Signals · 2027 horizon
Phase H · post-Librarian

Ship a signal. Get paid.

Publish your own agent inside the Kabal framework. Users subscribe, the squad runs your signal alongside the core four, and you earn a cut of the PnL it generates. Proof of edge, priced by the market.

How the marketplace could look

Ƕ
@paulo_onchain
PEPE-season
"Farms frog-coins in the first 48h. Bails before the whales arrive."
+142%
30d
51%
Win
2.4
Sharpe
Fee 15% of PnL
↳ 84 subs
Ω
@vaporwave_maxi
Night Crawl
"Trades exclusively 2am–6am UTC. Built from Asia-session microstructure."
+61%
30d
46%
Win
1.8
Sharpe
Fee 10% of PnL
↳ 212 subs
Δ
@delta_silent
Contra-Whale
"Shorts tokens within 60s of top-10 holder rotation. Discipline over greed."
-8%
30d
39%
Win
0.4
Sharpe
Fee 20% of PnL
↳ 14 subs

Mock data. Stats are what the marketplace would display — rolling-window PnL, win-rate, Sharpe, fee schedule, subscriber count. All numbers trace to on-chain memo proofs for every signal.

How it works

1
Publish
Ship a signal agent through the Kabal framework SDK. Pass the structural lint + backtest floor.
2
Prove
30-day paper baseline on live pump.fun data. Same memo trail as the core squad.
3
List
Card appears in /signals with rolling stats. You set the fee schedule.
4
Earn
Subscribers run your signal. Sentinel still governs. Fee auto-splits on every realized PnL.

Money flow

When a subscribed signal makes +1 SOL

Subscriber realized PnL (gross)+1.0000 SOL
Signal author fee (example 15%)−0.1500 SOL
Kabal platform cut (3% of PnL)−0.0300 SOL
Subscriber net+0.8200 SOL

Losing trades pay zero fee. Fees only apply to realized PnL above a per-signal high-water mark (so a 1-SOL win that follows a 0.5-SOL loss nets fee on 0.5, not 1.0).

Kabal's framework stays the tracker. Signal authors don't get access to user wallets, don't see Sentinel config, can't override caps. Think of it as "Steam Workshop for trading agents" — the platform runs the engine, authors ship content.

Risk and discipline

Every published signal runs inside the same three-tier risk architecture. Sentinel can veto a subscribed signal's trade the same way it vetoes Reaver or Ash. Authors publish logic, not exposure.

Signals below a rolling Sharpe threshold (say 0.5) get auto-hidden from the marketplace. Signals with drawdowns past 15% auto-pause until the author re-certifies. The goal is a curated market of actually-working edge, not infinite-content slop.

Questions

Can any signal author drain my wallet?
No. Signal authors publish decision logic. The trade still flows through Sentinel and Aegis. The author's fee comes out of realized PnL after the trade settles — never from your principal.
What stops spam signals?
A 30-day paper baseline + structural lint + a small listing fee. The baseline alone filters 90% of low-effort submissions — guesswork doesn't survive a month of live pump.fun data.
Is this just Copy Trading rebranded?
Different shape. Copy trading mirrors another trader's wallet. Signal marketplace runs an agent inside your own Kabal instance, under your caps, your Sentinel, your memo trail. You never expose capital outside your own vault.
When?
2027. After Librarian auto-learning is stable (Q4 2026), so new signals can inherit the squad's learning primitives rather than reinventing them.