Automated Crypto Trading Bot
A separate crypto research project alongside Claw Quant. The latest Momentum V1 system tests reproducible trading rules against market data, checks costs and risk, and records virtual trades. Live execution is disabled in the current implementation; this is not a production trading service or a claim of profitability.
Built by Matt Schueren.
From market data to a paper trade.
Deterministic rules handle strategy, risk, and simulated orders. AI helps with development and analysis.
- 01
Validate the data
Load public Coinbase market data and check timestamps, missing bars, and price consistency.
- 02
Rank & size
Apply momentum and basket-regime rules, then calculate volatility-aware targets within position caps.
- 03
Check costs & risk
Compare the estimated edge with trading costs and apply independent freshness, account, and loss checks.
- 04
Simulate & audit
Record virtual fills, rejected signals, and portfolio state in SQLite. Real-order submission is blocked; no live results feed is connected to this page.
What the bot checks
Inputs and controls used in research. No live metric values or performance feed are connected here.
- Momentum
- Price movement across completed bars drives candidate ranking.
- Basket regime
- The broader asset basket provides context for individual signals.
- Volatility & concentration
- Inverse-volatility sizing and position caps constrain exposure.
- Estimated trading costs
- Fees, spread, and slippage are compared with an estimated edge. That edge remains an uncalibrated research proxy.
- Data freshness
- Timestamp and missing-bar checks can prevent a signal from proceeding.
- Account & loss limits
- Unexpected account state or loss-limit breaches can halt the system; a halt does not automatically liquidate positions.
Testing the idea
Chronological replay, walk-forward research, and stress scenarios test the rules. Trial records, virtual fills, and rejected signals stay in SQLite so losing experiments remain part of the record. Live execution is disabled; simulations do not establish profitability.
Python · Coinbase market data · Momentum V1 · SQLite · Backtesting · Read-only Robinhood adapters