01
Find repeatable behavior
Use historical trades to identify wallets and market patterns with a record worth following.
ProducesTarget wallet cohort
Operator-controlled trading system
Every trade must earn the right to run.
niles turns live market data into reviewable trading decisions across Polymarket, Kalshi, and Alpaca. It can simulate or submit orders within limits an operator can inspect and change.
GitHub identifies approved dashboard operators. Trading and model credentials are not shared with GitHub.
THE DECISION ENGINE
Five stages. One recorded decision.
Each loop narrows the opportunity set before capital is involved. Open any stage to inspect the evidence, algorithms, decision rule, and operator controls.
02
Remove weak candidates
Apply deterministic screens before model scoring so cost and attention stay on credible opportunities.
ProducesAccepted queue or refusal
03
Build the probability case
Evaluate each survivor from several angles and record the probability, confidence, thesis, and cost.
ProducesProvider decision record
04
Form a trade decision
Run focused strategies independently, then use agreement strength to determine whether a trade can continue.
ProducesFull, half, or no position
05
Size, submit, and monitor
Turn an approved decision into bounded exposure, then watch confirmed positions for a reason to exit.
ProducesOrder intent and position lifecycle
THE LIVE-ORDER BOUNDARY
A good idea is not enough to place a trade.
A decision can exist without being eligible for execution. These checks form the final boundary between an approved thesis and an order.
Any gate failsNo live order. The refusal stays visible in the audit trail.
01
Operator
Approved identity, permitted account mode, live setting, and emergency stop state.
02
Market data
Current, supported, sufficiently liquid, and open for trading.
03
Model decision
A complete provider decision with sufficient confidence and available budget.
04
Strategy consensus
Recorded signals satisfy the configured agreement rule.
05
Risk
The proposed exposure stays inside every capital limit.
06
Execution
The venue account is reconciled and the order intent is safely persisted.
PRODUCT BOUNDARIES
Automation with explicit limits.
You decide where the bot can operate and how much exposure it can take. niles versions those settings, applies them on the next loop, and records what it did with them.
Explicit venue control
Venues do not become eligible because credentials exist. An operator must enable the venue and choose the permitted account mode.
Conservative scope
Kalshi supports standard binary markets on the primary subaccount. Alpaca supports long stocks and ETFs plus disabled-by-default, easy-to-borrow U.S. equity shorts. Unsupported contracts, options, crypto, hard-to-borrow locates, and margin-funded long purchases are refused.
Venue-confirmed results
Performance uses confirmed balances, positions, and fills. Simulated or unfilled orders do not appear as actual results.
OPERATOR ACCESS
Inspect the system before you change it.
Sign in to review current mode, blocked gates, model decisions, risk settings, audit events, and venue-confirmed performance.