PromptBase
PromptBase SelectGet Select
Close icon
General
Home
Marketplace
Create
Hire
Login
Chat
Sell
Explore

Prompt Details

Model
(claude-5-5-sonnet)
Token size
1,193
Example input
[CURRENT STACK AND ROLES]: EMA 50 and EMA 200 cross: trend and entry trigger. Exit on the opposite cross. [MARKET AND TIMEFRAME]: EURUSD, H1 [WEAKNESSES]: Many false entries when the market is ranging [TESTED AND TEST DATA]: Nothing rejected yet. Three years of H1 data, the last year can stay out of sample. [TARGET AND CONSTRAINTS]: Pine Script v6; maximum 2 new parameters
Example output
1. STACK MAP | Indicator | Family | Decision it changes | Notes | | EMA 50 / EMA 200 cross | Trend | Entry and exit | One rule drives both decisions | 2. OVERLAPS AND COVERAGE GAPS Overlaps: none, there is only one indicator. | Family | Status | Related weakness | | Trend | Covered (direction) | Not the strength of the trend | | Momentum | Not covered | Possible link with false entries | | Volatility | Not covered | Possible link with ranging markets | | Volume | Not covered | No link stated | | Structure and levels | Not covered | No link stated | | Session and time | Not covered | Possible link | The weakness points to a missing way to tell a ranging market from a trending one. 3. PROPOSALS A. Trend-strength filter (for example ADX). Family: trend (strength). Decision: filter on entries. Targets: false entries in ranges. Adds 2 parameters (length, threshold). Starting setting to test: length 14, threshold 20 (not optimal). Side effects: fewer trades, some trends entered later. Redundancy: medium, because ADX is built from price movement like the EMAs. How it could fail: the threshold is tuned to the sample, or the market regime changes. B. Volatility filter (for example ATR compared with its own average). Family: volatility. Decision: filter. Targets: quiet periods. Adds 2 parameters (ATR length, average length). Starting setting to test: ATR 14, average 50 (not optimal). Side effects: skips quiet periods, may skip the start of some trends. Redundancy: low. How it could fail: ranges that occur with high volatility. C. Session filter. Family: session and time. Decision: filter. Targets: low-activity hours. Adds 2 parameters (start hour, end hour). Starting setting to test: (TO COMPLETE: session hours). Side effects: fewer trades. Redundancy: low. How it could fail: false entries spread through the whole day. 4. RECOMMENDATION Start with A. Selection-table row applied: "ranging" points to trend strength. A answers the stated weakness directly and leaves the entry rule untouched. This is a hypothesis from the logic of the map, not a result. B is the second choice only if A is rejected in testing. 5. RULE SPECIFICATION Keep the base rule: long when EMA 50 crosses above EMA 200 on a closed bar, short on the opposite cross, exit on the opposite cross. Add: take an entry only if the ADX (length 14) was above 20 on the previous closed bar. 6. A/B TEST PLAN Same three years of H1 EURUSD data with the last year out of sample, same costs, same exit rule. Version A: entries without the filter. Version B: entries with the ADX filter. Record, in sample and out of sample: number of trades, expectancy in R, maximum drawdown, win rate. Reject the filter if the improvement disappears out of sample, or if the number of trades falls below (TO COMPLETE: minimum number of trades). 7. CODE REQUEST (Pine Script v6) Build one Pine Script v6 strategy for EURUSD on H1 with an input switch "Use ADX filter" (false = version A, true = version B). Base rule: long when EMA 50 crosses above EMA 200 on a closed bar, short on the opposite cross, exit on the opposite cross. Version B only: take an entry only if the ADX was above the threshold on the previous closed bar. Inputs with defaults: EMA fast 50, EMA slow 200, ADX length 14, ADX threshold 20. Costs: (TO COMPLETE: spread and commission per trade). Use closed-bar values only and no future data. Display the two EMAs and the filter state. Output in the strategy report, separately for in sample and out of sample: number of trades, expectancy in R, maximum drawdown, win rate. Explain every assumption. The user will compile and check the code. 8. UPDATED STACK STACK MAP v2 (pending test): EMA 50 / EMA 200 cross (trend, entry and exit) + ADX length 14, threshold 20 (trend strength, filter on entries). 9. MISSING INFORMATION (TO COMPLETE: spread and commission per trade; minimum number of trades; session hours if you later test proposal C.) No stop-loss or position size rule was provided beyond the opposite cross.
Claude

Trading Strategy Stack Builder

Add to Cart
Instant accessInstant access
Usage rightsCommercial use
Money-back guaranteeMoney‑back
By purchasing this prompt, you agree to our terms of service
CLAUDE-5-5-SONNET
Tested icon
Guide icon
4 examples icon
Free credits icon
Random indicators overfit a strategy. Give this prompt your stack, market, weaknesses and what you tested: it maps what each indicator covers, proposes up to 3 logical additions chosen by a fixed rule (real gap, one decision, low redundancy), recommends one, and writes an A/B test plan plus a ready-to-paste CODE REQUEST for Pine Script v6 or Python. No signals, no profit claims: you run the tests.
...more
Added 1 day ago
Report
Browse Marketplace