Evidence over narrative
Hypotheses are tested under explicit data, leakage, reproducibility and out-of-sample controls before they can support an investment decision.
SARVĀSVAM VENTURES
Investment · Quantitative Research
Research before capital.
We build, challenge and govern systematic investment research before proprietary capital is placed at risk.
CapitalProprietary
MethodSystematic research
GovernancePhase-gated
BaseMumbai, India
Scroll to enter
01 · The firm
Built around process, not prediction.
Markets are uncertain by construction. Our response is not to make forecasts sound certain, but to build research systems that make assumptions visible, weak ideas easier to reject and investment decisions easier to audit.
Construction before conviction
Signals matter only after costs, sizing, interactions, concentration, liquidity and implementation constraints are considered.
Survival before optimisation
Risk is a design constraint across research, allocation and monitoring—not a statistic appended after performance.
02 · Research engine
Every idea earns the right to advance.
The research process is deliberately staged. Progression is controlled, negative evidence is preserved, and an attractive backtest is never treated as sufficient evidence by itself.
Research governance onlineControlled progression
Observe
Identify a market behaviour, constraint or anomaly without assuming it is tradable.
Hypothesise
State the economic or behavioural premise before tuning implementation.
Experiment
Register assumptions, datasets, costs, variants and evaluation methodology.
Challenge
Attack leakage, fragility, sensitivity and competing explanations.
Validate
Require evidence beyond the sample in which the idea was discovered.
Admit
Promote only research that satisfies the applicable governance gate.
Allocate
Capital is considered only after research admission and portfolio-level constraints.
Observe · controlled state
Markets change.
Models can fail.
Evidence must survive both.
03 · Research integrity
A research record should remember what did not work.
Institutional discipline depends on traceability. Failed hypotheses, search iterations and control failures are part of the evidence—not material to be hidden when the narrative changes.
Illustrative governance states
Research memory
HYP-0142Economic premise challenged by cost stressRejected
HYP-0143Robustness evidence under reviewReview
HYP-0144Control conditions satisfied for next gateAdmitted
Illustrative interface only · no live strategy data shown
Control stack
Evidence before authority
01Reproducibility
Material experiments must be recoverable and independently rerunnable.
02Out-of-sample discipline
Validation boundaries are treated as governance controls, not optional diagnostics.
03Negative evidence
Rejected work remains part of the permanent research record.
04 · Risk
Risk is a system property.
Capital preservation begins before a position is opened. Research validity, implementation quality, portfolio interaction and operational control all contribute to the risk of an investment system.
Model fragility
Parameter sensitivity, leakage, unstable assumptions and weak generalisation are challenged before capital allocation.
Exposure discipline
Position sizing, concentration, liquidity and regime dependence are treated at portfolio level.
Implementation realism
Costs, slippage, timing and operational mechanics must survive realistic execution assumptions.
Controlled authority
Research, approval, deployment and monitoring are separated through explicit gates and preserved records.
Quietly formidable by design.
Sarvasvam is being built as a research-led investment organisation where process, evidence and governance compound before scale does.
Institutional enquiries