AI-native hedge fund
Fund Control Room

Command Center

From idea to live capital, around the clock.

Live operations across every deployed strategy — capital, today's P&L, open positions, and the research engine feeding the book.

Illustrative platform demo — showcases Saleb's live execution & monitoring interface using sample strategies; specific strategy logic is proprietary and not disclosed. Forward-paper stage — not a representation of actual returns.
Fund equity & P&LiCombined mark-to-market value of all live strategies, ticking intraday. Illustrative of the funded book.Live · intraday
Research pipelineiHow ideas flow from generation through validation, paper-testing, and into the live book.ideas → live
Where strategies stand
Open positions
StrategyInstrumentSideEntryLastUnrealized
Research enginestreaming
Catalyst Radar — next 7 daysiScheduled events that move markets — earnings, central-bank decisions, crypto token unlocks, expiry. The engine knows which strategies trade or stand down into each.scheduled · click to expand
Portfolio RiskiFund-level risk across the live book — net exposure, cross-strategy correlation, portfolio volatility/VaR, and concentration. Illustrative of the funded book.live book · institutional
Net exposure by market
By family
Cross-strategy correlation
✓Data integrity: survivorship-bias-free · point-in-time · delisted-inclusive · 2017–2026
Production Roster

Live Book

The production-roster view of Saleb's Live Book — how a validated strategy appears once promoted to real capital, with its risk, P&L and status monitored live. Strategies shown are illustrative samples of the deployment workflow. Click any row for the detail.

Illustrative platform demo — showcases Saleb's live execution & monitoring interface using sample strategies; specific strategy logic is proprietary and not disclosed. The fund is at forward-paper stage — not a representation of actual returns.

Strategies in production

StrategyMarketSinceTradesWinP&LSharpeCapitalStatus
The Proving Ground

Paper Testing

Validated strategies forward-testing on simulated capital before they earn a place in the Live Book. When forward results confirm the backtest, they get promoted — when they don't, they're flagged for demotion. Paper testing is where the backtest meets reality, and not everything survives it.

Illustrative platform demo — sample strategies shown to showcase the forward-paper monitoring interface; specific strategy logic is proprietary and not disclosed. Not a representation of actual returns.

In forward paper

StrategyMarketPaper sinceTradesWinP&LSharpeCapital
The Evaluation Log

Run History

Every evaluation we — or the AI — ever ran, filterable.

Sourced live from the fund's real run ledger: strategy evaluations across NSE and crypto. This is the rigour-on-past-runs record — most ideas die here, and we name exactly why.

⚠
In-sample vs. reality — why the gauntlet exists
The chasm between a gorgeous backtest and what actually survives. crypto-trend — a real run from the ledger.
Gross, in-sample (the seductive backtest)
weekly Sharpe 0.93looks deployable
Net, on the never-seen 2025 holdout (reality)
OOS net Sharpe 0.08indistinguishable from noise
KILLED Caught by the Deflated-Sharpe gate: DSR 0.08 against a bar of 0.95 — once you correct for the many configs tried, the gross 0.93 is luck. The 2025 chop confirmed it. This is the gap our 11-gate gauntlet exists to catch.
Cause-of-Death ledgeriEvery strategy that died, bucketed by failure mode — derived from the real why_dropped notes in the ledger. Naming our failure modes is how we do real science. Click a bucket to filter the log below.
StrategyMarket / UniverseFamilySharpenetperm-pBH-qDSRNVerdictDateRun by
Strategy Library

The research catalogue, by family

Every strategy the engine has researched, grouped by trading family. Live and paper strategies are tagged; click any card for the full research record.

Strategy Workbench

Build → run → validate → promote

Configure a strategy as a declarative spec, run it, and watch it clear the validation gauntlet. The same spec retargets to any market with zero code change.

✦
AI Strategy Suggester
— search the research library in plain English, then run it below.
AI
Signal↓
Regime filter⚙ configure
Regime variable
Rule
Re-run to see the before/after impact of this filter.
Meta-label↓
Sizing↓
Costs realistic fills
2017-08 → 2024-12
discovery
2025-01 → now
🔒 reserved holdout
queued…
allocating workers
⚗️
Configure a strategy and hit Run.
Or take the Guided Tour to see how a strategy reaches the Live Book.
Signal overlay
Growth of ₹1cr vs benchmark
Validationi11 independent checks — leak audit, realistic fills, significance, deflated Sharpe, cross-validation, held-out data, and forward paper — that a strategy must clear before it can trade real capital.
🔒 Held-out data test
This strategy cleared discovery and full validation. Spend 1 of 3 reserved held-out queries for this family to confirm it on never-seen 2025 data — the final check before paper.
Run historyiterate & compare
#specSharpenetDSRiDeflated Sharpe Ratio — the Sharpe corrected for the number of configurations tried, so a high reading isn't just luck from many trials.verdict
Live specthe engine runs this
▲ This declarative spec is the only thing that changes when you retarget markets. Zero code change.
The Growth Engine

Autopilot — the AI org that works around the clock

CEO sets direction → Analyst proposes ideas → CTO validates on the 2-speed gauntlet → Workers run the sweeps → survivors are promoted. Flip the top-right toggle to Autopilot and the AI drives the Workbench itself.

Data Lake — the foundationiEvery figure here is queried live from the fund's own data lake (lake.duckdb + the options IV-surface) when the demo is built — nothing is typed by hand. This is the survivorship-clean, point-in-time data the engine actually researches on, re-audited on every refresh.
Activity feedlive
Pipelineideas → live
Engine status
AI
✓