Introducing Quant Mode: Ask Sortefi About the Numbers, Not Just the Symbols
Until today Sortefi answered one kind of question: which symbols match my conditions? From today it answers a second kind too: what are the numbers? Name the symbols, ask in plain English, and Sortefi works the answer out on the history and shows you how.
We call it Quant mode. It sits next to screening in the app's sidebar, runs on the same data across the same five markets (US stocks, Indian NSE stocks, ETFs, forex and crypto perpetual futures), and follows the same rule we have always kept: if something important is missing, Sortefi asks one question instead of guessing.
Two ways to ask
| Screen | Quant | |
|---|---|---|
| Asks | Which symbols match? | What are the numbers? |
| Returns | A list, ready for TradingView | An answer, its tables and the method |
| Price | About $0.20 (₹20) | From $0.10 (₹10) |
Three real answers
These were asked of Quant on 13 Nov 2026. The numbers are copied from the answers, only rounded.
“Compare Apple and Microsoft on 3-year Sharpe ratio and maximum drawdown using daily candles”
AAPL: Sharpe 0.94, max drawdown −33.43% (26 Dec 2024 to 8 Apr 2025, 103 days). MSFT: Sharpe 0.57, max drawdown −34.91% (28 Oct 2025 to 25 Jun 2026, 240 days).
| Symbol | Sharpe | Max drawdown | Days |
|---|---|---|---|
| AAPL | 0.94 | −33.43% | 103 |
| MSFT | 0.57 | −34.91% | 240 |
“Are BANKNIFTY options expensive right now? Compare 30-day implied volatility with 20-day historical volatility over the last 6 months”
On 12 Nov 2026 Bank Nifty's 30-day implied volatility was 15.51% against 20-day historical volatility of 21.71%, 6.19 points below. Over the 6 months, implied was above historical on 19 of 126 sessions.
| Measure | Latest | 6-month average |
|---|---|---|
| Implied volatility (30-day) | 15.51% | 14.38% |
| Historical volatility (20-day) | 21.71% | 16.95% |
| Implied minus historical | −6.19 | −2.57 |
“Backtest BTC daily May 2024–Oct 2026: buy when the 12 EMA crosses above the 26 EMA, sell on the cross below, 0.1% fee and 0.05% slippage per trade. Return, max drawdown and win rate vs buy and hold.”
17 trades, win rate 29.41%. The rule returned +24.47% against +24.50% for buying and holding, with a maximum drawdown of −32.48% against −53.30%.
| Measure | Rule | Buy and hold |
|---|---|---|
| Return | +24.47% | +24.50% |
| Max drawdown | −32.48% | −53.30% |
| Trades | 17 | — |
| Win rate | 29.41% | — |
What you can ask
- Returns and investing: total and yearly returns over any window; a lump sum on a date or a monthly investment, with dividends reinvested or not.
- Risk: volatility, Sharpe and Sortino ratios, maximum drawdown with its dates, every fall of a given size and how long it took to recover.
- Against an index: beta, alpha and correlation, and how much one symbol moves when another does.
- How often, and what followed: how often a condition happened and what came next, streaks, opening gaps and how soon they filled, overnight versus intraday moves.
- Seasonality: returns by year, calendar month, weekday, day of the month or trading session.
- Statistics: skew and kurtosis, stationarity and random-walk tests, autocorrelation, cointegration and regression against a benchmark.
- Options, India and crypto: implied against realised volatility, IV rank, put-call ratio, max pain, and the payoff and Greeks of a strategy you state leg by leg.
- Portfolios: several symbols held together with your weights: combined return, volatility, drawdown and value at risk.
Any indicator or candlestick pattern Sortefi screens with can be used inside a question too: the average RSI on Mondays, or how often a hammer candle was followed by a green day.
Test a trading rule on past data
Describe a rule the way you would write it down: when to buy, when to sell, and if you like a stop, a target, long or short, and your costs. Sortefi runs it over the window you give and lists the trades (the last 100 when there are more), with the total return, yearly growth, drawdown, win rate and Sharpe ratio next to simply buying and holding the same symbol.
A few things are fixed so the result can be trusted:
- Signals are read on finished candles, and each trade fills at the next candle's open unless you say otherwise.
- Every backtest is checked for look-ahead: the rule is re-run on the history cut off at earlier dates, and a rule whose past signals would change is not used.
- If you don't state costs, you see the result with no costs and with 0.1% a side. For forex the default is no costs and a 1-pip spread.
- Ask for a settings search (“try periods 7 to 30”) and you get every setting, not only the best one, with a warning that a best setting found on the past may not hold, and a separate check period if you give one.
A backtest describes what a rule did. It isn't a forecast, and a rule that worked before can stop working.
What every answer shows
The headline numbers in a sentence or two, the tables behind them with a row for every symbol and its sample size, the method (candle size, the real first and last dates, every standard applied, such as a 0% risk-free rate or prices without dividends) and any data warnings. If a symbol has less history than you asked for, the answer says so.
What it won't do
Quant reports the past and present in the data. It doesn't forecast prices, recommend trades, pick the “best” strategy or give entry and exit levels. Those requests are declined, and declined requests aren't charged.
Price
From $0.10 (₹10 in India) an answer, from the same wallet as your screens. A question that reads a lot of history, compares many symbols or tests a rule costs more, and the exact charge is shown with every answer. Free prompts for new accounts cover quant answers too. No subscription.
The full list of what Quant can answer is in the Quant docs. If you ask it something it gets wrong, or can't answer yet, I'd like to hear about it: get in touch.
— Siddharth, founder of Sortefi