Prerequisites
Cryptohopper MCP configured in an MCP client — see the setup overview.
Adventurer tier or higher recommended. DCA range analysis benefits from a full year of history. Explorer (90-day cap) works for short-horizon DCA but under-samples drawdowns — see subscription tiers.
A token you have already decided to accumulate, and a rough sense of total capital to deploy.
A DCA range picker is not a bottom caller. It tells you "here's the price band where past drawdowns historically resolved" — so you can place orders in that zone rather than at random.
Setup steps
Issue the core prompt
Open your MCP client and send the prompt below. Adjust the token pair, capital amount, and window to match your plan.
Review the proposal
A solid response gives you a concrete upper and lower bound, a tranche plan with specific prices and sizes, and an honest fallback if the market breaks lower. Reject vague outputs.
Sanity-check against your own view
The model works from candle data only. It cannot account for upcoming token unlocks, macro events, or fundamental developments. Treat the proposal as a data-grounded starting point.
Place the DCA orders
Options: manually via your exchange or via the Cryptohopper REST API for end-to-end automation — see combine MCP + Trading API.
Re-run the range picker weekly
After a material price move the proposed range is stale. Weekly reruns during active accumulation keep the plan current.
Core prompt template
Using the Cryptohopper MCP, help me plan a DCA entry for SOL/USDT on Binance.
My total planned deployment: $10,000
My deployment window: 4 weeks
My risk preference: moderate (I want to average in without chasing)
Pull:
- The last 365 × 1d candles.
- The last 90 × 4h candles for short-term structure.
Compute:
- Current price.
- 90-day high and low.
- 365-day high and low.
- Distribution of daily closes over the last 90 days: 25th, 50th, 75th percentile.
- Typical drawdown size from recent swing highs (average depth of 5%+ pullbacks in the last 90 days).
Propose:
- An entry range (lower and upper bound), grounded in the drawdown distribution.
- A suggested split of the $10,000 across 4–6 tranches within that range (larger tranches lower).
- A fallback plan if price breaks below the proposed lower bound (buy the breakdown, pause, or reassess).
Show your reasoning for each choice.
Rules of thumb to encode
Add these to your prompt so results stay consistent across runs. The specific values matter less than having stable rules — a plan the agent reinvents each run is a plan you cannot trust.
Element | Starting rule |
Lower bound | ~25th percentile of recent daily closes, or recent swing low minus 1 ATR |
Upper bound | Current price or the 50th percentile (median close), whichever is lower |
Tranche count | 4–6. Fewer is fine; more creates operational overhead |
Tranche sizing | Larger tranches at lower prices — e.g. 10 / 15 / 20 / 25 / 30% of total |
Tranche spacing | Roughly even in price, biased tighter at the lower end |
Breakdown rule | If price drops a full ATR below the lower bound, pause and reassess |
Cost profile
Action | Adventurer | Hero |
365 × 1d candles | 20 units | 1 unit |
90 × 4h candles | 5 units | 1 unit |
Per plan | ~25 units | ~2 units |
4-week window (weekly refresh) | ~100 units | ~10 units |
See rate limits explained for full detail.
Troubleshooting
The proposed lower bound is above the current price
The market is in a downtrend and the agent is anchoring on historical levels that have already been broken. Accept that your entry starts from the current price, or instruct the agent: "If the current price is below the historical 25th percentile, use recent swing lows instead."
The tranche count changes between runs
Encode it explicitly: "Always propose exactly 5 tranches." The model varies by default unless instructed otherwise.
The plan doesn't match your conviction level
Pass it in explicitly. High conviction → weight tranches toward the current price to avoid missing upside. Speculative → weight toward lower bands to avoid overpaying.
HISTORY_LIMIT_EXCEEDED on the 365-day query
Your tier doesn't support that depth. Explorer caps at 90 days — use that, or upgrade to Adventurer. See subscription tiers.
The fallback plan is too generic
Instruct the agent to be specific: "Name the exact price level at which you would re-evaluate, and what conditions — volume pickup, RSI, trend reversal — would need to be true to resume buying."
You want to track whether the plan is working
Keep a simple log. After each tranche fills (or doesn't), note the price and compare against the proposal. Over a few cycles you learn whether the agent tends to be too conservative, too aggressive, or roughly right — more useful than trusting any single run.
You want to run this on many tokens at once
Each token costs ~25 call units on Adventurer, so a 10-token weekly re-run is ~250 units. Be honest about how many concurrent DCA positions you can actually manage — a plan you don't execute is just homework.