Credit drivers and costs
Detailed breakdown of every action that consumes credits and how to estimate your monthly usage.
This article covers every action that consumes credits in Coinrule, how much each costs, and how to estimate your monthly usage. A typical RSI+EMA strategy running on 1-hour candles incurs around 29 credits per month in recurring costs — well within the 500-credit monthly allowance on the Starter plan.
Credit cost breakdown
| Driver | Cost |
|---|---|
| AI strategy parsing (LLM) | Haiku: 1×, Sonnet: 3×, Opus: 15× — per 1,000 tokens |
| Strategy deployment (base) | 1 credit per deploy |
| Entry / exit conditions | 1 credit each |
| Indicators (above 3 free) | 2 credits each |
| Multi-market bonus | +2 credits |
| Data subscription (per watched market / day) | 0.5 credit |
| Sub-minute interval (per market / day) | 1 credit |
| Position open (execution) | 2 credits |
| Position close (execution) | 1 credit |
| Volume | 0.1 credit per $1,000 traded |
| Third-party signal (TradingView webhook) | 1 credit per signal |
How do I estimate monthly usage?
The scenario below is illustrative. Exact costs depend on how often your strategy triggers and your position sizes.
Example: RSI + EMA strategy on BTC/USDT, 1h candles, live on Binance
| Component | Approx. monthly cost |
|---|---|
| Data subscription (1 pair, 1h) | ~15 credits |
| AI parse (one-time, Haiku, ~5–15 credits) | — (one-time) |
| Strategy deploy (one-time, complexity-based) | — (one-time) |
| 2 trades/week × open + close | ~12 credits |
| Volume (2 × $1,000/trade) | ~1.6 credits |
| Total recurring | ~29 credits/mo |
A Starter plan includes 500 credits/month — enough for many similar strategies simultaneously.
How is the LLM cost calculated?
credits = ceil((input_tokens + 3 × output_tokens) / 1000) × model_multiplier
input_tokens= prompt + system context lengthoutput_tokens= length of the JSON the model returnsmodel_multiplier= see AI models overview
For most strategies, the LLM cost is a small fraction of the monthly total (3–15 credits per parse).
Data subscription cost formula
daily_cost = unique_watches × 0.5 + sub_minute_watches × 1
Where a "watch" is a unique (symbol, interval, venue) combination used by running strategies. sub_minute_watches is the count of pairs running on the 1m interval.
A 10-pair scan on 1h candles costs 5 credits/day in data (~150 credits/month).
Volume cost
daily_volume_cost = (total_traded_usd / 1000) × 0.1
Typically negligible for small accounts (below $5,000 portfolio).