For quants & systematic traders

Describe a strategy. Ivy builds it,
backtests it, and tells you whether
the results hold up.

Invarity is a platform for systematic trading on Indian markets, powered by Ivy — an AI agent that turns a plain-English idea into a readable strategy and a rigorous backtest, one that grades the evidence and flags short history, thin trades, and inflated ratios instead of hiding them.

Request a demo See how it works
Built for Indian markets · NSE equities · rupee-denominated
app.invarity.ai / Backtesting RSI strategy on RELIANCE
Backtest an RSI mean-reversion strategy on RELIANCE — RSI(14), buy below 30, sell above 70, with a 2% stop.
Here's the strategy, in plain language. The exact Python is in Strategies whenever you want it.
RSI Mean Reversion with Stop Loss
Strategy spec · v1.0.0
Open
Summary
Buy when 14-period RSI falls below 30, sell when RSI exceeds 70, with 2% stop-loss protection.
Entry rule
Go long when RSI(14) drops below 30 (oversold).
Exit rules
Exit above RSI 70, or if price falls 2% below entry.
Universe
Single stock (RELIANCE), daily timeframe.
Indicators
RSI(14)
Tunable parameters
rsi_period14oversold30overbought70stop_loss0.02
Backtest complete — here's the analysis, graded, with its reasoning shown.
Verdict
Needs more evidence
Cost survival: Yes Regime fit: Neutral Confidence: Low
Reliability notes
  • History is 0.9y (< 1y) — multi-year windows suppressed and ratios may be inflated.
  • Time in market is low — sparse exposure can inflate risk-adjusted ratios.
How it works

From a sentence to a verdict, in three moves.

You describe the idea in plain English. Ivy handles the code and the data, then hands back an analysis honest enough to act on.

01 — Describe it
Say it like you'd tell a colleague
Entry, exit, universe, sizing, risk — in plain language, in a chat. No SDK to learn and no data to wire up.
02 — Ivy builds & runs it
A readable spec, then a real backtest
Ivy writes a plain-language spec anyone can review — with the exact Python a click away in Strategies — and runs the backtest on historical prices.
03 — Read the verdict
Metrics, plus whether to trust them
Get an institutional metric catalogue and a plain-English verdict with reliability notes. Ask a follow-up or refine the idea in the same chat.
Readable by the whole desk

Every strategy, in plain language first.

Ivy explains each strategy the way a person would — the summary, the exact entry and exit rules, the universe, sizing, costs, and indicators — so anyone can read it, not just whoever wrote it. Need the precise logic? The generated Python sits one click away in Strategies.

A formatted spec built for non-technical readers.
Tunable parameters called out explicitly.
The exact runnable code available in Strategies.
RSI Mean Reversion with Stop Loss
Strategy spec · v1.0.0 · RSI 14-30-70 with 2% stop
Summary
Buy when 14-period RSI falls below 30, sell when RSI exceeds 70, with 2% stop-loss protection.
Entry rule
Enter long when 14-period RSI drops below 30 (oversold condition).
Exit rules
Exit when RSI rises above 70. Exit immediately if price drops 2% below entry (stop loss).
Universe
Single stock (RELIANCE) on daily timeframe.
Sizing & risk
90% of available cash per trade, 2% stop from entry.
Costs & slippage
Market orders at close; no explicit transaction costs.
Indicators
RSI(14)
Tunable parameters
rsi_period14oversold30overbought70stop_loss_pct0.02
from sdk import Strategy

class RSIStrategy(Strategy):
    def __init__(self, data_loader, config):
        self.rsi_period = 14
Full generated code lives in Strategies
Verdict
Reject
Confidence: Low Regime fit: Unknown Cost survival: No Overlap: Unknown
Reliability notes
  • History is 0.90y (< 1y) — multi-year windows suppressed and ratios may be inflated.
  • Only 0 trades — win/loss statistics are low-confidence.
  • Time in market is 0.0% — low exposure can inflate risk-adjusted ratios.
  • Zero-variance returns — Sharpe/Sortino are undefined.
The honest part

A verdict on every backtest — and why.

A good-looking equity curve can be an accident of a short window or a handful of trades. Ivy grades each run with a clear verdict and signals for confidence, regime fit, and cost survival — then lists the reliability notes behind it, so you know exactly when a result is too thin to trust.

Accept / reject with confidence and regime signals.
Reliability notes for short history, thin trades, low exposure.
Undefined metrics flagged, not quietly reported as zero.
What's in every analysis

The full picture behind the numbers.

An institutional metric catalogue
Returns and risk in depth — CAGR, Sharpe, Sortino, Calmar, max drawdown, volatility, VaR & CVaR, ulcer index, skew, kurtosis, profit factor, expectancy, and risk of ruin.
Sharpe1.24
Calmar1.26
Max DD−14.6%
Charts that show the whole run
Equity curve, the underwater drawdown, monthly and yearly return grids, and return distributions — the shape of the strategy, not just its summary line.
Measured against a benchmark
Every run is compared to a volatility-matched benchmark, with beta, alpha, correlation, and information ratio — so you can tell real edge from market beta.
alphabetacorr
Monte Carlo for robustness
Ivy simulates thousands of alternate paths to estimate the range of outcomes — median final return, median max drawdown, and the risk of ruin — so a single lucky sequence doesn't fool you.
median outcomerisk of ruin
Every trade, itemized
Round-trips, fills, and total costs, with win rate, average win and loss, expectancy, hold time, and turnover — the trade-level detail behind the headline metrics.
Win rate54%
Payoff1.3
Dig in, in chat — then export
Open any run back in chat to ask Ivy why a metric moved or refine the idea, and export the full analysis when you're ready to share it.
Open chatExport
Questions

The things quants ask first.

Do I need to know how to code?
No. You describe the strategy in plain English, and Ivy writes and runs it. The primary view is a formatted spec anyone can read; if you do work in code, the exact generated Python is available in the Strategies section.
What can I backtest?
Systematic strategies on Indian equities — for example an RSI mean-reversion rule on a single NSE stock, on a daily timeframe, in rupees. You set the entry, exit, universe, sizing, and risk; Ivy turns it into a runnable strategy.
How do I know whether a backtest is trustworthy?
Every run gets a verdict with signals for confidence, regime fit, and cost survival, plus reliability notes that call out short history, too few trades, low time in market, and undefined risk ratios. When a result is too thin to trust, Ivy says so rather than presenting a misleading number.
What metrics and charts do I get?
A full catalogue: returns (total, CAGR, best/worst periods), risk (volatility, max drawdown, VaR, CVaR, ulcer index, skew, kurtosis, risk of ruin), risk-adjusted ratios, benchmark comparison, and Monte Carlo outcomes — alongside the equity curve, drawdown, and monthly and yearly return charts.
Can I export my results?
Yes. You can export the full analysis to share or keep, and reopen any run in chat to discuss or refine it.

See Ivy on your strategies.

Book a walkthrough and we'll run one of your ideas through Ivy — the spec, the backtest, the verdict, and everything behind it.

Request a demo