Vriddhi वृद्धि — self-improving paper-trading agent
PAPER ONLY · no real money
loading report…
Equity curve — adaptive agent vs static ensemble vs buy-&-hold ($10k start)
Trailing-2yr holdout (out-of-time)
Meta-allocator weights over time (Hedge multiplicative weights)
Self-tuned learning rate η (meta-learning)
Cumulative PnL by strategy
Research artifact, not financial advice. Backtests are hypothetical,
assume perfect execution at daily closes, and cannot capture all real-world
frictions. Past (even simulated) performance does not guarantee future
results. Crypto assets are extremely volatile; you can lose everything.
See RISK.md in the repository.