Artificial Intelligence Crypto Trading Automation

Not writing for a long time; Although there were several reasons for not writing, it is better not to give excuses, but to give in the guise of originality - it has self-clarity.

Anyway TrickBD has a chain of tutorials on crypto trading and following up should hopefully take you from the basics of crypto trading to the practical execution and earning section – but at the end of CryptoRazz I had something going around in my head like “I won't be trading manually but automatically trading and those trades must be profitable as a single benefit to my crypto portfolio”. Earnings keep piling up”.

Yes…absolutely childish thinking because (1) Crypto market is always uncertain – so guaranteed profit is literally not practical (2) Profitable predictions can only be made by doing proper market analysis and thorough study before taking a trade – there is no alternative.

However, in reverse thinking, if in crypto trading I check the current market condition (legit analysis) with all the crypto trading history (data) in the past and take into account the future condition (future trend) - so that the margin of loss can be reduced to zero even if the probability of profit is minimal - then it is called fairly stable profitable crypto trading. goes

And when all this happens automatically it will be called Automatic Benefited Crypto Trading.

Now the matter is not as simple as it sounds – in the past all the crypto trading market history (data) that can describe each candlestick is possible for a human being to remember and practically utilize as it is almost impossible to verify or compare with the current market condition and taking trades according to the future up trend is also complicated.

In this case, we can create a system rules to verify current market conditions with all crypto trading market history data in the past to execute profitable trades with seemingly zero margin losses – or artificial intelligence (AI) can be used for current market analysis and future trend combination to be as emotionless as possible compared to humans (to avoid unwanted hype trading sessions). (without interruption) can calculate the correct entry point.

On the other hand, it is possible to create a crypto trading automation system that can be active 24/7 from market monitoring – observing trade sessions – opening and closing trades (permission dependent of course).

System Trading Logic

(1) Trend (EMA Crossover) — Crypto Market Indication:

if (efL != null && esL != null) {
    const sp = ((efL - esL) / price) * 100;
    tScore = clamp(sp * 40, -100, 100);
}

Analysis: 12-period and 26-period exponential moving average (EMA) calculations.

• If the 12-EMA is above the 26-EMA, the crypto market is considered to be in an uptrend (score is +).
• If the 12-EMA is below the 26-EMA, then the crypto market is considered to be in a downtrend (score – yes).
• The greater this distance, the higher the score (ranging from -100 to +100).

(2) Momentum (RSI) — To understand how active or inactive the crypto market is

if (rsiL <= 30) rScore = 100 - (rsiL / 30) * 40; else if (rsiL >= 70) rScore = -(100 - ((100 - rsiL) / 30) * 40);
else rScore = ((50 - rsiL) / 20) * 30;

System Logic: RSI is additional BUY/SELL indicative system function

• If the RSI goes below 30, it means the crypto market is oversold (oversold) — the system takes it as a buying opportunity (increases the score).
• If the RSI goes above 70, it means the crypto market is overbought (overbought) — the system takes it as a selling opportunity (lowers the score).
• If it is between 30-70, the RSI moves away from 50 – the score increases or decreases accordingly.

(3) MACD (Momentum Confirmation) — Trading Speed ​​Calculation

if (histL != null) {
    const dir = histL > 0 ? 1 : histL < 0 ? -1 : 0;
    const accel = histP != null ? (histL - histP) : 0;
    mScore = clamp(dir * 60 + clamp(accel * 500, -40, 40), -100, 100);
}

System Logic: By looking at the histogram (bar) of MACD we understand the direction and speed of momentum.

• Histogram direction: If the bar is above the zero line, it means bullish momentum (upward momentum) — 60 points are added if the direction is present. If it is below it is cut.
• Acceleration: How much the last histogram value has increased or decreased from the previous one — indicating whether the acceleration is accelerating or decelerating. This adds/subtracts up to 40 more points.
• If the speed is continuously increasing, the score will spread out and if the speed is decreasing then the score will decrease.

(4) Volatility Price High/Low Situation

if (bU != null && bL != null && bM != null) {
    const width = bU - bL || 1;
    const posn = (price - bM) / (width / 2);
    bScore = clamp(-posn * 70, -100, 100);
}

System Logic: The middle line of Bollinger Bands is a 20 period simple moving average. Where does the current price stand compared to that average?

• If the price is in the upper band (well above the average), it means the price has gone abnormally high — the system assumes a return, so gives a sell score (negative).
• If the price is in the lower band (well below the average), it means that the price has fallen abnormally — there is a possibility of a rebound, hence giving a buy score (positive).

(5) Adaptive Weighting System — How will trading decisions be made?

(A) General weighting

let w = normWeights(weights);
// Default: trend=30%, rsi=25%, macd=25%, boll=20%

System Logic: By default Trend is weighted at 30%, RSI at 25%, MACD at 25% and Bollinger at 20%.

(B) Regime-Adaptive Weighting (ADX-Based) Intelligent Movement

if (opts.regimeAdaptive && adxL != null) {
    const regime = clamp((adxL - 20) / 40, -0.4, 0.4);
    const adj = {
        trend: w.trend * (1 + regime),
        macd: w.macd * (1 + regime),
        rsi: w.rsi * (1 - regime),
        boll: w.boll * (1 - regime)
    };
    // Renormalization
}

System Logic: ADX indicator tells whether the crypto market is trending or ranging.

• If ADX is more than 25 (strong trend), the system automatically increases the weight of the trend and MACD, because the range measures (RSI, Bollinger) are less effective then.
• If ADX is less than 25 (along with moving crypto market), the system increases the weighting of RSI and Bollinger, as looking for a reversal by avoiding the trend is a more profitable way.

(C) Volume Confirmation — Confirmation logic development in supporting volume

if (opts.volumeConfirm && volRL != null) {
    if (volRL >= 1.5) confidence += 8;
    else if (volRL <= 0.5) confidence -= 10;
}

System Logic: Not only price analysis but also volume (amount of transactions) is important.

• If the volume is 1.5 times the average, that means there are a lot of buyers/sellers behind that signal — so confidence increases by 8%.
• If the volume is less than half the average, the signal is weak — so confidence drops by 10%.

(6) Final Score Calculation — Decision Making and Trade Executive Conditions

let score = clamp(
    tScore * w.trend + 
    rScore * w.rsi + 
    mScore * w.macd + 
    bScore * w.boll, 
    -100, 100
);

System Logic: The above four scores are added by multiplying them by their adjusted weights. This results in a final score between -100 and +100.

• Signal Determination:
• Score ≥ +20 → BUY order.
• Score ≤ -20 → SELL order.
• In the middle → HOLD instruction.

Confidence level function

const agreement = same / subs.length; 
const magnitude = Math.min(Math.abs(score) / 100, 1);
let confidence = Math.round((0.55 * agreement + 0.45 * magnitude) * 100);

System Logic:

• Agreement: How many of the four indicators agree? (eg if all indications give the same signal then agreement is 100%).
• Magnitude: How active is the final score? (Score +90 is very strong, +25 is weak).
• Final Confidence = (55% × Agreement) + (45% × Severity). If this confidence is less than the default 65%, the system will not trade even if the score shows BUY or SELL – this is the most effective filtering system.

7. AI Council Integration

(A) Local AI Core Function

System Logic: Local AI has 20+ analyst roles with each role analyzing the crypto market from a different perspective. Like—Technical Analyst (sees RSI, EMA), Macro Trend Analyst (ADX, sees trend direction), Social Sentiment Scanner (volume type analysis), Risk Analyst (loss potential) etc. After all, all conditions combine to make a decision.

(B) External AI Providers

switch (provider) {
    case 'openai':     // GPT-4o, GPT-4o-mini
    case 'deepseek':   // DeepSeek-Chat, DeepSeek-Reasoner
    case 'gemini':     // Gemini 2.0 Flash, Pro
    case 'openrouter': // Claude 3.5, Llama 3.2
}

System Logic: You can also use popular AI like OpenAI, DeepSeek, Gemini or OpenRouter to provide intelligence and clarity in the data analysis, compare and decision making section of the crypto market [in this case you can integrate any one desired model or integrate from multiple' model API – or switch any AI model from just OpenRouter].

(8) Hybrid Decision Making

const blendedConf = Math.round(
    (ext.confidence || 50) * 0.6 + 
    (localResult.confidence || 50) * 0.4
);

System Logic: External AI's opinion is given 60% weight and local council's opinion 40% weight [as the external AI is having the ability to perform current market analysis here] Final decision is taken by combination of both, so that even if one party makes a mistake, the other can correct him and gain the most correct decision.

(9) Risk Management Logic — Position Sizing

const riskUsd = equity * (risk.riskPerTrade / 100);
const sizeUsd = Math.min(riskUsd / (stopLossPct / 100), equity * 0.25, paper.cash);

System Logic:

• You can risk up to 2% (default) of total equity per trade.
• If the stop-loss is 3%, then the total position can be taken by calculating the risk of 2%.
• But without trading the entire amount at once — the maximum position size is kept up to 25% of the equity.

(10) Stop-Loss and Take-Profit

st.paper.openStops[symbol] = {
    sl: price * (1 - stopLossPct / 100),    // 3% default
    tp: price * (1 + takeProfitPct / 100),   // 6% default
};

System logic: 3% stop-loss is set below entry price (to cut loss), and 6% take-profit is set above (to book profit).

(11) Trailing Stop-Loss (Protection increases as profit increases)

if (risk.trailingEnabled) {
    const gainPct = ((price - avgEntry) / avgEntry) * 100;
    if (gainPct >= activatePct) { // Activate at 2% profit
        const newSl = price * (1 - trailDistancePct / 100); // 1.5% distance
        if (newSl > stops.sl) stops.sl = newSl;
    }
}

System logic: If the trade goes to 2% profit, the stop-loss is activated. Then, as the price rises, the stop-loss also moves upwards, maintaining a distance of 1.5%. As a result profit is somewhat secured [however low to gain security of minimum profit].

(12) Daily Loss Limit

const dayLoss = ((equity - dayStartEquity) / dayStartEquity) * 100;
if (dayLoss <= -dailyLossLimit) { stop(); }

System Logic: If there is a loss of 5% (default) from the amount of money at the beginning of the trade, then the system understands that the trading day is not going well and automatically shuts down the entire engine.

(13) Consecutive Loss Streak Management

if (lossLimit > 0 && consecutiveLosses >= lossLimit) {
    pausedUntil = Date.now() + (cooldown * 60000);
}

System Logic: If there are 3 trade losses in a row, then the system assumes that it may not understand the market conditions. It stops trading (goes into cooldown) for 60 minutes (default) Next function recall will recheck system data to ensure improved trade execution in synchronization with trading pattern ML.

(14) Trade Execution Flow

(A) Entry Conditions

if (Date.now() - lastAt < cooldownMin * 60000) return; if (label === 'BUY' && holding && holding.qty > 0) return;
if (label === 'SELL' && (!holding || holding.qty <= 0)) return;

System Logic: Before placing a trade the system checks – whether the cooldown (15 minutes) on this pair is over (to avoid repeated trades). If the BUY signal comes, but you already have that coin, don't buy a second time. If the SELL signal comes, but you don't have that coin, don't sell (short selling is not supported).

(B) Multi-timeframe confirmation

if (s.risk.multiTfEnabled) {
    const hEval = Strategy.evaluate(hCandles, weights, opts);
    if (hEval && hEval.label === 'SELL' && label === 'BUY') return;
    if (hEval && hEval.label === 'BUY' && label === 'SELL') return;
}

System Logic: Even if the 1 hour chart shows BUY, the system checks the 4 hour or 1 day chart. If the larger timeframe says SELL as opposed to BUY, the smaller timeframe signal is ignored (filtered out).

(C) Order Execution (Paper & Live)

// Paper trading
s.paper.cash -= amountUsd;
h.qty += qty;
h.avgEntry = (h.avgEntry * oldQty + price * qty) / newQty;

// Live Trading (Binance API)
await Exchange.placeLiveOrder({ symbol, side: 'BUY', type: 'MARKET', quantity: amountUsd });

System Logic: In paper mode only calculation changes (reduce cash, increase holding). Market orders are actually sent to Binance in live mode.

(15) Backtesting Engine (formerly Crypto Trading History Analysis)

function run(cfg) {
    for (let i = warmup; i < candles.length - 1; i++) {
        const evalRes = Strategy.evaluate(candles.slice(0, i + 1), weights, opts);
        // Position management and trade simulation
    }
    // Returns: Final Equity, Return %, Max Drawdown, Win Rate, Profit Factor
}

System Logic: This applies the strategy to past crypto market data—how much money would have been gained/lost if this system had been used. The warm-up period (60 candles) is omitted, because the indicators do not work properly at the beginning. Shows the results—last money, percentage return, maximum drawdown, win rate, and profit factor.

Summarize features and functions

• 4 Indicators Trend (EMA), Momentum (RSI), Confirmation (MACD), Volatility (Bollinger)
• Adaptive Weighting ADX changes the importance of the indicator accordingly by understanding whether the crypto market is trending or ranging.
• AI Council 20+ Local Analysts (Functional Systems) + External AI Providers OpenAI, DeepSeek, Gemini, OpenRouter—Analysis with any.
• Risk Management Position Sizing, SL/TP, Trailing Stop, Daily Loss Limit, Consecutive Loss Break.
• Verifying multi-timeframe small chart signals with larger charts
• Analysis and utilization of how backtesting strategies have performed historically.
• Practice with paper trading virtual cryptocurrencies
• Live Trading Automatic order placement in real Binance account (real life trade execution after confirming system performance).

Application

Crypto trading automation system application is available in browser based CryptoRazz's CryptoDevi. Also more advanced and automated functionality CryptoDev will also get almost the same function flow – it will have the following features:

• Vault — AES-GCM encrypted storage of Binance API, Telegram tokens, Supabase keys etc. in browser with passphrase.
• Market Data Engine — Live price, 24 hour change, volume and fear-grid index of all USDT pairs from Binance.
• Candle Data — Grabbing 15m, 1h, 4h or 1-day real-time candlestick data of any pair from Binance.
• AI Consensus Engine — 13 different analyst roles (Technical, Macro, Sentiment, Risk) vote together to derive BUY/SELL/HOLD signals and confidence.
• Paper Trading — Risk-free trading with virtual money, tracking portfolio value, P&L, win rate, max drawdown and sharp ratio.
• Equity Curve Chart — Graphs of portfolio value over time of paper trading for accurate comparison.
• Live Exchange (Binance) — Places HMAC-SHA256 signed real orders in the browser (account balance and open order monitoring).
• Backtester — RSI, MACD and Bollinger Bands rulesets run on real Binance data for the past 180 days for returns, win rates and sharp ratios.
• Risk Engine — Position size calculator, exposure limit check, circuit breaker and automated stop-loss/take-profit/trailing rules settings.
• Wallet (MetaMask) — Connect MetaMask-compatible wallets, view native balances, and send native tokens
• On-Chain Security Scanner — Viewing address balances, transaction lists, and contract source code scanning with the EtherScan-family API to assign heuristic risk scores.
• Whole Transaction Filter — Filter out transactions larger than a specified threshold from recent transactions.
• Live Gas Tracker — Shows current Safe, Standard and Fast gas prices for Ethereum.
• DeFi Calculators — Calculates Compound Yield, Permanent Loss, ROI, Mining ROI, DCA Backtest and FIFO Tax.
• Crypto Utility Toolkit — Multi-Hash (MD5/SHA1/SHA256/SHA512), Shannon Entropy, AES-GCM Encrypt/Decrypt, BIP-39 Checksum Validator, Base Converter and Markel Root Builder.
• Alerts & Automation — Setting alerts on price watchlists, sending alerts via browser notifications, Telegram bots and webhooks.
• Cloud Sync (Supabase) — Push/pull portfolio data with your Supabase project to sync across multiple devices.
• Autonomous Agent — Scans the market periodically, looks at AI consensus and executes automated trades in paper/live mode.
• Dashboard — showing live market overview, fear-grid gauge, paper portfolio summary, AI pulse and list of recent trades.
• Navigation & View Routing — Switch to more than 12 views from the sidebar (Dashboard, Signals, Agents, Paper, Live, Backtest, Risk, Wallet, Security, DeFi, Tools, Alerts, Cloud, Settings).
• Charting (Chart.js) — Renders equity curves and price charts (with Chart.js).
• Export/Import Backup — Downloads and restores entire app data as a JSON file.
• Trade Journal Export — Exports all paper trades in CSV format.

Disclaimer

Since it is a complete automation system and directly connected to the API based Binance or connected to the wallet from MetaMask, the logic flow test and virtual trade should be run only after confirmation; However with access to personal financial data it should be open-sourced rather than just kept in CryptoRazz's own system – so that there is clarity and transparency and if you can customize the system rule's production.

Local deployment

You can deploy CryptoDev and CryptoDevi's two program files to your own localhost (always active and must have Node.js) system; Or can deploy on your VPS system.

Both scripts are open source and can be downloaded as static files from GitHub.

Download Link :- https://github.com/HumayunShariarHimu/CryptoTrendingAutomation [Open source – you can customize it according to your trading technique if you want].

Welcome to the Web PsychoTech

Best wishes on your particular crypto trading journey – stay well.