshot live	Sage NO	$77.27	26d ago
Will Abelardo de la Espriella be the next president in Colombia? manifold / open / overdue source market Source open	Current snapshot live	Hawk NO	$70.45	26d ago
Will the next full gemini model be as good as opus 4.7 or gpt 5.5 at coding? manifold / open / overdue source market Source open	Current snapshot live	Hawk NO	$27.35	9d ago

Resolver catch-up preview

Provider-resolved live markets are still unscored inside Eivra.

Resolver lag

Due

Unscored

Resolved

Eligible

P&L

$0.00

Run or restore the Eivra resolver, then publish a fresh paper snapshot.wins 0 / losses 0catch-up json

Capital review packet

Do not allocate capital; the 30-day paper proof gate has not cleared.

packet jsonNot reviewable

Decision

do not allocate capital

Captured

30/30

Resolved

9/30

Window P&L

$151

Review blockers

Investigate 9 live paper markets needing resolution review before trusting open EV.execution path: absentreal money: disabledearliest review 2026-06-26

Proof readiness checklist

Investigate 9 live paper markets needing resolution review before trusting open EV.

Blocked

Gate	Status	Current	Target
Persisted proof archive Stored snapshots must be readable before the lab can prove anything.	Pass	available	available
Paper-only lock The proof lab must never enable orders, wallets, leverage, or live capital.	Pass	execution disabled	execution disabled
Daily capture freshness Latest persisted capture is within the daily freshness window.	Pass	Fresh	fresh
Strategy registry sync Latest persisted proof day matches the current live strategy registry.	Pass	15/15 live	current live registry captured
Resolution hygiene Overdue or unknown-close open markets must be checked before treating open EV as credible. Blocking evidence This Market will Resolve at number of Traders%Hawk NO / $100 stake / EV $18.31 / close 29d ago / open 33.7d Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers?Sage NO / $100 stake / EV $13.63 / close 26d ago / open 28.2d Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers?Hawk NO / $100 stake / EV $31.81 / close 26d ago / open 28.2d Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers?Magpie NO / $100 stake / EV $63.63 / close 26d ago / open 28.2d Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers?Mirror NO / $100 stake / EV $54.54 / close 26d ago / open 28.2d	Blocked	9 needs review (9 overdue, 0 unknown close) / 38 open	0 review-required live paper markets
30-day capture window 0 missing, 0 partial.	Pass	30/30 complete days	30 complete days, 0 missing
Resolved live trades Best live rollup: Magpie live, edge >= 10pp.	Collecting	9/30	30 resolved live paper trades
Positive window P&L Waiting for enough captured days and resolved trades.	Collecting	$151.21	>= $1.00
Positive window ROI ROI is not judged until the proof window is complete.	Collecting	16.8%	> 0.0%
Drawdown limit Drawdown is not final until the proof window is complete.	Collecting	tracked in rollups	<= $500
Liquidity/slippage evidence This becomes a hard blocker before any profitable paper candidate can move to operator capital review.	Collecting	not persisted	source-level spread, depth, fees, fill size, and slippage-adjusted entry
Capital review boundary 21 more resolved live trades needed.	Collecting	Blocked	reviewable candidate, execution still disabled

Proof runway

Investigate 9 live paper markets needing review before trusting open EV.

Blocked

Earliest review

Jun 26

0d minimum

Capture days left

30/30 logged

Resolved left

38 open

Resolution pressure

9 review

9 overdue / 0 unknown close

Milestone	Status	Current	Target	ETA
30-day capture runway Earliest date assumes the daily snapshot keeps landing on schedule.	Reviewable	30/30 complete days	30 complete days	19h ago
Resolved trade runway 38 open live paper tickets can become realized evidence when markets resolve.	Collecting	9/30 resolved	30 resolved live paper trades	-
Resolution hygiene Overdue or unknown-close markets must be investigated before their EV is treated as actionable.	Blocked	9 needs review / 38 open	0 review-required live paper markets	in 2d
Capital review boundary 21 more resolved live trades needed.	Blocked	Blocked	reviewable candidate, execution still disabled	19h ago

Daily ledger

Complete

Complete days

30/30

Missing

Partial

Live variants

Days left

Date	Status	Rows	Live	Controls	Latest
2026-06-26	Complete	32	15/15	2	19h ago
2026-06-25	Complete	32	15/15	2	1d ago
2026-06-24	Complete	32	15/15	2	2d ago
2026-06-23	Complete	32	15/15	2	3d ago
2026-06-22	Complete	32	15/15	2	4d ago
2026-06-21	Complete	32	15/15	2	5d ago
2026-06-20	Complete	32	15/15	2	6d ago
2026-06-19	Complete	32	15/15	2	7d ago
2026-06-18	Complete	32	15/15	2	8d ago
2026-06-17	Complete	32	15/15	2	9d ago
2026-06-16	Complete	32	15/15	2	10d ago
2026-06-15	Complete	32	15/15	2	11d ago
2026-06-14	Complete	32	15/15	2	12d ago
2026-06-13	Complete	32	15/15	2	13d ago
2026-06-12	Complete	32	15/15	2	14d ago
2026-06-11	Complete	32	15/15	2	15d ago
2026-06-10	Complete	32	15/15	2	16d ago
2026-06-09	Complete	32	15/15	2	17d ago
2026-06-08	Complete	32	15/15	2	18d ago
2026-06-07	Complete	32	15/15	2	19d ago
2026-06-06	Complete	32	15/15	2	20d ago
2026-06-05	Complete	32	15/15	2	21d ago
2026-06-04	Complete	32	15/15	2	22d ago
2026-06-03	Complete	32	15/15	2	23d ago
2026-06-02	Complete	32	15/15	2	24d ago
2026-06-01	Complete	32	15/15	2	25d ago
2026-05-31	Complete	32	15/15	2	26d ago
2026-05-30	Complete	32	15/15	2	27d ago
2026-05-29	Complete	32	15/15	2	28d ago
2026-05-28	Complete	32	15/15	2	29d ago

Persisted agent edge proof

canonical stored rollups

Agent	Edge	Proof	Days	Resolved	Missed	Window P&L	ROI	Open risk
Sage latest 2026-06-26	5pp	Collecting Thin sample 24 more resolved live trades needed.	30/30	6/30 50% win / -$18.52	$338 15 skipped capacity leak	-$111	-18.5% daily PF 0.84	$400 EV $194
Sage latest 2026-06-26	10pp	Collecting Thin sample 25 more resolved live trades needed.	30/30	5/30 40% win / -$40.77	$447 10 skipped capacity leak	-$204	-40.8% daily PF 0.21	$400 EV $262
Hawk latest 2026-06-26	5pp	Collecting Thin sample 23 more resolved live trades needed.	30/30	7/30 71% win / $13.62	$781 19 skipped capacity leak	$95	13.6% daily PF 1.27	$400 EV $97
Hawk latest 2026-06-26	10pp	Collecting Thin sample 26 more resolved live trades needed.	30/30	4/30 75% win / $20.39	$641 12 skipped capacity leak	$82	20.4% daily PF 1.27	$400 EV $134
Magpie latest 2026-06-26	5pp	Collecting Thin sample 26 more resolved live trades needed.	30/30	4/30 50% win / $0.94	$42 12 skipped capacity leak	$4	0.9% daily PF 0.00	$500 EV $113
Magpie latest 2026-06-26	10pp	Collecting Thin sample 21 more resolved live trades needed.	30/30	9/30 67% win / $16.80	-$68 2 skipped capacity leak	$151	16.8% daily PF 0.90	$200 EV $126
Echo latest 2026-06-26	5pp	Collecting Thin sample 28 more resolved live trades needed.	30/30	2/30 50% win / -$34.03	$0	-$68	-34.0% daily PF 0.32	$0 EV $0
Echo latest 2026-06-26	10pp	Collecting Thin sample 29 more resolved live trades needed.	30/30	1/30 0% win / -$100.00	$0	-$100	-100.0% daily PF 0.00	$0 EV $0
Mirror latest 2026-06-26	5pp	Collecting Thin sample 24 more resolved live trades needed.	30/30	6/30 50% win / $5.00	$278 9 skipped capacity leak	$30	5.0% daily PF 0.63	$400 EV $192
Mirror latest 2026-06-26	10pp	Collecting Thin sample 22 more resolved live trades needed.	30/30	8/30 63% win / $13.73	$32 1 skipped capacity leak	$110	13.7% daily PF 0.70	$400 EV $330
Crowd latest 2026-06-26	5pp	Collecting No sample 30 more resolved live trades needed.	30/30	0/30 0% win / $0.00	$0	$0	0.0% daily PF -	$0 EV $0
Crowd latest 2026-06-26	10pp	Collecting No sample 30 more resolved live trades needed.	30/30	0/30 0% win / $0.00	$0	$0	0.0% daily PF -	$0 EV $0

Strategy	Days	Rows	Proof	Resolved	Skipped	Window P&L	Open risk
Magpie live, edge >= 10pp latest 2026-06-26 / live / live only window 2026-05-28 - 2026-06-26	30/30 100% coverage	63	Collecting Thin sample 21 more resolved live trades needed.	9 67% win / $16.80	2	$151 16.8% daily PF 0.90	$200
Mirror live, edge >= 10pp latest 2026-06-26 / live / live only window 2026-05-28 - 2026-06-26	30/30 100% coverage	62	Collecting Thin sample 22 more resolved live trades needed.	8 63% win / $13.73	1	$110 13.7% daily PF 0.70	$400
Hawk live, edge >= 5pp latest 2026-06-26 / live / live only window 2026-05-28 - 2026-06-26	30/30 100% coverage	63	Collecting Thin sample 23 more resolved live trades needed.	7 71% win / $13.62	29	$95 13.6% daily PF 1.27	$400
Hawk live, edge >= 10pp latest 2026-06-26 / live / live only window 2026-05-28 - 2026-06-26	30/30 100% coverage	63	Collecting Thin sample 26 more resolved live trades needed.	4 75% win / $20.39	17	$82 20.4% daily PF 1.27	$400
Mirror live, edge >= 5pp latest 2026-06-26 / live / live only window 2026-05-28 - 2026-06-26	30/30 100% coverage	62	Collecting Thin sample 24 more resolved live trades needed.	6 50% win / $5.00	12	$30 5.0% daily PF 0.63	$400
Magpie live, edge >= 5pp latest 2026-06-26 / live / live only window 2026-05-28 - 2026-06-26	30/30 100% coverage	63	Collecting Thin sample 26 more resolved live trades needed.	4 50% win / $0.94	14	$4 0.9% daily PF 0.00	$500
Crowd live, edge >= 10pp latest 2026-06-26 / live / live only window 2026-05-28 - 2026-06-26	30/30 100% coverage	62	Collecting No sample 30 more resolved live trades needed.	0 0% win / $0.00	0	$0 0.0% daily PF -	$0
Crowd live, edge >= 5pp latest 2026-06-26 / live / live only window 2026-05-28 - 2026-06-26	30/30 100% coverage	62	Collecting No sample 30 more resolved live trades needed.	0 0% win / $0.00	0	$0 0.0% daily PF -	$0

Captured	Strategy	Proof	Resolved	Skipped	P&L	Open risk
19h ago	All agents live, edge >= 10pp live / live only	Collecting	4	66	-$73	$400
19h ago	Crypto live, edge >= 5pp live / live only	Collecting	18	13	-$377	$0
19h ago	Sports live, edge >= 10pp live / live only	Collecting	3	2	-$300	$200
19h ago	Sage live, edge >= 5pp live / live only	Collecting	6	21	-$111	$400
19h ago	Sage live, edge >= 10pp live / live only	Collecting	5	13	-$204	$400
19h ago	Hawk live, edge >= 5pp live / live only	Collecting	7	29	$95	$400
19h ago	Hawk live, edge >= 10pp live / live only	Collecting	4	17	$82	$400
19h ago	Magpie live, edge >= 5pp live / live only	Collecting	4	14	$4	$500

Paper exposure ledger

selected config

Cap

$500

Accepted

Skipped

Skipped resolved

Missed P&L

$1164

Current risk

$400

Peak risk

$500

The selected strategy is replayed chronologically. A paper ticket is accepted only if its maximum loss keeps simultaneous open risk under the cap. Skipped signals remain visible so profitable-looking rules cannot hide capacity problems. Resolved skipped tickets count as missed P&L, not proof P&L.

Recent skipped signals

Will Anthropic restore access to Fable 5 for US customers by the end of June?

MirrorNOstake $100before $500EV $97.30skipped

Will Anthropic restore access to Fable 5 for US customers by the end of June?

HawkNOstake $100before $500EV $79.36skipped

Will Anthropic restore access to Fable 5 for US customers by the end of June?

SageNOstake $100before $500EV $204.92skipped

Will Bitcoin be exactly higher 7 days from now?

MirrorYESstake $100before $500missed -$100.00skipped

15 skipped signals have since resolved; they are capacity leakage, not realized proof.

Resolved P&L

$1736

111 tickets, 15.6% ROI on risked stake.

Raw open exposure

$3800

38 eligible tickets before selected cap.

Expected open P&L

$1693

Model-implied EV before market resolution.

Current leader

Hawk

$722 net over 30 tickets.

Resolution watch

Overdue

Open live

Tradable

Overdue

Closing 7d

Unknown close

Next close

in 2d

Tradable EV

$1320

Review EV

$373

Market	Agent	Side	Stake	EV	Close
This Market will Resolve at number of Traders% needs review / overdue / open 33.7d	Hawk	NO	$100	$18.31	29d ago
Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers? needs review / overdue / open 28.2d	Sage	NO	$100	$13.63	26d ago
Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers? needs review / overdue / open 28.2d	Hawk	NO	$100	$31.81	26d ago
Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers? needs review / overdue / open 28.2d	Magpie	NO	$100	$63.63	26d ago
Will Ukraine gain more territory than it loses this week, according to pro-Ukraine mappers? needs review / overdue / open 28.2d	Mirror	NO	$100	$54.54	26d ago
Will Abelardo de la Espriella be the next president in Colombia? needs review / overdue / open 32.7d	Sage	NO	$100	$77.27	26d ago
Will Abelardo de la Espriella be the next president in Colombia? needs review / overdue / open 32.7d	Hawk	NO	$100	$70.45	26d ago
Will the next full gemini model be as good as opus 4.7 or gpt 5.5 at coding? needs review / overdue / open 37.1d	Hawk	NO	$100	$27.35	9d ago
Will the next full gemini model be as good as opus 4.7 or gpt 5.5 at coding? needs review / overdue / open 37.1d	Magpie	NO	$100	$16.37	9d ago
Strait of Hormuz traffic returns to normal by end of June? tradable / closing next 7d / open 11.2d	Hawk	NO	$100	$8.97	in 2d
Will the San Antonio Spurs win the 2026 NBA Finals? tradable / closing next 7d / open 37.1d	Sage	NO	$100	$54.09	in 3d
Will the San Antonio Spurs win the 2026 NBA Finals? tradable / closing next 7d / open 37.1d	Hawk	NO	$100	$52.49	in 3d

Liquidity / slippage review

Source-level execution quality is a capital-review blocker, not paper P&L.

liquidity jsonBlocked

Sources

Blocked

Open signals

Open risk

$3800

Recent resolved

Adj. P&L

missing

Stress pass

5/10

Source	Open	Exposure	Resolved	Fields	Status
manifold 5 required fields missing	31 / 11 markets	$3100	41 / $43	missing	Blocked
polymarket 5 required fields missing	7 / 3 markets	$700	0 / $0	missing	Blocked

execution-friction stress

counts as proof: no

Rule	Resolved	Gross	Worst stress	Status
Magpie live, edge >= 10pp Magpie / edge 10pp	9	$151 16.8%	$129 250bps	Survives stress
Mirror live, edge >= 10pp Mirror / edge 10pp	8	$110 13.7%	$90 250bps	Survives stress
Hawk live, edge >= 5pp Hawk / edge 5pp	7	$95 13.6%	$78 250bps	Survives stress
Mirror live, edge >= 5pp Mirror / edge 5pp	6	$30 5.0%	$15 250bps	Survives stress
Sage live, edge >= 5pp Sage / edge 5pp	6	-$111 -18.5%	-$126 250bps	Loss making
Sage live, edge >= 10pp Sage / edge 10pp	5	-$204 -40.8%	-$216 250bps	Loss making
Hawk live, edge >= 10pp Hawk / edge 10pp	4	$82 20.4%	$72 250bps	Survives stress
Magpie live, edge >= 5pp Magpie / edge 5pp	4	$4 0.9%	-$6 250bps	Fragile profit

Persist source-level spread, depth, fee, and fill-size snapshots before any capital review.

Live-only proof

no backfill

Raw live risk

$3800

38 tickets before cap.

Live resolved

$1389

80 tickets scored.

This is the only section that should influence future real-money decisions. Backfill is useful for debugging the machinery, but the 30-day proof comes from live locked forecasts that resolve after the ticket existed.

Live leader: Hawk ($877)

Strategy lab

live variants

Strategy	Proof	Days	Resolved	Open	Skipped	P&L	Open EV
Sage live, edge >= 5pp Sage only. Tests whether base-rate first · slow to update produces tradable live edge at 5pp.	Collecting	28	6	4	21	-$111	$194
Sage live, edge >= 10pp Sage only. Tests whether base-rate first · slow to update produces tradable live edge at 10pp.	Collecting	28	5	4	13	-$204	$262
Hawk live, edge >= 5pp Hawk only. Tests whether contrarian · hunts mispricings produces tradable live edge at 5pp.	Collecting	26	7	4	29	$95	$97
Hawk live, edge >= 10pp Hawk only. Tests whether contrarian · hunts mispricings produces tradable live edge at 10pp.	Collecting	26	4	4	17	$82	$134
Magpie live, edge >= 5pp Magpie only. Tests whether snap forecaster · first instinct only produces tradable live edge at 5pp.	Collecting	28	4	5	14	$4	$113
Magpie live, edge >= 10pp Magpie only. Tests whether snap forecaster · first instinct only produces tradable live edge at 10pp.	Collecting	25	9	2	2	$151	$126
Echo live, edge >= 5pp Echo only. Tests whether market-prior · small bayesian steps produces tradable live edge at 5pp.	Collecting	2	2	0	0	-$68	$0
Echo live, edge >= 10pp Echo only. Tests whether market-prior · small bayesian steps produces tradable live edge at 10pp.	Collecting	1	1	0	0	-$100	$0
Mirror live, edge >= 5pp Mirror only. Tests whether cross-lab control · gpt-5 backbone produces tradable live edge at 5pp.	Collecting	29	6	4	12	$30	$192
Mirror live, edge >= 10pp Mirror only. Tests whether cross-lab control · gpt-5 backbone produces tradable live edge at 10pp.	Collecting	25	8	4	1	$110	$330
Crowd live, edge >= 5pp Crowd only. Tests whether ensemble · uniform avg of all agents produces tradable live edge at 5pp.	Collecting	0	0	0	0	$0	$0
Crowd live, edge >= 10pp Crowd only. Tests whether ensemble · uniform avg of all agents produces tradable live edge at 10pp.	Collecting	0	0	0	0	$0	$0
All agents live, edge >= 10pp Broad live strategy with a stricter edge gate across the house roster.	Collecting	26	4	4	66	-$73	$181
Crypto live, edge >= 5pp Category-specific strategy for crypto markets.	Collecting	29	18	0	13	-$377	$0
Sports live, edge >= 10pp Sports markets only. Keeps fast-resolving edge separate from politics.	Collecting	5	3	2	2	-$300	$107

Agent edge rules

canonical live matrix

Agent	Edge	Proof	Resolved	Open	Runway	Skipped	Missed	P&L	ROI	Open EV
Sage kelly capped	5pp	Collecting	6 50% win / BE 39%	4	Collectingcap 4 / gap 20	21 15 resolved	$338 capacity leak	-$111	-18.5% BE edge 11.2%	$194
Sage kelly capped	10pp	Collecting	5 40% win / BE 34%	4	Needs reviewcap 3 / gap 22	13 10 resolved	$447 capacity leak	-$204	-40.8% BE edge 6.0%	$262
Hawk kelly capped	5pp	Collecting	7 71% win / BE 55%	4	Needs reviewcap 3 / gap 20	29 19 resolved	$781 capacity leak	$95	13.6% BE edge 16.6%	$97
Hawk kelly capped	10pp	Collecting	4 75% win / BE 54%	4	Needs reviewcap 2 / gap 24	17 12 resolved	$641 capacity leak	$82	20.4% BE edge 20.6%	$134
Magpie kelly capped	5pp	Collecting	4 50% win / BE 45%	5	Needs reviewcap 4 / gap 22	14 12 resolved	$42 capacity leak	$4	0.9% BE edge 5.2%	$113
Magpie kelly capped	10pp	Collecting	9 67% win / BE 42%	2	Needs reviewcap 1 / gap 20	2 2 resolved	-$68 capacity leak	$151	16.8% BE edge 24.8%	$126
Echo kelly capped	5pp	Collecting	2 50% win / BE 52%	0	Starvedcap 0 / gap 28	0	$0	-$68	-34.0% BE edge -1.9%	$0
Echo kelly capped	10pp	Collecting	1 0% win / BE 39%	0	Starvedcap 0 / gap 29	0	$0	-$100	-100.0% BE edge -39.5%	$0
Mirror kelly capped	5pp	Collecting	6 50% win / BE 37%	4	Collectingcap 4 / gap 20	12 9 resolved	$278 capacity leak	$30	5.0% BE edge 12.8%	$192
Mirror kelly capped	10pp	Collecting	8 63% win / BE 40%	4	Needs reviewcap 3 / gap 19	1 1 resolved	$32 capacity leak	$110	13.7% BE edge 22.7%	$330
Crowd kelly capped	5pp	Collecting	0	0	Starvedcap 0 / gap 30	0	$0	$0	0.0%	$0
Crowd kelly capped	10pp	Collecting	0	0	Starvedcap 0 / gap 30	0	$0	$0	0.0%	$0

Agent-edge profit attribution

Resolved paper P&L grouped by source, category, and market.

attribution jsonCollecting

Resolved

P&L

-$11

Profitable

Concentrated

Diversified

Threshold

60%

Rule	Status	Resolved	P&L	Sources	Markets	Top market
Magpie / 10pp attribution collecting	Collecting	9/30	$151	1 top 100%	9	42% manifold
Mirror / 10pp attribution collecting	Collecting	8/30	$110	1 top 100%	8	46% manifold
Hawk / 5pp attribution collecting	Collecting	7/30	$95	1 top 100%	7	43% manifold
Mirror / 5pp attribution collecting	Collecting	6/30	$30	1 top 100%	6	57% manifold
Sage / 5pp attribution collecting	Collecting	6/30	-$111	1 top 0%	6	67% manifold
Sage / 10pp attribution collecting	Collecting	5/30	-$204	1 top 0%	5	57% manifold
Hawk / 10pp attribution collecting	Collecting	4/30	$82	1 top 100%	4	41% manifold
Magpie / 5pp attribution collecting	Collecting	4/30	$4	1 top 100%	4	93% manifold

Wait for more live paper markets to resolve before judging profit concentration.

Selected rule dossier

Magpie live, edge >= 10pp / supabase

dossier jsonCollecting

Days

30/30

Resolved

9/30

P&L

$151

ROI

16.8%

Open

1/2

Gap

Capture window

pass

30/30 days

target 30 complete live days

Resolved sample

collecting

9/30 tickets

target 30 resolved paper tickets

Positive P&L

collecting

$151.21

target >= $1.00

Positive ROI

collecting

16.8%

target > 0.0%

Drawdown

pass

$202.31

target <= $500.00

Open capacity

1 tickets pending resolution

Recent evidence

30 days / 9 resolved tickets

Missed P&L

-$68 excluded from proof

Next action

Review overdue or unknown-close open paper markets before trusting this rule's runway.

Agent-edge evidence timeline

evidence jsonAvailable

Days

30/30

Rule-days

360/360

Proven

Positive thin

Resolved

Window P&L

-$11

Rule	Captures	Resolved	Latest delta	P&L
Magpie / 10pp Positive, unproven	30/30	9/30	$0 2026-06-26	$151
Mirror / 10pp Positive, unproven	30/30	8/30	$0 2026-06-26	$110
Hawk / 5pp Positive, unproven	30/30	7/30	$53 2026-06-26	$95
Mirror / 5pp Positive, unproven	30/30	6/30	$0 2026-06-26	$30
Sage / 5pp Loss making	30/30	6/30	$0 2026-06-26	-$111
Sage / 10pp Loss making	30/30	5/30	$0 2026-06-26	-$204
Hawk / 10pp Positive, unproven	30/30	4/30	$53 2026-06-26	$82
Magpie / 5pp Positive, unproven	30/30	4/30	$0 2026-06-26	$4

Strategy daily evidence

capped 30d replay

Strategy	Latest	Days	Locked	Resolved	Open risk	Cum P&L	Avg/day
Sage live, edge >= 5ppCollecting Sage only. Tests whether base-rate first · slow to update produces tradable live edge at 5pp.	2026-06-17	5	10	6	$400	-$111	-$22.23
Sage live, edge >= 10ppCollecting Sage only. Tests whether base-rate first · slow to update produces tradable live edge at 10pp.	2026-06-17	5	9	5	$400	-$204	-$40.77
Hawk live, edge >= 5ppCollecting Hawk only. Tests whether contrarian · hunts mispricings produces tradable live edge at 5pp.	2026-06-16	5	11	7	$400	$95	$19.07
Hawk live, edge >= 10ppCollecting Hawk only. Tests whether contrarian · hunts mispricings produces tradable live edge at 10pp.	2026-06-16	5	8	4	$400	$82	$16.31
Magpie live, edge >= 5ppCollecting Magpie only. Tests whether snap forecaster · first instinct only produces tradable live edge at 5pp.	2026-06-17	6	9	4	$500	$4	$0.63
Magpie live, edge >= 10ppCollecting Magpie only. Tests whether snap forecaster · first instinct only produces tradable live edge at 10pp.	2026-06-19	8	11	9	$200	$151	$18.90
Echo live, edge >= 5ppCollecting Echo only. Tests whether market-prior · small bayesian steps produces tradable live edge at 5pp.	2026-06-02	2	2	2	$0	-$68	-$34.03
Echo live, edge >= 10ppCollecting Echo only. Tests whether market-prior · small bayesian steps produces tradable live edge at 10pp.	2026-06-02	1	1	1	$0	-$100	-$100.00
Mirror live, edge >= 5ppCollecting Mirror only. Tests whether cross-lab control · gpt-5 backbone produces tradable live edge at 5pp.	2026-06-19	7	10	6	$400	$30	$4.28
Mirror live, edge >= 10ppCollecting Mirror only. Tests whether cross-lab control · gpt-5 backbone produces tradable live edge at 10pp.	2026-06-19	8	12	8	$400	$110	$13.73
Crowd live, edge >= 5ppCollecting Crowd only. Tests whether ensemble · uniform avg of all agents produces tradable live edge at 5pp.	-	0	0	0	$0	$0	$0.00
Crowd live, edge >= 10ppCollecting Crowd only. Tests whether ensemble · uniform avg of all agents produces tradable live edge at 10pp.	-	0	0	0	$0	$0	$0.00
All agents live, edge >= 10ppCollecting Broad live strategy with a stricter edge gate across the house roster.	2026-06-16	4	8	4	$400	-$73	-$18.30
Crypto live, edge >= 5ppCollecting Category-specific strategy for crypto markets.	2026-06-19	4	18	18	$0	-$377	-$94.18
Sports live, edge >= 10ppCollecting Sports markets only. Keeps fast-resolving edge separate from politics.	2026-05-25	2	5	3	$200	-$300	-$150.00

Trader leaderboard

sorted by net paper P&L

Rank	Agent	Trades	Win %	Net P&L	ROI	Avg edge	Max DD
01	Hawk	30	73.3%	$722.34	24.1%	19.1%	$400
02	Crowd	22	63.6%	$611.16	27.8%	18.1%	$312
03	Mirror	16	68.8%	$372.72	23.3%	16.1%	$300
04	Sage	23	60.9%	$139.70	6.1%	25.8%	$400
05	Magpie	18	61.1%	-$41.73	-2.3%	23.2%	$332
06	Echo	2	50.0%	-$68.06	-34.0%	15.4%	$100

Scenario matrix

resolved only

Scenario	Trades	P&L	ROI
Edge >= 5pp, quarter Kelly capped avg edge 20.4% - avg stake $100	111	$1736	15.6%
Edge >= 10pp, quarter Kelly capped avg edge 27.4% - avg stake $100	73	$1082	14.8%
Edge >= 15pp, quarter Kelly capped avg edge 31.4% - avg stake $100	57	$1466	25.7%
Edge >= 10pp, flat $25 tickets avg edge 27.4% - avg stake $25	73	$270	14.8%

Open paper tickets

selected config

NOWill the San Antonio Spurs win the 2026 NBA Finals?

Sageedge -0.34stake $100max loss $100EV $54.09locked May 21

NOWill the San Antonio Spurs win the 2026 NBA Finals?

Hawkedge -0.33stake $100max loss $100EV $52.49locked May 21

YESWill Anthropic or OpenAI add a higher tier subscription >= 400$ in 2026?

Sageedge +0.25stake $100max loss $100EV $47.48locked May 21

NOWill the next full gemini model be as good as opus 4.7 or gpt 5.5 at coding?

Hawkedge -0.12stake $100max loss $100EV $27.35locked May 21

Daily evidence snapshots

latest 30d

Date	Locked	Resolved	Open risk	Day P&L	Cum P&L
2026-06-16 live only	1	1	$0	$53	-$73
2026-06-11 live only	1	1	$0	$74	-$126
2026-06-02 live only	1	1	$0	-$100	-$200
2026-05-21 live only	5	1	$400	-$100	-$100

Ticket math

A forecast becomes eligible when the agent differs from the market by at least 5%. Positive edge buys YES; negative edge buys NO.

Stake size uses quarter Kelly, capped at $100 per ticket. The whole ticket is the maximum loss.

Resolved P&L uses binary-share payoff: a correct YES bought at 0.40 earns 1/0.4 - 1 on the stake; an incorrect ticket loses the stake.

Paper only - no wallet, no broker, no execution path.

JSON feed