🚀 Quick Start
Get EDEA running on your exchange in under 5 minutes.
Create your EDEA account
Go to edea-bot.com and sign up with your email. You'll receive a 6-digit verification code by email. Your 3-day free trial starts immediately — no charge until it ends.
Connect your Kraken exchange
Log into Kraken → Security → API → Generate new key. Enable Read and Trade only. Never enable Withdraw. Paste your key + secret into EDEA Settings → Exchange.
Choose a strategy and configure risk
Use the Strategy Finder (✨ button above the strategy grid) to get a personalised recommendation. Set your max trade size (€10–€50 to start), stop loss (1.5–2.0%), and max open positions (3).
Add pairs to your watchlist and start the engine
Go to Settings → Watchlist and add 4–6 pairs (e.g. BTC/EUR, ETH/EUR, SOL/EUR). Hit Start Engine. EDEA scans your watchlist every 2 minutes, 24/7.
[TICK] messages. If you see [SKIP] messages, the algorithm is correctly filtering out weak setups. This is normal — the bot is being selective, not broken.
What to expect
- The bot scans every 2 minutes on all watchlist pairs simultaneously
- A trade only places when the strategy score exceeds the threshold AND the Triple Barrier quality gate passes (historical TP hit rate >a minimum quality bar for that pair)
- Most scans produce no trade — this is correct behaviour, not a bug
- The ratchet automatically locks profits as positions move in your favour
- You can pause the engine at any time from the Overview page — open positions remain protected
🔑 Connect Your Exchange
EDEA currently supports Kraken. More exchanges coming in 2027.
Creating a Kraken API key
Log into Kraken and open API settings
Go to kraken.com → Account → Security → API. Click Generate new key.
Set permissions correctly
Enable Query Funds, Query Open Orders, Create & Modify Orders, Cancel/Close Orders. Never enable Withdraw. EDEA cannot request withdrawal access — but confirm this yourself.
Optional — IP whitelist
Restrict your API key to EDEA's server IP for extra security. Contact support for the current IP range. This is optional but recommended for high-balance accounts.
Paste into EDEA
Go to EDEA → Settings → Exchange. Paste your API key and secret. EDEA stores them encrypted (AES-256) and never exposes them in logs or the UI.
Checking your connection
After saving your API key, the Overview page will show your Kraken account balance. If the balance shows as €0 or you see a connection error, check that your API key has the correct permissions and that it hasn't expired or been revoked on Kraken's side.
⚡ Your First Trade
What happens from engine start to your first position.
The scan cycle
Every 2 minutes, Google Cloud Scheduler triggers a global tick. EDEA evaluates every active pair in your watchlist against your selected strategy. For each pair it computes a score from 0 to the strategy's max score. A trade places only when:
- Score > the threshold (the global entry threshold)
- Triple Barrier quality gate passes: historical take-profit hit rate >a minimum quality bar for this pair
- You have budget remaining (total open positions < max positions)
- No whale pump/dump signal is active for this pair
Reading the Engine Terminal
Open the Engine Terminal tab to see real-time logs. Common messages on your first day:
| Message | What it means |
|---|---|
| [TICK] Scan: 6 pairs | top=BTC/EUR score=2.4 | Normal scan. No trade placed — score is above the entry threshold threshold. |
| [SKIP] BTC/EUR — a score below threshold below threshold the threshold | Score insufficient. Market does not meet entry criteria. Not a bug. |
| [SKIP] ETH/EUR — Triple Barrier blocked: 28% TP rate | Historical success rate too low for this pair. Correctly blocked. |
| [BUY] SOL/EUR @ 145.20 | score=4.1 | Trade placed. All quality filters passed. |
| [TP] SOL/EUR — Take Profit hit @ 148.80 | Position closed at target. Profit realised. |
📋 Strategy Overview
EDEA has 15 trading strategies + the AI Agent. Each runs on the same 2-minute tick engine with the same score threshold and Triple Barrier quality gate.
Strategy Selection Guide
| Situation | Recommended Strategy |
|---|---|
| New user, wants simplicity | EDEA Fusion — adapts automatically to any market |
| Clear uptrend on the chart | Trend Rider or Breakout Scalper |
| Market is sideways, no direction | Grid Intelligence or Reversal Hunter |
| Volatility breakout expected | Volatility Squeeze |
| Wants mathematical sophistication | Hurst Adaptive or Calibrated Parrondo (Premium+) |
| Highest conviction, fewer trades | Apex Consensus (VIP+) |
| Learning how EDEA works | Parrondo Adaptive — simplest to understand |
| Short timeframes (1m–15m) | Breakout Scalper or CSS Flow Rider |
| Full autonomy, no oversight | AI Agent (Enterprise) |
Availability by plan
| Plan | Strategies available |
|---|---|
| Basic (€9.99/mo) | Parrondo Adaptive, Trend Rider |
| Standard (€19.99/mo) | + Reversal Hunter, Breakout Scalper, Grid Intelligence, Volatility Squeeze, EDEA Fusion, CSS Flow Rider, Scorecard Consensus |
| Premium (€39.99/mo) | + Calibrated Parrondo, Ichimoku Cloud, Hurst Adaptive |
| VIP (€79.99/mo) | + Apex Consensus |
| Enterprise (€149.99/mo) | + AI Agent + all features |
| Trial (3 days) | All 15 strategies + AI Agent available during trial |
The flagship EDEA strategy. Based on the mathematical Parrondo Paradox: two strategies that individually lose can produce a winner when alternated intelligently. Game A focuses on momentum (buy when the trend is accelerating). Game B focuses on mean reversion (buy when price is oversold and likely to bounce). EDEA monitors performance and switches games when the active game accumulates net losses.
Game A — Momentum signals
Game B — Mean Reversion signals
Game switching
When the active game's net score drops below a negative threshold, EDEA switches to the other game and resets the counter. In the Engine Terminal this appears as [SWITCH] Game A → Game B. The switching itself is the source of the positive expected value — not either game alone.
A strict trend-following strategy. Only enters when multiple EMAs are fully stacked in the same direction, ADX confirms meaningful trend strength, and momentum is building. Deliberately skips choppy, ranging, and sideways markets. In exchange for this strictness, Trend Rider produces high-conviction entries in genuine trends.
Entry signals
Identifies statistically oversold conditions where selling has been excessive and a bounce is probable. Multiple independent confirmation signals must agree simultaneously. The strictness is intentional — it avoids entering during genuine downtrends by requiring evidence that buyers are actually returning.
Entry signals
Enters the precise moment a consolidation zone breaks upward on elevated volume. Genuine breakouts (new price highs with significantly elevated volume) have strong follow-through. Fake breakouts (new highs on weak volume) fail quickly. Volume confirmation is the critical filter.
Entry signals
A range-accumulation strategy. Identifies when price is in a tight, well-defined consolidation range and enters near the lower boundary. Critical rule: if ADX exceeds 25, the strategy passes regardless of other signals — in trending markets, lower Bollinger Band levels are not support.
Entry signals
Identifies the specific moment when compressed volatility is about to release. The "squeeze" occurs when Bollinger Bands compress inside the Keltner Channel. When the BBs expand back outside the Keltner Channel, a directional move is statistically imminent.
Entry signals
The most comprehensive strategy. Runs all other strategies simultaneously on every tick and combines their scores using regime-dependent weights. In trending markets, Trend Rider and Breakout Scalper scores count most. In ranging markets, Grid Intelligence and Reversal Hunter count most. The regime is detected live via HMM and Hurst exponent.
Regime weights
Rides real institutional order flow using order book imbalance (OBI), Hurst exponent, EMA20 trend direction, and OBV accumulation. Designed to detect when large buyers are quietly accumulating before a move.
Entry signals
Multi-layer confirmation consensus using EMA structure (45%), momentum indicators (35%), ATR volatility filter, and OBV gate. Requires all layers to agree before entering. Lower frequency than individual signal strategies — higher per-trade quality.
Layer weights
An enhanced version of Parrondo Adaptive with three improvements: signal weights are calibrated via walk-forward backtesting on the recent price history of each specific pair; the starting game is selected empirically from backtest results; and VIP/Enterprise plans add a Gemini AI macro context score as a capped signal contribution.
Improvements over base Parrondo
The only EDEA strategy that projects support and resistance 2recent candles into the future via the Senkou Span cloud. Uses it as a multi-layer confirmation system covering trend direction, short-term momentum, historical momentum, and projected support/resistance ahead. Requires minimum 60 candles of history.
Entry signals
Uses the DFA (Detrended Fluctuation Analysis) Hurst exponent to classify the statistical character of the current price series. H above the persistence threshold = persistent/trending → applies momentum signals. H below the mean-reversion threshold = anti-persistent/mean-reverting → applies mean-reversion signals. H in the statistical noise zone = noise → returns score 0 and refuses to trade.
Hurst zones
The highest-conviction strategy. Uses eight completely independent signal systems, each voting separately. A trade only occurs when the majority of all eight systems independently agree. Any single system showing a clearly bearish reading applies a −2.0 soft veto. Apex produces the fewest trades of any strategy — when it fires, it represents the highest available consensus.
Eight independent vote systems
Replaces the traditional scoring engine with Gemini AI that autonomously selects pairs, timing, and position sizing. Unlike all other strategies, AI Agent does not use the the threshold score threshold or Triple Barrier quality gate. The AI receives full market context on every tick and makes its own trading decisions. All reasoning is logged in the Engine Terminal.
AI inputs (every tick)
🛡️ Stop Loss & Take Profit
EDEA places stop loss and take profit orders simultaneously when entering every position.
Stop Loss
The stop loss closes your position if the price drops by your configured percentage from the entry price. It is placed immediately when the position opens. If your stop loss is set to 1.5%, a position entered at €100 will close at €98.50.
- Recommended range: 1.0–2.5% for most crypto pairs
- Too tight (<0.5%): normal volatility closes positions prematurely
- Too wide (>6%): each loss is large; recovery requires a bigger win
- Configure in Settings → Risk → Stop Loss %
Take Profit
A take profit target closes the position at a set profit percentage. Unlike the ratchet (which is dynamic), take profit is a fixed level. When EDEA's ratchet is active, the ratchet often exits before the take profit target is reached on strong moves.
ATR-based dynamic stops
Some strategies use ATR (Average True Range) to set dynamic stop losses that breathe with market volatility. When ATR is high (volatile market), the stop widens slightly to avoid false exits. This is automatic and does not override your configured stop loss percentage — it adjusts within a safe range.
🔒 Ratchet System
EDEA's proprietary progressive trailing stop. The most important feature for protecting profits on winning trades.
How it works
- Position opens with a standard stop loss
- As price rises, the ratchet floor rises with it — locking in gains progressively
- The floor never moves down. Once locked at a level, it stays there
- Once the profit covers trading fees, the floor locks above your break-even point
- When price falls back to the floor, the position exits automatically — profit secured
Three ratchet profiles
| Profile | How it behaves | Best for |
|---|---|---|
| Conservative | Tight floor, locks gains quickly, exits sooner | Volatile markets, shorter holds |
| Balanced | Adaptive floor, balances protection and runway | Most users — default |
| Aggressive | Wide floor, gives position more room to run | Strong trending markets |
Ratchet Universal
Standard+ plans include Ratchet Universal, which extends ratchet protection to all positions in your account — including manual trades you open yourself on Kraken. Any position EDEA detects gets ratchet protection automatically.
💰 Position Sizing
How EDEA determines how much to invest in each trade.
Max trade size
Your Max Trade Size setting determines the maximum amount EDEA will invest in a single position. EDEA will never invest more than this amount in one trade, regardless of signal strength.
- Start with €10–€50 for your first two weeks
- Increase gradually as you validate the bot's performance on your pairs
- Your total risk at any time is: Max Trade Size × Max Open Positions
Max open positions
The maximum number of simultaneous trades. When this limit is reached, EDEA will not open new positions until an existing one closes.
| Plan | Max positions allowed |
|---|---|
| Basic | 3 |
| Standard | 5 |
| Premium | 10 |
| VIP | 20 |
| Enterprise | Unlimited |
Budget guardrail
EDEA will not open a new position if doing so would exceed your plan's maximum budget. This is a hard limit — even if the strategy generates a signal, no trade places if you are at your budget ceiling.
🧠 HMM Regime Gauge
A Hidden Markov Model trained in real-time on exchange candles to classify the current market regime.
What it does
The HMM Gauge analyses 300–500 candles for your selected pair and timeframe, fits a 3-state Hidden Markov Model (Trending / Ranging / Reverting), and tells you:
- The current most likely regime and its confidence probability
- Transition probabilities: how likely is the regime to change next period?
- A statistical price distribution (P10/P25/P75/P90 quantiles) for the current window
- Swing high and low levels from recent candle history
- BOCPD changepoint detection — alerts when the regime has statistically changed
How to use it for trading decisions
- If HMM shows Trending with high confidence → favour Trend Rider or Breakout Scalper
- If HMM shows Ranging → favour Grid Intelligence or Reversal Hunter
- If the BOCPD alert fires (regime change detected) → reassess your strategy choice
- The P25 price level is a statistically fair entry zone in ranging conditions
- The P75 level is a statistically elevated zone — caution for new longs
🧭 Signal Compass (CSS)
Real-time order flow analysis combining four independent signals into a directional score.
The four CSS components
| Component | What it measures |
|---|---|
| OBI — Order Book Imbalance | The ratio of bid volume to ask volume at the top of the order book. Positive OBI = more buyers than sellers right now. |
| TFI — Transaction Flow Imbalance | Whether recent trades were more buy-initiated or sell-initiated. Measures real executed flow, not just orders. |
| Hurst Exponent | Statistical character of price series — persistent (trending) vs anti-persistent (mean-reverting). |
| HMM Regime | Current market regime from the Hidden Markov Model. |
Reading the CSS score
- Score range: typically −100 to +100
- Strong positive (>+50): institutional buying pressure detected
- Strong negative (<−50): selling pressure dominant
- Near zero: balanced market, no strong directional signal
- A "Gate Open" signal means CSS has crossed a threshold — the bot may be more likely to trigger
🔎 Pattern Matching
A k-NN historical similarity engine that finds the 20 most structurally similar moments in price history.
How it works
Pattern Matching takes a window of recent candles and searches all available historical data for the 20 moments most structurally similar to the current setup. It then shows the distribution of what happened next across those 20 historical matches.
What the panel shows
- P25/P75 range: the middle 50% of outcomes from the 20 closest historical matches
- Bullish probability: percentage of matched moments that resulted in price rising
- Bearish probability: percentage of matched moments that resulted in price falling
- Median expected move: the typical magnitude of the move that followed
📡 Correlation Lab
Two tools for analysing relationships between pairs: lead-lag analysis and 24h relative strength.
Correlation & Lead-Lag
Measures how two pairs move together and whether one tends to move before the other. A high correlation (near 1.0) means the pairs move in the same direction most of the time. A positive lead-lag means Pair A tends to move before Pair B.
- Use BTC/EUR or ETH/EUR as the lead pair — they often move before altcoins
- If BTC/EUR is rising and your altcoin hasn't moved yet, the lag signal suggests a follow-up move
- Re-scan before each session — lead-lag relationships shift over time
24h Relative Strength
Normalises two pairs to 100 at the same 24h start point and overlays them on the same chart. The gap between the lines is the divergence — how much one pair has outperformed the other over the past 24 hours.
- Large divergence + high correlation = likely mean-reversion trade — the laggard often catches up
- Large divergence + low correlation = pair-specific news — investigate before trading
- Spread narrowing from peak = convergence already happening — trade may be late
- Always check the HMM regime of the laggard before entering
📦 What are Packs?
A crypto basket tool — create a group of assets, confirm weights, and EDEA executes all orders simultaneously.
Pack types
| Type | How the basket is built | Plan required |
|---|---|---|
| Manual Pack | You select every asset and weight | Standard+ |
| Momentum Pack | Bot selects top momentum assets from your watchlist | Standard+ |
| CSS Flow Pack | Bot selects assets with strongest order flow signal | Standard+ |
| Scorecard Pack | Bot selects based on Scorecard Consensus scores | Standard+ |
| Reversal Pack | Bot selects most oversold assets | Premium+ |
How packs differ from individual trades
- A pack is a single grouped position — P&L is shown for the group as a whole, plus per-asset breakdown
- All orders in a pack fire simultaneously — no sequential entry
- Packs are managed independently from the algo bot — you can run both at the same time
- Close a pack with one click — all assets sell simultaneously
➕ Creating a Pack
Step-by-step guide to building and launching your first pack.
Open the Pack Wizard
Go to Trading → Packs tab → click New Pack. The wizard opens.
Choose a pack type
Select Manual, Momentum, CSS Flow, Scorecard, or Reversal. For your first pack, start with Manual or Momentum.
Set market cap filter (optional)
Filter assets by market cap: All / Large cap (top 15 by volume) / Mid cap (16–35) / Small cap (36+). Smaller caps carry more risk.
Review and confirm
The bot proposes assets and weights. You review every asset and weight before anything is bought. Adjust as needed, then confirm. All orders fire simultaneously.
⌨️ Engine Terminal Messages
Complete reference for all messages that appear in the Engine Terminal tab.
| Message | What it means |
|---|---|
| [TICK] Scan: N pairs | top=X score=Y | Hurst Adaptive | Normal scan tick. Strategy shown. No trade placed (score is above the entry threshold threshold). This is the most common message. |
| [SKIP] PAIR — Triple Barrier blocked: Weak setup — TP hit rate below quality bar | Historical take-profit success rate too low for this pair. Trade blocked regardless of score. Not a bug — the quality gate is working. |
| [SKIP] PAIR — a score below threshold below threshold the threshold | Score insufficient. Market does not meet entry criteria for the selected strategy. |
| [BUY] PAIR @ price | Game A | score=4.1 | Trade placed. Score above threshold, all quality filters passed. |
| [WHALE_PUMP] PAIR — Buy 70% | Delta +1.7% | Large coordinated buy detected. Bot blocked entry to avoid buying into a pump. |
| [ASSESSMENT] Market: TRENDING → Game A confirmed | One-time regime assessment. Starting game confirmed for Parrondo strategy. |
| [SWITCH] Game A → Game B (score dropped below a negative threshold) | Parrondo switching triggered. Game A net loss score hit threshold. Switched to mean reversion. This is by design. |
| [TSL] PAIR — TSL triggered @ price | Trailing stop loss fired. Position exited with profit locked in. |
| [SL] PAIR — Stop Loss @ price | Hard stop loss executed. Position closed to limit loss. |
| [TP] PAIR — Take Profit hit @ price | Take profit target reached. Position closed at target. |
| [tick] Config reload error for user | Firestore config reload failed on this tick. Previous config used. Escalate if persistent — contact support. |
| [session] No token provided for user | Session authentication failed. User likely not logged in or token expired. Log out and back in. |
SKIP messages explained
SKIP messages are the most frequently misunderstood. They do not mean something is broken. They mean the algorithm evaluated the pair and correctly decided not to enter because the quality criteria were not met. Seeing many SKIP messages means your quality gate is working — you are being protected from weak setups.
❓ Frequently Asked Questions
Common questions from EDEA users.