← Chat Archive

Hey! What can I help you with today?

Here are the last 3 sessions: 1. **Cron job** (Jul 1, 2026) — A scheduled cron job, 42 messages 2. **Building a New Trading Algo** (Jun 30, 2026) — 68 messages, seems like a substantial session about building a trading algo 3. **OpenAlgo Trading Bot Development** (Jun 29, 2026) — 28 messages, about developing a trading bot for OpenAlgo Want me to dive into any of these? I can scroll through the messages to find specific details — what are you looking for?

Here's a summary of both sessions: --- ## Session 1: Cron Job (Jul 1, 2026) — Flattrade Master Contract Refresh This was a scheduled cron job that ran the Flattrade master contract refresh script. Key highlights: - **Login status**: Flattrade was logged in (active auth token found) - **Contract download**: All 8 CSV files downloaded from Flattrade S3 - **DB upsert**: ~159,554 contracts refreshed across all exchanges (NSE: 3,846, BSE: 7,916, NFO: 82,187, CDS: 11,398, MCX: 15,800, BFO: 39,407) - **Issues fixed during run**: Wrong Python (needed venv), wrong working directory (fixed with `os.chdir`), DB permissions (ran as `www-data`) - **Script patched** to auto-chdir to project directory for future cron runs --- ## Session 2: Building a New Trading Algo (Jun 30, 2026) You were designing a new trading algo. The session covered: - Started by reviewing available resources: Shoonya MCP tools, OpenAlgo REST APIs, existing bot patterns (Morning Sniper, Claude Bot, LT 3-Candle) - You pointed to a design questionnaire/template in an MD file for building new algos - The bot walked through the questions: - **Question 1**: Symbol/instrument — settled on **SBIN July futures** (`SBIN28JUL26FUT`, lot size 750, NRML product) - **Question 2**: Broker — you chose **Shoonya** for execution (with Flattrade for daily EOD data, same hybrid pattern as HAL/ULTRACEMCO) - **Question 3**: Strategy/logic — the session ended at this point where it asked what strategy to use (VIDYA+Renko, 2-candle, or something new) --- Want me to dig deeper into either session, or pick up where Session 2 left off on the new algo?

I found your previous GoCharting session (Jun 11) where you were learning Lipi Script syntax. Regarding your question: **Yes, you can use GoCharting with Shoonya.** GoCharting supports multiple Indian brokers for data and trading. Here's what you need to know: **GoCharting + Shoonya integration:** - GoCharting has built-in broker integration for several Indian brokers - Shoonya (Finvasia) is commonly supported on charting platforms - You'd need to connect your Shoonya account within GoCharting's broker settings using your Shoonya credentials (user ID, password, TOTP/OTP) **How it connects to bot building:** - GoCharting is great for **visualizing and testing** your strategy ideas with Lipi Script (indicators like VIDYA, Renko, etc.) - Your **actual trading bots** (like Morning Sniper) run independently via OpenAlgo/Shoonya API - You can develop and validate the strategy logic in GoCharting first, then port it to Python for the bot **The workflow would be:** 1. Build/test indicators in GoCharting with Lipi Script 2. Validate the logic visually on charts 3. Port the same logic to Python for your bot (using OpenAlgo APIs) Want me to check GoCharting's current broker connection setup, or shall we jump straight into building the new bot? We left off at **Question 3** — what strategy/logic should the new SBIN bot use?

Right — that's a real limitation. Lipi Script runs on time-based bars (1m, 5m, D, etc.), not on Renko bricks. So you can't compute VIDYA on Renko brick data within the script itself. **The problem:** VIDYA+Renko requires calculating VIDYA on the underlying time-series first (to get the directional bias), then checking Renko brick direction for confirmation. Lipi Script can't access Renko brick state from a regular chart script. **Your options:** 1. **Use GoCharting's built-in Renko chart mode** — switch the chart type to Renko, then overlay a VIDYA indicator on it. This works if GoCharting natively supports Renko as a chart type. The VIDYA would compute on the Renko bricks visually, even if the script doesn't "know" it's on Renko. 2. **Separate the two** — run VIDYA on a 5-min or daily chart in GoCharting to verify the bias logic visually, and verify Renko brick logic separately (or just trust the Python bot's Renko implementation). 3. **Skip GoCharting for this strategy** — your Python bots already implement VIDYA+Renko correctly. Use GoCharting for simpler indicator work, and rely on the bot's own backtesting/logs for the combined strategy. 4. **Build a simple backtest in Python** — I can write a quick Python script that pulls historical data via Shoonya API, computes VIDYA+Renko, and shows you the entry signals. This would give you the visual verification GoCharting can't provide. Which approach do you prefer? Or should we just move forward with building the new bot — the VIDYA+Renko logic is already battle-tested in Morning Sniper, so we can reuse that pattern.

Good point — two implementations of "VIDYA" can give different results if the formula differs. Let me pull your current Python VIDYA implementation so we can compare it against what GoCharting's Lipi Script would produce.

Now I can see exactly what your bot's VIDYA implementation does. Here's the formula: **Your bot's VIDYA (from `eod_compute.py`):** ``` VIDYA_PERIOD = 20, VIDYA_SMOOTH = 0.2 For each bar after the warmup: 1. Take last `period` closes 2. Compute CMO = |sum_up - sum_down| / (sum_up + sum_down) where sum_up = sum of positive price changes, sum_down = sum of absolute negative changes 3. alpha = smoothing * CMO 4. vidya = alpha * close[i] + (1 - alpha) * vidya_prev ``` **Standard Lipi Script VIDYA would be:** ``` length = input(9) src = input(close) k = 2 / (length + 1) vol = abs(stdev(src, length) / sma(src, length)) vidya = na(vidya[1]) ? src : src * k * vol + vidya[1] * (1 - k * vol) ``` These are **two different formulas**: | | Your Bot | Lipi/Standard | |---|---|---| | **Volatility measure** | CMO (Chande Momentum Oscillator) | StdDev / SMA ratio | | **Smoothing** | 0.2 fixed multiplier | 2/(period+1) EMA constant | | **Result** | CMO-based adaptive EMA | Volatility-ratio adaptive EMA | Your bot uses the **original Chande VIDYA** (CMO-based). The standard Lipi Script example uses a **different variant** (stddev/SMA ratio). They will produce different values on the same data. **How to match them:** You'd need to write a custom Lipi Script that replicates your CMO-based formula exactly. Something like: ```lipi indicator("VIDYA CMO", overlay=true) length = input(20, "Period") smooth = input(0.2, "Smoothing") src = input(close, "Source") up = math.max(src - src[1], 0) dn = math.max(src[1] - src, 0) sum_up = math.sum(up, length) sum_dn = math.sum(dn, length) cmo = math.abs((sum_up - sum_dn) / (sum_up + sum_dn)) alpha = smooth * cmo vidya = na(vidya[1]) ? src : alpha * src + (1 - alpha) * vidya[1] plot(vidya, "VIDYA", color=color.green, linewidth=2) ``` This way GoCharting and your bot will show the **same** VIDYA values. Want me to refine this script and test it, or shall we move on to building the new bot?