bodhi.

Solana Meme Coin Trading Intelligence · Paper Trading
Checking...
At a glance
Closed Trades
all time
Win Rate
target: above 50%
Open Positions
max 10
Random Baseline
~50%
coin flip benchmark
Returns
Net P&L
all closed trades
Total Won
Total Lost
Avg Win
per winning trade
Avg Loss
per losing trade
Biggest Win
moonshot
Learning curve
Win rate calculated over rolling 50-trade windows. The dotted line is random chance (50%). An upward trend means learning is happening.
Feature health
Each feature is one thing Bodhi observes about a token before deciding to trade. Click any feature to learn what it means and how Bodhi uses it. A weight near zero means the feature has not influenced decisions yet. Amber means the weight is exactly zero — worth watching.
Recent trades
Token
Entry Price
PnL
Result
Decision
Loading trades...
The journey

Where we are. Where we're going.

Bodhi is a scientific experiment, not a trading system. We are building a machine that discovers market truth for itself — without human assumptions, without hand-coded signals. One stage at a time. One proof at a time.

Stage 0
Foundation
Built the infrastructure. Feed, bandit, trade loop. Replaced v2's 33-process chaos with 3 clean files.
Complete
Stage 1
Multi-Armed Bandit
Bodhi observes 7 features and learns which ones predict wins. No human assumptions. Let the market teach.
Now — Active
Stage 2
Observer Bodhi
A second bot learns from all market data — not just its own trades. Compare on-policy vs off-policy learning.
Locked
Stage 3
Neural Network RL
When bandit weights stabilise, upgrade the brain. Deep learning discovers patterns no human could encode.
Locked
Stage 4
Live Trading
Real money, only after 30+ days profitable paper trading, statistical significance proven, risk agreed in writing.
Locked
Go-live criteria — no exceptions
📊
Statistical significanceWin rate above 50% with p < 0.05 over 500+ trades. Not a lucky streak — proven math.
🧪
Out-of-sample validationPattern confirmed on data not used in training. If it only works on what it learned from, it is not real.
📅
30 consecutive days profitableNot one good week. Thirty days of paper trading profit in a row, without changing anything.
✍️
Risk agreed in writingMaximum stake decided and written down before the first live trade. No exceptions, no improvising.
The casino doesn't win every hand. It doesn't need to. It has a mathematical edge and the discipline to apply it consistently, without emotion, without changing the rules because it had a bad night. That is what we are building.
The Bodhi Principle Small edge Consistent application No emotion Time
Scout
Pump.fun graduation event paper trading

Scout monitors Pump.fun for the moment a token graduates — the transition from a controlled bonding curve to open trading on Raydium DEX. This is a defined, on-chain event with measurable mechanics: when a token sells through 100% of its bonding curve, it migrates to Raydium and liquidity unlocks. Scout detects that migration in real time, enters a paper $50 trade, and collects price snapshots at 1, 5, 15, 30, 60, and 240 minutes post-graduation to map the typical price trajectory.

Entry filters: minimum $50k pool liquidity, token age ≤ 30 minutes. Exit rules: +50% take profit, −30% stop loss, 15-minute hard timeout. The snapshot dataset is building one answer: do Pump.fun graduates pump, dump, or consolidate after bonding curve exit?

Bot Status
Offline
Events Today
graduations detected
Trades Today
Today PnL
paper USD
All Time PnL
Win Rate
all time
Recent Graduations
Last 50 detected events
Token
Age
Market Cap
Liquidity
Status
Monitoring for graduation events…
Edge Analysis
Pattern emerges as data accumulates
1 min
5 min
15 min
30 min
1 hour
4 hours
Accumulating data…
Minimum 50 events required for statistical relevance
Trade History
Token
Entry
Exit
Reason
PnL USD
PnL %
No closed trades yet
Running total
Seisma
Funding rate squeeze detector — perp markets

Seisma monitors all Drift Protocol perpetual futures markets for extreme funding rate imbalances. When funding turns sharply negative (below −50% annualised) and long open interest is lopsided (OI ratio ≥ 1.6), overleveraged longs are under mechanical stress — a forced unwind becomes probable. Seisma reads that pressure and paper trades a short before the squeeze releases. The same logic applies in reverse: extreme positive funding combined with short-heavy OI signals a long squeeze.

Signals require two consecutive poll confirmations (2 min) to filter noise. DexScreener spot price confirms at trigger. Exit rules: +40% take profit, −25% stop loss, 10-minute hard timeout. The core hypothesis: funding rate extremes + OI imbalance = mechanical squeeze that is predictable before it releases.

Bot Status
Offline
Signals Today
conditions detected
Trades Today
Today PnL
paper USD
All-Time Win Rate
all time
All-Time PnL
Recent Signals
Last 20 squeeze conditions detected
Market
Dir
Funding
OI Ratio
Status
Monitoring funding rates…
Squeeze Analysis
Pattern emerges as data accumulates
False Signal Rate
stop-loss exits
avg min to exit
Short squeeze
Long squeeze
-50% → -100%
-100% → -200%
below -200%
+50% → +100%
above +100%
Market Conditions
Live funding rates across monitored perp markets — red below -50% ann, green above +50% ann
Market
Funding Rate
OI Ratio
Signals 30d
Win Rate 30d
Monitoring funding rates…
Trade History
Paper trades — $50 position size
Market
Dir
Entry
Exit
Reason
PnL %
PnL $
No closed trades yet
Vigil
The data collector — token graduation intelligence
Tokens Collected
total in database
Date Range
graduation dates covered
Avg Quality Score
0–100 scale
Backfill Status
Live Monitor
realtime collector
Backfill progress
Moonshot probability calculator
Strategy inputs
Win rate (%)19
Avg win multiplier2.3×
Avg loss % of bet40%
Bet size ($)$50
Moonshot rate % of wins3%
Moonshot multiplier50×
What if Vigil improves detection?
Improved moonshot rate6%
New bets to moonshot
New PnL / 100 bets
Improvement factor
Expected outcomes
Kelly fraction
Bets to moonshot
Capital needed
PnL / 100 bets
Moonshot value
Break-even bets
Recent graduations
Token
Graduated
Market Cap
Quality
Outcome
No tokens collected yet