KiteTrader
KiteTrader
— — PAPER Engine
Scanning market…

Index

Breadth

Patterns found

Top movers

SECTOR ROTATION · ranked by strength vs index

Global cue for NIFTY

Morning brief

World markets that move India

India

US-listed ADRs (overnight read for the NSE stock)

Results & corporate calendar

Market headlines

Company news (48 h)

Sector news sentiment

Sources

Free public feeds (Yahoo Finance, ET, Moneycontrol, LiveMint, Business Standard, Google News, NSE/BSE). Any source can change or block without notice — the table shows which are working. Add or fix result dates in data/results_manual.csv (symbol,date,purpose).

Grade = confluence count (market trend, stock trend, relative strength, volume, pattern quality, reward:risk, ADX). Entry/stop/targets come from chart structure (support, resistance, pattern levels) clamped to 1–3×ATR. Tap a card for the chart. Rule-based aids, not investment advice.

How to read the breakout life-cycle

How the AI trades

1 · DataDaily candles for the universe + NIFTY 50 (Adani group excluded)
2 · Features~40 per stock-day: momentum, volatility, trend, RSI/MACD/ADX, volume, 52-wk position, market regime, breadth, cross-sectional ranks
3 · LabelsTriple-barrier: does a trade hit +target before −stop within H days, after costs?
4 · ModelsLogistic regression + gradient-boosted trees + extra-trees, averaged
5 · ValidationPurged walk-forward: train on the past, test on the next quarter, gap so labels never overlap
6 · VerdictAUC, lift vs baseline, quarter-by-quarter stability, 1-trade-a-day simulation
7 · LiveRank today's stocks, explain each pick, hand the best to the risk engine (daily budget, stops)
8 · GuardrailsBlocks trading on "No edge", stale model, data drift or live under-performance; weekly retrain

Train model

Goal finder — what does 90% accuracy actually cost?

Win% (real) = 4,000-path simulation on bootstrapped real index moves with 6 checkpoints a day, IV rising when the market falls, and full charges (STT 0.15% on option sales, exchange, GST, ₹20/order). Win% (IV) = same on implied-vol paths. Max loss = the most the hedged structure can ever lose per lot.

Strategy builder

or add legs from the chain with B / S

Payoff

Build or pick a structure

OPTION POSITIONS · auto-managed every 30 s: target, stop, time exit

OPTION CHAIN

Hedge my portfolio

Accuracy vs profit — read this first

Win rate alone does not make money. A strategy needs win% × avg win > loss% × avg loss after costs. Any rule can reach 90% "accuracy" by taking tiny profits and keeping wide stops — and then one loss wipes out nine wins. This lab shows both numbers for every rule, tested on past data with real costs, and the optimizer judges combinations on data it never saw (out-of-sample).

Run

STRATEGY SCORECARD

TODAY'S HIGH-PROBABILITY SIGNALS

p = this stock's own historical hit rate for that rule (shrunk toward the universe rate when the sample is small)

Sector allocation

HOLDINGS

Score 0–100 combines trend (EMA20/50/200), momentum (RSI, MACD), 52-week-high proximity, volume and ADX. Signals are screening aids, not advice. Tap a row for the chart.

Virtual trading

Fast-forward replay — test your current Algo settings on past months in minutes

Uses the strategies and risk limits currently saved in the Algo tab (daily strategies: Setup trader, 20-day breakout, RSI dip-buy, High-probability).

COMPARE RUNS · change one setting at a time, then compare

Engine

Risk limits

Position size = min(capital/trade ÷ price, risk/trade ÷ (price − stop), max order value ÷ price). Orders are LIMIT at LTP ± buffer, throttled to 5/s.

Strategies

ENGINE POSITIONS

Readiness checklist

Step-by-step: from zero to live

  1. Zerodha account: make sure DDPI is active (Console → Account → DDPI) so the app can sell delivery shares. Enable F&O only if you will trade options.
  2. Kite Connect app: sign up at developers.kite.trade → Create new app → type Connect (₹500/month; the free Personal plan has no historical data, which this app needs). Keep the API key and secret handy.
  3. Server with a fixed IP: AWS Lightsail → Create instance → Region Mumbai → Linux/Ubuntu 24.04 → plan Small 2 GB. Then Networking → Attach static IP, and in the instance's Networking tab add a firewall rule for HTTPS (443).
  4. Install: Lightsail → Account → SSH keys → download the Mumbai default key. With it and kitetrader.zip in Downloads, open PowerShell and run:
    cd $HOME\Downloads; scp -i .\LightsailDefaultKey-ap-south-1.pem .\kitetrader.zip ubuntu@SERVER-IP:~
    Then open the instance's browser terminal (Connect using SSH) and run:
    sudo apt-get install -y unzip && unzip kitetrader.zip && sudo bash kitetrader/deploy/setup_server.sh
    It prints your app URL, dashboard password, server IP and Kite redirect URL.
  5. Connect Kite: run sudo kt kite and paste the API key/secret. In the Kite developer console set the Redirect URL and add the server IP under Profile → IP Whitelist (one change per week).
  6. Open the app on your phone (the https URL), log in, tap Login to Kite. Do this every trading morning before 9:15 — Telegram can remind you (sudo kt telegram).
  7. Paper-trade on real data for at least 2 weeks with a fresh virtual run. Review the Virtual tab; use Replay to tune.
  8. Go live small: Algo → Apply Safe mode preset, set Live capital cap to ₹25,000–50,000 → Save. Tick the confirmations here, run sudo kt live on, then press Switch to LIVE below.
  9. Every day: log in to Kite, glance at the News tab and Telegram, and use KILL if anything looks wrong. sudo kt live off locks real trading instantly.

Phone alerts (Telegram)

Event log

Engine orders