Multi-agent system
Quantropy runs a multi-agent system. Specialist agents parse intent, compile specs, execute backtests and stress-test results.
Product
From a thought to a quant-level backtest
01
Describe your strategy in plain English
No DSL, no Python, no config files. Write it the way you'd say it: legs, deltas, days to expiration, entry filters, profit targets, stops and rolls. Quantropy asks for anything ambiguous instead of silently guessing.
quantropy · strategy console
Sell 30-delta SPY put credit spreads with 5-point wings at 45 DTE, close at 50% profit or 21 DTE. Skip entries when VIX > 30. Testing period 2018-2025.
02
AI converts it into a precise backtest
Your description is compiled into an explicit, reviewable strategy spec: every rule visible, every assumption stated. You approve the interpretation before a single trade is simulated, so you always know exactly what was tested.
quantropy · compiled options spec
Universe
SPY · S&P 500 ETF options
2018–2025
Leg specs
SELL1× put · 0.30 delta
BUY1× put · 5-point lower strike
Entry logic
45 DTE target · enter at market open
VIX < 30 · delta-targeted strikes · one entry daily
Exit rules
PROFIT50% credit
TIME21 DTE
STOP200% credit
Risk & sizing
1 spread per $25k notional
03
Run it against precise options data
Backtests replay the actual option chain: quotes, spreads and greeks as they printed, so fills reflect what you could realistically have traded. End-of-day data hides the costs that decide whether an edge survives.
quantropy · options chain
| Strike | Bid | Ask | IV | Delta |
|---|---|---|---|---|
| 560P | 3.12 | 3.18 | 17.4% | -0.31 |
| 555P | 2.44 | 2.49 | 18.1% | -0.26 |
| 550P | 1.88 | 1.93 | 18.9% | -0.21 |
| 545P | 1.41 | 1.46 | 19.8% | -0.17 |
| 540P | 1.02 | 1.07 | 20.9% | -0.13 |
04
The AI copilot suggests improvements
Quantropy examines the backtest for weak points, unintended exposure, fragile assumptions and missed opportunities. It explains what may be holding performance back, then proposes concrete refinements you can test rather than changing the strategy blindly.
quantropy · backtest diagnostics
- High impact
Exit timing
Winners give back 18% of peak profit before the 21 DTE exit.
SUGGESTED EDITTest a trailing profit rule - Refine
Entry filter
Trades entered with IV rank below 20 produce a negative expectancy.
SUGGESTED EDITAdd IV rank ≥ 20 - Risk
Position sizing
Overlapping positions create 1.8× the intended downside exposure.
SUGGESTED EDITCap concurrent risk
05
Analyze results and iterate
Compare variants side by side, tighten the rules that matter, discard the ones that don't. Every change re-runs against the same, so you can see exactly which adjustments moved the equity curve. Export results once you like them.
quantropy · strategy results
RETURN
+214.6%
CAGR
+18.4%
Sharpe
1.31
Sortino
1.87
DRAWDOWN
-14.8%
Equity curve
StrategySPY buy & hold1Y3Y5YALL20182020202220242025
06
Automated execution
ON THE ROADMAPOnce the strategy holds up across regimes, promote it to automated execution. Entries and exits route to your broker while Quantropy keeps monitoring each fill against live conditions — every trade logged for review.
quantropy · automated execution log
Trade log
| Entry | Exit | Structure | P&L | Return |
|---|---|---|---|---|
| 2024-11-08 | 2024-11-29 | SPY 30Δ put spread · 5pt | +60 | +15.0% |
| 2024-09-20 | 2024-10-11 | SPY 30Δ put spread · 5pt | -340 | -85.0% |
| 2024-07-19 | 2024-08-09 | SPY 30Δ put spread · 5pt | +58 | +14.5% |
Automation live · entries & exits routed to broker
Market data research
Or just ask the market a question
Maybe during strategy development you'd like to confirm an assumption about market behaviour. Just ask a question.
And get an explicit, reviewable definition: event, measurement window, universe, date range --> approved before any number runs. Covers earnings reactions, volatility events, macro prints. Returns occurrences, hit rate, average/median move, and the return distribution. If the sample's too thin to mean anything, Quantropy says so instead of spinning a story.
quantropy · market research
In how many cases when Apple earnings show a P/E increase of more than 5% does the market go up that day?
Question parsed
Event: AAPL earnings where P/E rose > 5% QoQ · Window: same-day SPX return · Universe: 2010–2025 · N = 47 occurrences.
Under the hood
Historical data depth
Years of tick-level chain history with quotes, implied volatility surfaces and greeks. Corporate actions and expiries handled for you.
Purpose-built financial LLMs
Our LLMs are developed in-house and trained to work precisely with financial data: options chains, greeks, tick history and market microstructure.
Claim your early access spot
Drop your email and we'll send your invite as spots open, along with your 30% lifetime discount code if you sign up until the end of 2026.
No spam. Product updates and your invite only.