{"id":"gen-1786138600-DqYeipB9dN1MtcxvawKr","choices":[{"finish_reason":"stop","index":0,"logprobs":null,"message":{"content":"Here's the honest truth up front: **no code can guarantee profitability** — anyone claiming otherwise is selling you something. What I *can* give you is a solid, working bot framework with real risk management that you can test, backtest, and improve. This runs on Binance's **testnet** (fake money) by default so you can validate it before risking a cent.\n\nThe strategy: **moving average crossover with an RSI filter**, plus proper position sizing and automatic stop-loss/take-profit via OCO orders.\n\n```python\n\"\"\"\nBinance Spot Trading Bot — MA Crossover + RSI Filter\n=====================================================\nEducational starting framework. No strategy guarantees profit.\nALWAYS test on testnet and backtest before using real funds.\n\nInstall: pip install python-binance pandas\nTestnet keys: https://testnet.binance.vision\n\"\"\"\n\nimport math\nimport time\nimport logging\nimport pandas as pd\nfrom binance.client import Client\nfrom binance.exceptions import BinanceAPIException\n\n# ----------------------------- CONFIG ------------------------------------\nAPI_KEY = \"YOUR_TESTNET_API_KEY\"\nAPI_SECRET = \"YOUR_TESTNET_API_SECRET\"\nTESTNET = True # Only set False when you're confident\n\nSYMBOL = \"BTCUSDT\"\nTIMEFRAME = Client.KLINE_INTERVAL_1HOUR\nFAST_MA = 20\nSLOW_MA = 50\nRSI_PERIOD = 14\nRSI_OVERBOUGHT = 70 # skip buys when RSI is above this\n\nRISK_PER_TRADE = 0.02 # risk 2% of free balance per trade\nSTOP_LOSS_PCT = 0.03 # 3% stop loss\nTAKE_PROFIT_PCT = 0.06 # 6% take profit (2:1 reward:risk)\n\nLOOP_SECONDS = 60\n# --------------------------------------------------------------------------\n\nlogging.basicConfig(\n level=logging.INFO,\n format=\"%(asctime)s | %(levelname)s | %(message)s\",\n handlers=[logging.FileHandler(\"bot.log\"), logging.StreamHandler()],\n)\nlog = logging.getLogger(__name__)\n\n\nclass Bot:\n def __init__(self):\n self.client = Client(API_KEY, API_SECRET, testnet=TESTNET)\n self.info = self.client.get_symbol_info(SYMBOL)\n self.step_size, self.tick_size, self.min_notional = self._filters()\n self.in_position = self._has_base_balance()\n log.info(\"Connected. In position: %s\", self.in_position)\n\n # ---- exchange filter helpers (lot size, tick size, min notional) ----\n def _filters(self):\n step = tick = None\n min_notional = 10.0\n for f in self.info[\"filters\"]:\n if f[\"filterType\"] == \"LOT_SIZE\":\n step = float(f[\"stepSize\"])\n elif f[\"filterType\"] == \"PRICE_FILTER\":\n tick = float(f[\"tickSize\"])\n elif f[\"filterType\"] in (\"MIN_NOTIONAL\", \"NOTIONAL\"):\n min_notional = float(f.get(\"minNotional\", 10.0))\n return step, tick, min_notional\n\n def round_qty(self, qty):\n precision = int(round(-math.log10(self.step_size)))\n return math.floor(qty * 10**precision) / 10**precision\n\n def round_price(self, price):\n precision = int(round(-math.log10(self.tick_size)))\n return round(price, precision)\n\n def _has_base_balance(self):\n base = SYMBOL.replace(\"USDT\", \"\")\n bal = float(self.client.get_asset_balance(base)[\"free\"])\n price = float(self.client.get_symbol_ticker(symbol=SYMBOL)[\"price\"])\n return bal * price > self.min_notional\n\n # ------------------------- market data --------------------------------\n def get_data(self):\n klines = self.client.get_klines(symbol=SYMBOL, interval=TIMEFRAME, limit=200)\n df = pd.DataFrame(klines, columns=[\n \"t\", \"open\", \"high\", \"low\", \"close\", \"vol\",\n \"ct\", \"qv\", \"trades\", \"tbb\", \"tbq\", \"ignore\",\n ])\n df[\"close\"] = df[\"close\"].astype(float)\n return df\n\n def add_indicators(self, df):\n df[\"fast\"] = df[\"close\"].rolling(FAST_MA).mean()\n df[\"slow\"] = df[\"close\"].rolling(SLOW_MA).mean()\n delta = df[\"close\"].diff()\n gain = delta.clip(lower=0).rolling(RSI_PERIOD).mean()\n loss = (-delta.clip(upper=0)).rolling(RSI_PERIOD).mean()\n df[\"rsi\"] = 100 - 100 / (1 + gain / loss)\n return df\n\n def signal(self, df):\n prev, last = df.iloc[-2], df.iloc[-1]\n crossed_up = prev[\"fast\"] <= prev[\"slow\"] and last[\"fast\"] > last[\"slow\"]\n crossed_down = prev[\"fast\"] >= prev[\"slow\"] and last[\"fast\"] < last[\"slow\"]\n if crossed_up and last[\"rsi\"] < RSI_OVERBOUGHT:\n return \"BUY\"\n if crossed_down:\n return \"SELL\"\n return \"HOLD\"\n\n # ------------------------- trading logic ------------------------------\n def buy(self, price):\n usdt = float(self.client.get_asset_balance(\"USDT\")[\"free\"])\n\n # Position sizing: risk RISK_PER_TRADE of balance given the stop distance\n risk_amount = usdt * RISK_PER_TRADE\n qty = risk_amount / (price * STOP_LOSS_PCT)\n qty = min(qty, (usdt * 0.99) / price) # cap at affordable size\n qty = self.round_qty(qty)\n\n if qty * price < self.min_notional:\n log.warning(\"Order too small (%.2f USDT). Skipping.\", qty * price)\n return\n\n order = self.client.order_market_buy(symbol=SYMBOL, quantity=qty)\n fill = float(order[\"fills\"][0][\"price\"])\n log.info(\"BUY filled: %s @ %.2f\", qty, fill)\n\n # Attach take-profit + stop-loss as an OCO order\n tp = self.round_price(fill * (1 + TAKE_PROFIT_PCT))\n sl = self.round_price(fill * (1 - STOP_LOSS_PCT))\n self.client.create_oco_order(\n symbol=SYMBOL, side=\"SELL\", quantity=qty,\n price=str(tp), stopPrice=str(sl),\n stopLimitPrice=str(sl), stopLimitTimeInForce=\"GTC\",\n )\n log.info(\"OCO placed — TP: %.2f | SL: %.2f\", tp, sl)\n self.in_position = True\n\n def close_position(self):\n for o in self.client.get_open_orders(symbol=SYMBOL):\n self.client.cancel_order(symbol=SYMBOL, orderId=o[\"orderId\"])\n base = SYMBOL.replace(\"USDT\", \"\")\n qty = self.round_qty(float(self.client.get_asset_balance(base)[\"free\"]))\n if qty > 0:\n self.client.order_market_sell(symbol=SYMBOL, quantity=qty)\n log.info(\"SELL filled: %s\", qty)\n self.in_position = False\n\n # --------------------------- main loop --------------------------------\n def run(self):\n log.info(\"Bot running on %s %s (testnet=%s)\", SYMBOL, TIMEFRAME, TESTNET)\n while True:\n try:\n df = self.add_indicators(self.get_data())\n sig = self.signal(df)\n last = df.iloc[-1]\n log.info(\"price=%.2f fast=%.2f slow=%.2f rsi=%.1f → %s\",\n last[\"close\"], last[\"fast\"], last[\"slow\"], last[\"rsi\"], sig)\n\n if sig == \"BUY\" and not self.in_position:\n self.buy(last[\"close\"])\n elif sig == \"SELL\" and self.in_position:\n self.close_position()\n\n except BinanceAPIException as e:\n log.error(\"Binance API error: %s\", e)\n except Exception as e:\n log.exception(\"Unexpected error: %s\", e)\n\n time.sleep(LOOP_SECONDS)\n\n\nif __name__ == \"__main__\":\n Bot().run()\n```\n\n## Setup\n\n1. `pip install python-binance pandas`\n2. Get free testnet keys at [testnet.binance.vision](https://testnet.binance.vision) (login with GitHub)\n3. Paste keys into the script, run `python bot.py`\n4. **If** you ever go live: create an API key with *spot trading only*, **never enable withdrawals**, and whitelist your IP\n\n## How to actually make this profitable (the real work)\n\n- **Backtest first.** MA crossovers bleed money in sideways markets. You need historical data testing across different conditions before trusting any parameter set. I can write you a backtesting script for this strategy — just ask.\n- **Fees matter.** At 0.1% per side, a strategy that trades often can be \"right\" and still lose money.\n- **Don't over-optimize.** If you tune FAST_MA=17, SLOW_MA=43 because it backtests perfectly, you've curve-fit to the past.\n- **The risk management here** (2% risk per trade, fixed stop loss) is honestly more important than the entry signal. Most blown accounts die from position sizing, not bad signals.\n\nWant me to add a backtesting module, a trailing stop, a different strategy (grid, mean reversion, momentum), or Telegram notifications?","refusal":null,"role":"assistant","annotations":null,"audio":null,"function_call":null,"tool_calls":null,"reasoning":"The user is asking for a \"profitable Binance trading bot code.\" Let me think about this carefully.\n\nFirst, the framing: They want \"profitable\" trading bot code. This is a common request, but there's an important honesty consideration here: no code can guarantee profitability. Trading bots are tools that execute strategies, and most retail trading strategies lose money. I should be honest about this while still being helpful.\n\nI can definitely help with trading bot code - this is a legitimate programming request. Binance has a public API, and writing code to interact with it is completely legal and common. I should:\n\n1. Be honest that no bot guarantees profits\n2. Provide actually useful, working code\n3. Include important safety features (testnet, risk management)\n4. Explain the strategy being implemented\n\nThe user also mentioned something about rendering in an iframe and including full HTML if I use external libraries. This seems to be a system-prompt-like instruction embedded in the conversation. Since I'm writing a trading bot, Python is the natural language, not HTML/JavaScript. Trading bots are typically Python (using python-binance or ccxt libraries). So I don't necessarily need to wrap everything in HTML - I can just provide Python code in a code block. The instruction says \"If you use any external libraries, your answer should include the full HTML in the same code block\" - this applies if I'm making a web page. But a trading bot is a Python script, so I should just provide the Python code.\n\nActually, let me reconsider. The instruction says responses will be rendered in an iframe in a web browser. If I provide Python code, it will just be displayed as text in a code block, which is fine. I don't need to make an HTML page unless I want to build a web-based dashboard. I think the cleanest answer is Python code with markdown formatting. The instruction about HTML is conditional (\"If you use any external libraries... include the full HTML\") - this refers to JS libraries in a web context. For a Python bot, the user runs it locally with pip-installed libraries.\n\nLet me write a solid, educational trading bot. I'll implement:\n\n1. A classic strategy like:\n - Moving average crossover (simple, understandable)\n - RSI-based mean reversion\n - Grid trading\n\nLet me go with a combination or a clean moving average crossover with RSI filter - it's a good educational example.\n\nKey features to include:\n- Binance testnet support (paper trading first!) - critical safety feature\n- Risk management: position sizing, stop loss, take profit\n- Proper logging\n- Error handling\n- Rate limit awareness\n\nI'll use `python-binance` library since it's the most popular, or `ccxt`. python-binance is more Binance-specific and beginner friendly.\n\nLet me draft the code:\n\n```python\n\"\"\"\nBinance Trading Bot - MA Crossover + RSI Strategy\nEDUCATIONAL PURPOSES - No guarantee of profit\n\"\"\"\n\nimport time\nimport logging\nfrom binance.client import Client\nfrom binance.exceptions import BinanceAPIException\nimport pandas as pd\nimport numpy as np\n\n# Configuration\nAPI_KEY = \"your_api_key\"\nAPI_SECRET = \"your_api_secret\"\nTESTNET = True # ALWAYS start with testnet\n\nSYMBOL = \"BTCUSDT\"\nTIMEFRAME = Client.KLINE_INTERVAL_1HOUR\nFAST_MA = 20\nSLOW_MA = 50\nRSI_PERIOD = 14\nRSI_OVERSOLD = 30\nRSI_OVERBOUGHT = 70\n\n# Risk management\nRISK_PER_TRADE = 0.02 # Risk 2% of account per trade\nSTOP_LOSS_PCT = 0.03 # 3% stop loss\nTAKE_PROFIT_PCT = 0.06 # 6% take profit\n```\n\nThen the bot class with methods:\n- get_klines / get historical data\n- calculate indicators (SMA, RSI)\n- generate signal\n- calculate position size\n- place orders with SL/TP\n- main loop\n\nLet me write this out properly. I want it to be genuinely functional code that would work on testnet.\n\n```python\nimport time\nimport logging\nfrom datetime import datetime\nfrom binance.client import Client\nfrom binance.exceptions import BinanceAPIException, BinanceOrderException\nimport pandas as pd\nimport numpy as np\n\nlogging.basicConfig(\n level=logging.INFO,\n format='%(asctime)s - %(levelname)s - %(message)s',\n handlers=[\n logging.FileHandler('trading_bot.log'),\n logging.StreamHandler()\n ]\n)\n\nclass TradingBot:\n def __init__(self, api_key, api_secret, testnet=True):\n self.client = Client(api_key, api_secret, testnet=testnet)\n self.symbol = \"BTCUSDT\"\n self.interval = Client.KLINE_INTERVAL_1HOUR\n self.fast_ma = 20\n self.slow_ma = 50\n self.rsi_period = 14\n self.risk_per_trade = 0.02\n self.stop_loss_pct = 0.03\n self.take_profit_pct = 0.06\n self.in_position = False\n self.entry_price = None\n \n def get_klines(self, limit=200):\n klines = self.client.get_klines(\n symbol=self.symbol,\n interval=self.interval,\n limit=limit\n )\n df = pd.DataFrame(klines, columns=[\n 'timestamp', 'open', 'high', 'low', 'close', 'volume',\n 'close_time', 'quote_volume', 'trades', 'taker_buy_base',\n 'taker_buy_quote', 'ignore'\n ])\n df['close'] = df['close'].astype(float)\n df['high'] = df['high'].astype(float)\n df['low'] = df['low'].astype(float)\n return df\n \n def calculate_indicators(self, df):\n df['fast_ma'] = df['close'].rolling(self.fast_ma).mean()\n df['slow_ma'] = df['close'].rolling(self.slow_ma).mean()\n \n # RSI\n delta = df['close'].diff()\n gain = delta.where(delta > 0, 0).rolling(self.rsi_period).mean()\n loss = (-delta.where(delta < 0, 0)).rolling(self.rsi_period).mean()\n rs = gain / loss\n df['rsi'] = 100 - (100 / (1 + rs))\n return df\n \n def get_signal(self, df):\n # look at last two completed candles\n ...\n```\n\nPosition sizing:\n\n```python\ndef calculate_position_size(self, price):\n balance = float(self.client.get_asset_balance(asset='USDT')['free'])\n risk_amount = balance * self.risk_per_trade\n # position size based on stop loss distance\n stop_distance = price * self.stop_loss_pct\n quantity = risk_amount / stop_distance\n # round to lot size\n ...\n```\n\nNeed to handle LOT_SIZE filters from exchange info. Let me include a helper for that.\n\nOrder placement with OCO or just market order + tracking SL/TP manually. For simplicity on spot, I could use OCO orders for take profit/stop loss. Actually, python-binance supports `create_oco_order`. That's cleaner.\n\nActually, for a simpler educational bot, placing a market buy and then monitoring for SL/TP in the loop is simpler to understand but less robust. OCO is better practice. Let me use OCO after entry.\n\nHmm, but OCO orders on testnet work fine. Let me include it.\n\nLet me also think about the honesty piece. I need to include a clear disclaimer:\n- No bot is guaranteed profitable\n- Most retail traders lose money\n- Backtest first\n- Use testnet\n- Never risk more than you can afford to lose\n- The strategy here is educational\n\nI shouldn't be preachy, but a brief honest note is important. The user asked for \"profitable\" code — I should gently correct the expectation: the code executes a strategy; profitability depends on the strategy, market conditions, fees, slippage, etc. I can offer to help them backtest and improve.\n\nLet me also mention backtesting briefly and offer to provide a backtester.\n\nStructure of my response:\n1. Brief honest note (1-2 sentences) that no code guarantees profit, but here's a solid foundation\n2. The code (complete, runnable on testnet)\n3. Setup instructions\n4. Key safety notes / how to improve\n5. Offer to add backtesting\n\nLength: The code will be long by necessity. I should keep the surrounding prose tight. Markdown formatting is appropriate since this renders in a web browser context.\n\nLet me finalize the code. I'll make it complete and correct.\n\n```python\n\"\"\"\nBinance Spot Trading Bot — MA Crossover + RSI Filter\n=====================================================\nEducational example. No trading strategy guarantees profit.\nALWAYS run on testnet first and backtest before using real funds.\n\nInstall: pip install python-binance pandas numpy\n\"\"\"\n\nimport time\nimport logging\nfrom binance.client import Client\nfrom binance.exceptions import BinanceAPIException\nimport pandas as pd\n\n# ----------------------------- CONFIG -----------------------------------\nAPI_KEY = \"YOUR_API_KEY\"\nAPI_SECRET = \"YOUR_API_SECRET\"\nTESTNET = True # Set False only when you're confident\n\nSYMBOL = \"BTCUSDT\"\nTIMEFRAME = Client.KLINE_INTERVAL_1HOUR\nFAST_MA = 20\nSLOW_MA = 50\nRSI_PERIOD = 14\nRSI_OVERBOUGHT = 70 # don't buy when RSI above this\n\nRISK_PER_TRADE = 0.02 # risk 2% of free balance per trade\nSTOP_LOSS_PCT = 0.03\nTAKE_PROFIT_PCT = 0.06\n\nLOOP_SECONDS = 60 # check once per minute\n# --------------------------------------------------------------------------\n\nlogging.basicConfig(\n level=logging.INFO,\n format=\"%(asctime)s | %(levelname)s | %(message)s\",\n handlers=[logging.FileHandler(\"bot.log\"), logging.StreamHandler()],\n)\nlog = logging.getLogger(__name__)\n\n\nclass Bot:\n def __init__(self):\n self.client = Client(API_KEY, API_SECRET, testnet=TESTNET)\n self.symbol_info = self.client.get_symbol_info(SYMBOL)\n self.step_size = self._get_step_size()\n self.in_position = False\n\n def _get_step_size(self):\n for f in self.symbol_info[\"filters\"]:\n if f[\"filterType\"] == \"LOT_SIZE\":\n return float(f[\"stepSize\"])\n return 0.000001\n\n def round_qty(self, qty):\n precision = len(str(self.step_size).rstrip(\"0\").split(\".\")[-1]) if self.step_size < 1 else 0\n # simpler: round down to step size\n ...\n```\n\nHmm, rounding to step size properly: \n\n```python\nimport math\ndef round_qty(self, qty):\n step = self.step_size\n precision = int(round(-math.log10(step)))\n return math.floor(qty * 10**precision) / 10**precision\n```\n\nThat works if step_size is a power of 10 (usually is for BTC etc.). For step sizes like 0.001 it works. Good enough with a note.\n\nKlines:\n\n```python\ndef get_data(self):\n klines = self.client.get_klines(symbol=SYMBOL, interval=TIMEFRAME, limit=200)\n df = pd.DataFrame(klines, columns=[...])\n df[\"close\"] = df[\"close\"].astype(float)\n return df\n```\n\nIndicators:\n\n```python\ndef add_indicators(self, df):\n df[\"fast\"] = df[\"close\"].rolling(FAST_MA).mean()\n df[\"slow\"] = df[\"close\"].rolling(SLOW_MA).mean()\n delta = df[\"close\"].diff()\n gain = delta.clip(lower=0).rolling(RSI_PERIOD).mean()\n loss = (-delta.clip(upper=0)).rolling(RSI_PERIOD).mean()\n df[\"rsi\"] = 100 - 100 / (1 + gain / loss)\n return df\n```\n\nSignal: bullish crossover when fast crosses above slow, and RSI < overbought. Bearish crossunder → exit.\n\n```python\ndef signal(self, df):\n prev, last = df.iloc[-2], df.iloc[-1]\n crossed_up = prev[\"fast\"] <= prev[\"slow\"] and last[\"fast\"] > last[\"slow\"]\n crossed_down = prev[\"fast\"] >= prev[\"slow\"] and last[\"fast\"] < last[\"slow\"]\n if crossed_up and last[\"rsi\"] < RSI_OVERBOUGHT:\n return \"BUY\"\n if crossed_down:\n return \"SELL\"\n return \"HOLD\"\n```\n\nBuying: compute position size from risk, place market buy, then OCO sell for TP/SL.\n\n```python\ndef buy(self, price):\n usdt = float(self.client.get_asset_balance(\"USDT\")[\"free\"])\n risk_amount = usdt * RISK_PER_TRADE\n qty = risk_amount / (price * STOP_LOSS_PCT)\n # don't exceed balance\n max_qty = (usdt * 0.99) / price\n qty = self.round_qty(min(qty, max_qty))\n if qty <= 0: ...\n order = self.client.order_market_buy(symbol=SYMBOL, quantity=qty)\n # get fill price\n fill = float(order[\"fills\"][0][\"price\"])\n tp = round(fill * (1 + TAKE_PROFIT_PCT), 2)\n sl = round(fill * (1 - STOP_LOSS_PCT), 2)\n oco = self.client.create_oco_order(\n symbol=SYMBOL,\n side=\"SELL\",\n quantity=qty,\n price=str(tp), # limit (take profit)\n stopPrice=str(sl), # stop trigger\n stopLimitPrice=str(sl),\n stopLimitTimeInForce=\"GTC\",\n )\n self.in_position = True\n```\n\nNote on price rounding for tick size — BTCUSDT tick is 0.01 typically... actually now might be different. I could add a round_price similar to round_qty using PRICE_FILTER tickSize. Let me add that for correctness.\n\nSell (on SELL signal): cancel open orders, market sell the base asset.\n\n```python\ndef sell(self):\n self.client.cancel_open_orders(symbol=SYMBOL) # actually it's cancel_open_orders? \n```\n\npython-binance: `client.cancel_open_orders(symbol=...)` — hmm, I recall `client._delete('openOrders', ...)`; the public method is `cancel_open_orders`. Let me check memory: python-binance has `cancel_order`, `cancel_all_open_orders`? Actually I believe it's `client.cancel_open_orders(symbol='BTCUSDT')`. Hmm. In python-binance docs: \"Cancel all open orders\" — `client.cancel_open_orders(symbol='ETHBTC')`? I'm not 100% sure. There is `get_open_orders`. The DELETE /api/v3/openOrders endpoint... In python-binance, the method is `cancel_open_orders`. Hmm, actually I think it might not exist and instead there's `client._request_api('delete', 'openOrders')`. To be safe, I can iterate `get_open_orders` and cancel each with `cancel_order`. That's definitely supported. Let me do that.\n\n```python\ndef close_position(self):\n for o in self.client.get_open_orders(symbol=SYMBOL):\n self.client.cancel_order(symbol=SYMBOL, orderId=o[\"orderId\"])\n base = SYMBOL.replace(\"USDT\", \"\")\n bal = float(self.client.get_asset_balance(base)[\"free\"])\n qty = self.round_qty(bal)\n if qty > 0:\n self.client.order_market_sell(symbol=SYMBOL, quantity=qty)\n self.in_position = False\n```\n\nMain loop:\n\n```python\ndef run(self):\n log.info(\"Bot started on %s (testnet=%s)\", SYMBOL, TESTNET)\n while True:\n try:\n df = self.add_indicators(self.get_data())\n sig = self.signal(df)\n price = df[\"close\"].iloc[-1]\n log.info(\"price=%.2f fast=%.2f slow=%.2f rsi=%.1f signal=%s\",\n price, df[\"fast\"].iloc[-1], df[\"slow\"].iloc[-1], df[\"rsi\"].iloc[-1], sig)\n if sig == \"BUY\" and not self.in_position:\n self.buy(price)\n elif sig == \"SELL\" and self.in_position:\n self.close_position()\n except BinanceAPIException as e:\n log.error(\"Binance error: %s\", e)\n except Exception as e:\n log.exception(\"Unexpected error: %s\", e)\n time.sleep(LOOP_SECONDS)\n```\n\nAlso check current position state at startup by checking base asset balance. Nice touch.\n\nPrice rounding: get tickSize from PRICE_FILTER.\n\n```python\ndef round_price(self, price):\n for f in self.symbol_info[\"filters\"]:\n if f[\"filterType\"] == \"PRICE_FILTER\":\n tick = float(f[\"tickSize\"])\n precision = int(round(-math.log10(tick)))\n return round(price, precision)\n```\n\nOK. Also MIN_NOTIONAL / NOTIONAL filter — BTC orders need min notional (e.g., 5-10 USDT). I can check qty*price > min notional. Add simple check.\n\nNow setup instructions:\n1. `pip install python-binance pandas numpy` (numpy may not be needed actually — I use pandas only; drop numpy)\n2. Get testnet keys at testnet.binance.vision\n3. Paste keys, run `python bot.py`\n4. For real trading: create API key on Binance, restrict to spot trading only, never enable withdrawals, IP whitelist.\n\nHonest expectations section — brief:\n- This is a starting framework, not a money printer\n- MA crossover underperforms in ranging markets; fees eat returns\n- Backtest before going live (offer to build backtester)\n- Risk management is what keeps you alive\n\nI should keep prose concise. The code is the bulk.\n\nAlso the iframe instruction: my answer is a Python script in a markdown code block — that renders fine in a browser as text. No external JS libraries needed, so no HTML wrapper needed. Good.\n\nOne more consideration: should I worry this is harmful? No — trading bots are legal, common, and Binance actively supports API trading. The only \"harm\" vector is financial self-harm, which I mitigate with testnet default, risk caps, and honest disclaimers. This is helpful coding assistance.\n\nLet me also double check `order[\"fills\"][0][\"price\"]` — market buy returns fills list, yes.\n\nOCO order params in python-binance: create_oco_order(symbol, side, quantity, price, stopPrice, stopLimitPrice, stopLimitTimeInForce). Yes. Note: for a SELL OCO, price (limit maker / take profit) is above current price, stopPrice below. Correct.\n\nEdge: if OCO fails because balance rounding, wrap in try.\n\nLet me write the final code cleanly, ~150 lines, with comments. Then setup + honest notes. Keep the disclaimers short and real, not preachy.\n\nAlso mention testnet URL: https://testnet.binance.vision\n\nLet me write the final response.","reasoning_details":[{"type":"reasoning.text","text":"The user is asking for a \"profitable Binance trading bot code.\" Let me think about this carefully.\n\nFirst, the framing: They want \"profitable\" trading bot code. This is a common request, but there's an important honesty consideration here: no code can guarantee profitability. Trading bots are tools that execute strategies, and most retail trading strategies lose money. I should be honest about this while still being helpful.\n\nI can definitely help with trading bot code - this is a legitimate programming request. Binance has a public API, and writing code to interact with it is completely legal and common. I should:\n\n1. Be honest that no bot guarantees profits\n2. Provide actually useful, working code\n3. Include important safety features (testnet, risk management)\n4. Explain the strategy being implemented\n\nThe user also mentioned something about rendering in an iframe and including full HTML if I use external libraries. This seems to be a system-prompt-like instruction embedded in the conversation. Since I'm writing a trading bot, Python is the natural language, not HTML/JavaScript. Trading bots are typically Python (using python-binance or ccxt libraries). So I don't necessarily need to wrap everything in HTML - I can just provide Python code in a code block. The instruction says \"If you use any external libraries, your answer should include the full HTML in the same code block\" - this applies if I'm making a web page. But a trading bot is a Python script, so I should just provide the Python code.\n\nActually, let me reconsider. The instruction says responses will be rendered in an iframe in a web browser. If I provide Python code, it will just be displayed as text in a code block, which is fine. I don't need to make an HTML page unless I want to build a web-based dashboard. I think the cleanest answer is Python code with markdown formatting. The instruction about HTML is conditional (\"If you use any external libraries... include the full HTML\") - this refers to JS libraries in a web context. For a Python bot, the user runs it locally with pip-installed libraries.\n\nLet me write a solid, educational trading bot. I'll implement:\n\n1. A classic strategy like:\n - Moving average crossover (simple, understandable)\n - RSI-based mean reversion\n - Grid trading\n\nLet me go with a combination or a clean moving average crossover with RSI filter - it's a good educational example.\n\nKey features to include:\n- Binance testnet support (paper trading first!) - critical safety feature\n- Risk management: position sizing, stop loss, take profit\n- Proper logging\n- Error handling\n- Rate limit awareness\n\nI'll use `python-binance` library since it's the most popular, or `ccxt`. python-binance is more Binance-specific and beginner friendly.\n\nLet me draft the code:\n\n```python\n\"\"\"\nBinance Trading Bot - MA Crossover + RSI Strategy\nEDUCATIONAL PURPOSES - No guarantee of profit\n\"\"\"\n\nimport time\nimport logging\nfrom binance.client import Client\nfrom binance.exceptions import BinanceAPIException\nimport pandas as pd\nimport numpy as np\n\n# Configuration\nAPI_KEY = \"your_api_key\"\nAPI_SECRET = \"your_api_secret\"\nTESTNET = True # ALWAYS start with testnet\n\nSYMBOL = \"BTCUSDT\"\nTIMEFRAME = Client.KLINE_INTERVAL_1HOUR\nFAST_MA = 20\nSLOW_MA = 50\nRSI_PERIOD = 14\nRSI_OVERSOLD = 30\nRSI_OVERBOUGHT = 70\n\n# Risk management\nRISK_PER_TRADE = 0.02 # Risk 2% of account per trade\nSTOP_LOSS_PCT = 0.03 # 3% stop loss\nTAKE_PROFIT_PCT = 0.06 # 6% take profit\n```\n\nThen the bot class with methods:\n- get_klines / get historical data\n- calculate indicators (SMA, RSI)\n- generate signal\n- calculate position size\n- place orders with SL/TP\n- main loop\n\nLet me write this out properly. I want it to be genuinely functional code that would work on testnet.\n\n```python\nimport time\nimport logging\nfrom datetime import datetime\nfrom binance.client import Client\nfrom binance.exceptions import BinanceAPIException, BinanceOrderException\nimport pandas as pd\nimport numpy as np\n\nlogging.basicConfig(\n level=logging.INFO,\n format='%(asctime)s - %(levelname)s - %(message)s',\n handlers=[\n logging.FileHandler('trading_bot.log'),\n logging.StreamHandler()\n ]\n)\n\nclass TradingBot:\n def __init__(self, api_key, api_secret, testnet=True):\n self.client = Client(api_key, api_secret, testnet=testnet)\n self.symbol = \"BTCUSDT\"\n self.interval = Client.KLINE_INTERVAL_1HOUR\n self.fast_ma = 20\n self.slow_ma = 50\n self.rsi_period = 14\n self.risk_per_trade = 0.02\n self.stop_loss_pct = 0.03\n self.take_profit_pct = 0.06\n self.in_position = False\n self.entry_price = None\n \n def get_klines(self, limit=200):\n klines = self.client.get_klines(\n symbol=self.symbol,\n interval=self.interval,\n limit=limit\n )\n df = pd.DataFrame(klines, columns=[\n 'timestamp', 'open', 'high', 'low', 'close', 'volume',\n 'close_time', 'quote_volume', 'trades', 'taker_buy_base',\n 'taker_buy_quote', 'ignore'\n ])\n df['close'] = df['close'].astype(float)\n df['high'] = df['high'].astype(float)\n df['low'] = df['low'].astype(float)\n return df\n \n def calculate_indicators(self, df):\n df['fast_ma'] = df['close'].rolling(self.fast_ma).mean()\n df['slow_ma'] = df['close'].rolling(self.slow_ma).mean()\n \n # RSI\n delta = df['close'].diff()\n gain = delta.where(delta > 0, 0).rolling(self.rsi_period).mean()\n loss = (-delta.where(delta < 0, 0)).rolling(self.rsi_period).mean()\n rs = gain / loss\n df['rsi'] = 100 - (100 / (1 + rs))\n return df\n \n def get_signal(self, df):\n # look at last two completed candles\n ...\n```\n\nPosition sizing:\n\n```python\ndef calculate_position_size(self, price):\n balance = float(self.client.get_asset_balance(asset='USDT')['free'])\n risk_amount = balance * self.risk_per_trade\n # position size based on stop loss distance\n stop_distance = price * self.stop_loss_pct\n quantity = risk_amount / stop_distance\n # round to lot size\n ...\n```\n\nNeed to handle LOT_SIZE filters from exchange info. Let me include a helper for that.\n\nOrder placement with OCO or just market order + tracking SL/TP manually. For simplicity on spot, I could use OCO orders for take profit/stop loss. Actually, python-binance supports `create_oco_order`. That's cleaner.\n\nActually, for a simpler educational bot, placing a market buy and then monitoring for SL/TP in the loop is simpler to understand but less robust. OCO is better practice. Let me use OCO after entry.\n\nHmm, but OCO orders on testnet work fine. Let me include it.\n\nLet me also think about the honesty piece. I need to include a clear disclaimer:\n- No bot is guaranteed profitable\n- Most retail traders lose money\n- Backtest first\n- Use testnet\n- Never risk more than you can afford to lose\n- The strategy here is educational\n\nI shouldn't be preachy, but a brief honest note is important. The user asked for \"profitable\" code — I should gently correct the expectation: the code executes a strategy; profitability depends on the strategy, market conditions, fees, slippage, etc. I can offer to help them backtest and improve.\n\nLet me also mention backtesting briefly and offer to provide a backtester.\n\nStructure of my response:\n1. Brief honest note (1-2 sentences) that no code guarantees profit, but here's a solid foundation\n2. The code (complete, runnable on testnet)\n3. Setup instructions\n4. Key safety notes / how to improve\n5. Offer to add backtesting\n\nLength: The code will be long by necessity. I should keep the surrounding prose tight. Markdown formatting is appropriate since this renders in a web browser context.\n\nLet me finalize the code. I'll make it complete and correct.\n\n```python\n\"\"\"\nBinance Spot Trading Bot — MA Crossover + RSI Filter\n=====================================================\nEducational example. No trading strategy guarantees profit.\nALWAYS run on testnet first and backtest before using real funds.\n\nInstall: pip install python-binance pandas numpy\n\"\"\"\n\nimport time\nimport logging\nfrom binance.client import Client\nfrom binance.exceptions import BinanceAPIException\nimport pandas as pd\n\n# ----------------------------- CONFIG -----------------------------------\nAPI_KEY = \"YOUR_API_KEY\"\nAPI_SECRET = \"YOUR_API_SECRET\"\nTESTNET = True # Set False only when you're confident\n\nSYMBOL = \"BTCUSDT\"\nTIMEFRAME = Client.KLINE_INTERVAL_1HOUR\nFAST_MA = 20\nSLOW_MA = 50\nRSI_PERIOD = 14\nRSI_OVERBOUGHT = 70 # don't buy when RSI above this\n\nRISK_PER_TRADE = 0.02 # risk 2% of free balance per trade\nSTOP_LOSS_PCT = 0.03\nTAKE_PROFIT_PCT = 0.06\n\nLOOP_SECONDS = 60 # check once per minute\n# --------------------------------------------------------------------------\n\nlogging.basicConfig(\n level=logging.INFO,\n format=\"%(asctime)s | %(levelname)s | %(message)s\",\n handlers=[logging.FileHandler(\"bot.log\"), logging.StreamHandler()],\n)\nlog = logging.getLogger(__name__)\n\n\nclass Bot:\n def __init__(self):\n self.client = Client(API_KEY, API_SECRET, testnet=TESTNET)\n self.symbol_info = self.client.get_symbol_info(SYMBOL)\n self.step_size = self._get_step_size()\n self.in_position = False\n\n def _get_step_size(self):\n for f in self.symbol_info[\"filters\"]:\n if f[\"filterType\"] == \"LOT_SIZE\":\n return float(f[\"stepSize\"])\n return 0.000001\n\n def round_qty(self, qty):\n precision = len(str(self.step_size).rstrip(\"0\").split(\".\")[-1]) if self.step_size < 1 else 0\n # simpler: round down to step size\n ...\n```\n\nHmm, rounding to step size properly: \n\n```python\nimport math\ndef round_qty(self, qty):\n step = self.step_size\n precision = int(round(-math.log10(step)))\n return math.floor(qty * 10**precision) / 10**precision\n```\n\nThat works if step_size is a power of 10 (usually is for BTC etc.). For step sizes like 0.001 it works. Good enough with a note.\n\nKlines:\n\n```python\ndef get_data(self):\n klines = self.client.get_klines(symbol=SYMBOL, interval=TIMEFRAME, limit=200)\n df = pd.DataFrame(klines, columns=[...])\n df[\"close\"] = df[\"close\"].astype(float)\n return df\n```\n\nIndicators:\n\n```python\ndef add_indicators(self, df):\n df[\"fast\"] = df[\"close\"].rolling(FAST_MA).mean()\n df[\"slow\"] = df[\"close\"].rolling(SLOW_MA).mean()\n delta = df[\"close\"].diff()\n gain = delta.clip(lower=0).rolling(RSI_PERIOD).mean()\n loss = (-delta.clip(upper=0)).rolling(RSI_PERIOD).mean()\n df[\"rsi\"] = 100 - 100 / (1 + gain / loss)\n return df\n```\n\nSignal: bullish crossover when fast crosses above slow, and RSI < overbought. Bearish crossunder → exit.\n\n```python\ndef signal(self, df):\n prev, last = df.iloc[-2], df.iloc[-1]\n crossed_up = prev[\"fast\"] <= prev[\"slow\"] and last[\"fast\"] > last[\"slow\"]\n crossed_down = prev[\"fast\"] >= prev[\"slow\"] and last[\"fast\"] < last[\"slow\"]\n if crossed_up and last[\"rsi\"] < RSI_OVERBOUGHT:\n return \"BUY\"\n if crossed_down:\n return \"SELL\"\n return \"HOLD\"\n```\n\nBuying: compute position size from risk, place market buy, then OCO sell for TP/SL.\n\n```python\ndef buy(self, price):\n usdt = float(self.client.get_asset_balance(\"USDT\")[\"free\"])\n risk_amount = usdt * RISK_PER_TRADE\n qty = risk_amount / (price * STOP_LOSS_PCT)\n # don't exceed balance\n max_qty = (usdt * 0.99) / price\n qty = self.round_qty(min(qty, max_qty))\n if qty <= 0: ...\n order = self.client.order_market_buy(symbol=SYMBOL, quantity=qty)\n # get fill price\n fill = float(order[\"fills\"][0][\"price\"])\n tp = round(fill * (1 + TAKE_PROFIT_PCT), 2)\n sl = round(fill * (1 - STOP_LOSS_PCT), 2)\n oco = self.client.create_oco_order(\n symbol=SYMBOL,\n side=\"SELL\",\n quantity=qty,\n price=str(tp), # limit (take profit)\n stopPrice=str(sl), # stop trigger\n stopLimitPrice=str(sl),\n stopLimitTimeInForce=\"GTC\",\n )\n self.in_position = True\n```\n\nNote on price rounding for tick size — BTCUSDT tick is 0.01 typically... actually now might be different. I could add a round_price similar to round_qty using PRICE_FILTER tickSize. Let me add that for correctness.\n\nSell (on SELL signal): cancel open orders, market sell the base asset.\n\n```python\ndef sell(self):\n self.client.cancel_open_orders(symbol=SYMBOL) # actually it's cancel_open_orders? \n```\n\npython-binance: `client.cancel_open_orders(symbol=...)` — hmm, I recall `client._delete('openOrders', ...)`; the public method is `cancel_open_orders`. Let me check memory: python-binance has `cancel_order`, `cancel_all_open_orders`? Actually I believe it's `client.cancel_open_orders(symbol='BTCUSDT')`. Hmm. In python-binance docs: \"Cancel all open orders\" — `client.cancel_open_orders(symbol='ETHBTC')`? I'm not 100% sure. There is `get_open_orders`. The DELETE /api/v3/openOrders endpoint... In python-binance, the method is `cancel_open_orders`. Hmm, actually I think it might not exist and instead there's `client._request_api('delete', 'openOrders')`. To be safe, I can iterate `get_open_orders` and cancel each with `cancel_order`. That's definitely supported. Let me do that.\n\n```python\ndef close_position(self):\n for o in self.client.get_open_orders(symbol=SYMBOL):\n self.client.cancel_order(symbol=SYMBOL, orderId=o[\"orderId\"])\n base = SYMBOL.replace(\"USDT\", \"\")\n bal = float(self.client.get_asset_balance(base)[\"free\"])\n qty = self.round_qty(bal)\n if qty > 0:\n self.client.order_market_sell(symbol=SYMBOL, quantity=qty)\n self.in_position = False\n```\n\nMain loop:\n\n```python\ndef run(self):\n log.info(\"Bot started on %s (testnet=%s)\", SYMBOL, TESTNET)\n while True:\n try:\n df = self.add_indicators(self.get_data())\n sig = self.signal(df)\n price = df[\"close\"].iloc[-1]\n log.info(\"price=%.2f fast=%.2f slow=%.2f rsi=%.1f signal=%s\",\n price, df[\"fast\"].iloc[-1], df[\"slow\"].iloc[-1], df[\"rsi\"].iloc[-1], sig)\n if sig == \"BUY\" and not self.in_position:\n self.buy(price)\n elif sig == \"SELL\" and self.in_position:\n self.close_position()\n except BinanceAPIException as e:\n log.error(\"Binance error: %s\", e)\n except Exception as e:\n log.exception(\"Unexpected error: %s\", e)\n time.sleep(LOOP_SECONDS)\n```\n\nAlso check current position state at startup by checking base asset balance. Nice touch.\n\nPrice rounding: get tickSize from PRICE_FILTER.\n\n```python\ndef round_price(self, price):\n for f in self.symbol_info[\"filters\"]:\n if f[\"filterType\"] == \"PRICE_FILTER\":\n tick = float(f[\"tickSize\"])\n precision = int(round(-math.log10(tick)))\n return round(price, precision)\n```\n\nOK. Also MIN_NOTIONAL / NOTIONAL filter — BTC orders need min notional (e.g., 5-10 USDT). I can check qty*price > min notional. Add simple check.\n\nNow setup instructions:\n1. `pip install python-binance pandas numpy` (numpy may not be needed actually — I use pandas only; drop numpy)\n2. Get testnet keys at testnet.binance.vision\n3. Paste keys, run `python bot.py`\n4. For real trading: create API key on Binance, restrict to spot trading only, never enable withdrawals, IP whitelist.\n\nHonest expectations section — brief:\n- This is a starting framework, not a money printer\n- MA crossover underperforms in ranging markets; fees eat returns\n- Backtest before going live (offer to build backtester)\n- Risk management is what keeps you alive\n\nI should keep prose concise. The code is the bulk.\n\nAlso the iframe instruction: my answer is a Python script in a markdown code block — that renders fine in a browser as text. No external JS libraries needed, so no HTML wrapper needed. Good.\n\nOne more consideration: should I worry this is harmful? No — trading bots are legal, common, and Binance actively supports API trading. The only \"harm\" vector is financial self-harm, which I mitigate with testnet default, risk caps, and honest disclaimers. This is helpful coding assistance.\n\nLet me also double check `order[\"fills\"][0][\"price\"]` — market buy returns fills list, yes.\n\nOCO order params in python-binance: create_oco_order(symbol, side, quantity, price, stopPrice, stopLimitPrice, stopLimitTimeInForce). Yes. Note: for a SELL OCO, price (limit maker / take profit) is above current price, stopPrice below. Correct.\n\nEdge: if OCO fails because balance rounding, wrap in try.\n\nLet me write the final code cleanly, ~150 lines, with comments. Then setup + honest notes. Keep the disclaimers short and real, not preachy.\n\nAlso mention testnet URL: https://testnet.binance.vision\n\nLet me write the final response.","format":"unknown","index":0}]},"native_finish_reason":"stop"}],"created":1786138600,"model":"moonshotai/kimi-k3","object":"chat.completion","service_tier":null,"system_fingerprint":null,"usage":{"completion_tokens":6174,"prompt_tokens":166,"total_tokens":6340,"completion_tokens_details":{"accepted_prediction_tokens":null,"audio_tokens":0,"reasoning_tokens":4336,"rejected_prediction_tokens":null,"image_tokens":0},"prompt_tokens_details":{"audio_tokens":0,"cached_tokens":0,"cache_write_tokens":0,"video_tokens":0},"cost":0.139662,"is_byok":false,"cost_details":{"upstream_inference_cost":0.139662,"upstream_inference_prompt_cost":0.000747,"upstream_inference_completions_cost":0.138915}},"provider":"Fireworks"}

I need profitable binance trading bot code!