Let's be real: Gambling is mathematically designed to make you lose. The house edge in Aviator is ~3%. Over time, you WILL lose money. No bot, no strategy, no signal group can change that.
So why does this exist? Because if you're going to play anyway, you should at least play with discipline, math, and a stop-loss — not with emotions, gut feelings, and "one more round" at 3 AM. This bot replaces your worst enemy (yourself at the table) with cold, calculated decisions.
Think of it like this: A helmet doesn't prevent accidents. But it keeps your skull intact when one happens. This bot is your helmet.
Adaptive, probability-based strategy bot for the Aviator game (by Spribe). Uses quantitative trading principles — Kelly Criterion sizing, expected value filtering, volatility guards, and streak detection — to make mathematically honest betting decisions.
This is NOT prediction. It is adaptive risk management based on live session data.
- You will NOT get rich using this bot
- The house always has a 3% edge — no strategy eliminates that
- This bot maximizes your chance of hitting a short-term target and then stopping
- If you're gambling money you can't afford to lose, close this page and call a helpline
Gambling Helpline (India): 1800-599-0019 | International: https://www.gamblersanonymous.org
- Open Chrome →
chrome://extensions/ - Enable Developer mode (top-right toggle)
- Click Load unpacked → select the
aviator-strategy-botfolder - Navigate to any site running the Aviator game
- Click the extension icon to show the panel
The bot cannot predict crash points. But it CAN:
- Track a rolling window of recent crash points
- Calculate real-time statistics from that data
- Dynamically adjust bet size and cashout target
- Only bet when expected value is favorable
- Cut exposure during high-variance periods
Maintains a rolling window of the last 50 crash points. After every round, it calculates:
- Mean — average crash value
- Standard Deviation — measure of volatility
- % above 1.5x — how often rounds exceed 1.5x
- % above 2x — how often rounds exceed 2x
- % above 3x — how often rounds exceed 3x
- Recent low streak — consecutive crashes below 1.5x (danger signal)
The bot waits for at least 10 data points before making any bet decisions. During this collection phase it only observes.
Used by professional gamblers and quant funds to size positions based on edge and probability.
Kelly % = (p × b − q) / b
Where:
p = probability of winning (from history)
q = 1 − p (probability of losing)
b = net odds (cashout multiplier − 1)
The bot uses Half-Kelly (standard quant practice — reduces variance at the cost of slightly lower returns). Bet size is capped at 5% of bankroll maximum. If Kelly returns zero or negative, the round is skipped entirely — negative Kelly means the bet has no edge.
The bot does NOT use a fixed target. It picks the optimal cashout each round based on what the data supports:
| Condition | Target | Logic |
|---|---|---|
| 55%+ of recent rounds went above 2x | 2.0x | Session is running hot |
| 55%+ went above 1.5x | 1.5x | Normal conditions |
| 45%+ went above 1.5x | 1.3x | Session is cold — conservative |
| Below 45% | Skip | No edge exists |
When crash points are highly variable (high standard deviation), the bot reduces exposure:
| StdDev | Volatility Level | Action |
|---|---|---|
| < 1.5 | Low | Normal bet size (100%) |
| 1.5–3.0 | Medium | Reduce bet by 40% |
| 3.0–5.0 | High | Reduce bet by 70% |
| > 5.0 | Extreme | Skip round entirely |
The bot adjusts behavior based on consecutive outcomes:
| Condition | Response |
|---|---|
| 4+ consecutive losses | Reduce bet to 50% (never martingale) |
| 6+ consecutive losses | Skip round entirely |
| 3+ recent crashes below 1.5x | Skip round |
| 4+ consecutive wins | Hold steady — do NOT increase bets |
Critical rule: The bot NEVER uses Martingale (doubling after loss). Ever.
Before every round, combines all modules into a single decision:
- Calculate rolling stats from history
- Determine dynamic cashout target
- Check if target has positive EV:
EV = (winProb × payout) − lossProb - If EV ≤ 0 → skip
- Calculate Kelly bet size for that target
- Apply volatility multiplier
- Apply streak adjustment
- Final sanity checks (minimum bet, bankroll cap)
If the bot skips 3 consecutive rounds, it forces a minimum 1% bet at 1.5x to keep collecting data.
| Rule | Threshold | Action |
|---|---|---|
| Drawdown from peak | 25% drop from session high | Bot stops |
| Take-Profit | 40% gain from starting bankroll | Bot stops |
| Max Rounds | Configurable (default 100) | Bot stops |
| Bankrupt | Bankroll reaches ₹0 | Bot stops |
Note: Stop-loss is measured from session peak (highest bankroll reached), not just the starting amount. This is a trailing stop.
The bot implements 9 distinct betting strategies. Each has a fundamentally different approach to bet sizing, target selection, and risk management.
Uses Kelly Criterion to size bets based on detected edge. Dynamically picks cashout target (1.1x–2.0x) from rolling 50-round history. Skips rounds when no positive expected value is found. Dual-panel split: 70% safe + 30% hunter.
Framework: Dubins-Savage Bold Play + Browne's Goal-Reaching.
Dynamically adjusts aggression based on where you stand between your stop-loss floor and profit-target ceiling. The closer you are to ruin, the bolder the bets — the closer you are to your goal, the smaller and safer.
Aggression Zones:
| Zone | Bankroll Position | Behavior |
|---|---|---|
| TARGET | > 95% toward profit goal | Lock-in mode: tiny bets, 1.2x cashout |
| COMFORT | 70–95% toward goal | Conservative: small bets, 1.5x target |
| NEUTRAL | 30–70% toward goal | Standard: Kelly-based sizing |
| DANGER | 10–30% toward goal | Aggressive: bigger bets, higher targets |
| CRITICAL | < 10% from stop-loss | Desperation: bold play, max payout targets |
Key math:
- Gambler's Ruin probability computed in real-time:
P(reach target) = (1 - (q/p)^current) / (1 - (q/p)^target) - Bet sizes and targets derived from distance-to-goal, NOT from crash history
- Completely stateless — all decisions derived from current bankroll position relative to boundaries
Best for: Sessions with a clear profit target and stop-loss, where you want the bot to "fight" its way to the goal with calibrated aggression.
A multi-phase "observe then fire" strategy. Fully autonomous — no manual target or window configuration needed.
Lifecycle:
INITIAL_OBSERVE (50 rounds) → Analysis → READY → [trigger check] → FIRE
↓ (after each bet)
OBSERVE (50 rounds) → Re-analysis → READY → [trigger check] → FIRE
↓ (repeat until window exhausted or hit)
DONE → restart from INITIAL_OBSERVE
Phase 1 — Observation (50 rounds, configurable): The bot watches crash data without betting. It analyzes 6 multiplier tiers (10x, 15x, 20x, 30x, 50x, 100x) and scores each by:
- Observed frequency vs. theoretical expectation
- Edge (observed rate − theoretical rate)
- Payout ratio
Phase 2 — Window Sizing: After analysis, the bot picks the "hottest" tier and calculates how many bets (window size) are needed to achieve a target cumulative hit probability:
window_size = ceil(ln(1 - confidence) / ln(1 - P(hit per round)))
Example at 15x target (P ≈ 6.5% per round):
| Confidence | Window Size | Cost (at ₹10/bet) | P(at least 1 hit) |
|---|---|---|---|
| 70% (Conservative) | ~18 bets | ₹180 | 70% |
| 50% (Balanced) | ~10 bets | ₹100 | 50% |
| 30% (Aggressive) | ~5 bets | ₹50 | 30% |
Phase 3 — Trigger (Wait Condition): Before firing, the bot can optionally wait for a signal:
| Trigger | Condition | Rationale |
|---|---|---|
| Immediate | Fire on next round | Default — no delay after analysis |
| Dry Spell | Last 5 rounds ALL below 2x | Patience: wait for a "cold" streak |
| Overdue | Target not seen in last 20 rounds | Patience: target feels "due" |
Note: Dry Spell and Overdue do NOT improve mathematical odds (each round is independent). They add discipline/patience by delaying entry.
Phase 4 — Fire: Places the bet with calculated amount and target. If dynamic target is enabled (default), re-observes 50 rounds before EACH individual bet in the window and re-analyzes — the target may shift between bets.
Configurable Parameters:
| Setting | Default | Description |
|---|---|---|
| Observation Rounds | 50 | How many rounds to watch before each bet |
| Max Risk per Window | 5% | Total bankroll % risked across all bets in a window |
| Window Confidence | 50% | Target P(at least 1 hit) — controls window size |
| Dynamic Target | On | Recalculate target between bets within a window |
| Fire Trigger | Immediate | When to pull trigger after observation completes |
Best for: High-multiplier hunting with patience. You accept many rounds of pure observation for occasional big payouts.
Always bets 1% of bankroll. Always cashes out at 1.5x (64.7% win rate). Never changes. Maximum bankroll survival for long sessions with minimal drawdown.
Fixed 1% of bankroll per round. Cashout at 2.0x (48.5% win rate). Balanced risk/reward — wins less often but profits more per win.
Positive progression: doubles bet after each win (1x→2x→4x). Resets to base after any loss or 3 consecutive wins. Capitalizes on hot streaks safely. Never increases after losses.
Structured 4-step sequence on consecutive wins: bet 1, 3, 2, 6 units. Reset to step 1 on any loss. Profits from 4-win streaks (pays 12 units on 12-unit investment). Self-limiting.
Gentle negative progression: +1 unit on loss, −1 on win, never below 1 unit. Much safer than Martingale — linear growth instead of exponential. Slow recovery, high bankroll survival.
Ultra-conservative cycle system. Only increases bet by 1 unit after a win AND only if the cycle is negative. Resets once cycle reaches +1 unit profit. Patient capital preservation.
Evaluates performance every 25 rounds using a composite scoring system and switches to the best-performing strategy automatically.
Rules:
- Minimum 15 rounds on a strategy before switching is considered
- Excludes Goal Seeker and Sniper from rotation (they have dedicated lifecycles)
- Scores based on: win rate, drawdown, EV, volatility-adjusted returns
The floating sidebar shows live quant data:
Decision Engine section:
- EV (expected value of current decision)
- Win Probability (from rolling history)
- Dynamic Cashout Target
- Kelly % being used
- Volatility level (Low/Medium/High/Extreme)
- Bet size for this round
- Decision reason (why it bet or skipped)
Rolling Stats section:
- Mean crash, StdDev
- % above 1.5x, 2x, 3x
- Sample count
- Sparkline of last 20 crash points (color-coded: green >2x, orange >1.5x, red <1.5x)
Bankroll section:
- Starting, Current, Peak, P/L
Session section:
- Rounds played, Win rate, Skipped rounds, Consecutive losses/wins
| Setting | Default | Description |
|---|---|---|
| Starting Bankroll | ₹1000 | Your initial capital |
| Max Rounds | 100 | Auto-stop after this many rounds |
| Drawdown Stop | 5% | Stop if bankroll drops this % from peak |
| Take-Profit | 110% | Stop if bankroll reaches this % of start |
| Simulation Mode | On | Dry run with generated crash points |
When enabled (default), the bot:
- Generates crash points with a weighted distribution mimicking real game behavior
- Runs full quant logic (Kelly, EV, volatility) against simulated data
- Shows timing that mirrors real game phases (betting → flying → crash → pause)
- Does NOT interact with the game DOM
Disable simulation only when you want the bot to evaluate against real detected crash multipliers.
manifest.json — Chrome extension manifest (V3)
background.js — Service worker for storage/lifecycle
content.js — Main bot: game observation, decision loop, simulation
strategies.js — All 9 strategy implementations + auto-switch logic
panel.js — Floating dashboard UI + configuration
panel.css — Panel styles
icons/ — Extension icons
- The house edge is ~3% (provably fair RNG)
- NO strategy can guarantee long-term profit against a negative-expectation game
- Observation does NOT predict future rounds — each crash is independent
- "Hot" tiers, dry spells, and overdue signals are pattern-matching heuristics, not causal predictors
- What math CAN do: optimize bet sizing, manage risk, maximize P(reaching a goal before ruin)
- The Gambler's Ruin theorem proves that with infinite play, the house always wins
- These strategies optimize discipline and risk management, not luck
This is for educational and entertainment purposes. The house edge (~3%) means no strategy guarantees long-term profit. This bot minimizes bad decisions — it does not eliminate risk.