/**
* Merged Patterns Tool
* Combines: fractals, zigzag_indicator, trend_detection
*/
import { z } from 'zod'
import {
calculateFractals,
calculateZigZag,
} from '../../signal-generation/technical-indicators'
import { detectChangeOfCharacter } from '../../signal-generation/analysis/market-structure'
export const patternsInputSchema = z.object({
type: z.enum([
'fractals',
'zigzag',
'change_of_character'
]).describe('Type of pattern indicator to calculate'),
// Price data
closes: z.array(z.number()).min(5).describe('Array of closing prices'),
highs: z.array(z.number()).describe('Array of high prices'),
lows: z.array(z.number()).describe('Array of low prices'),
// ZigZag specific
deviation: z.number().min(0.1).max(20).optional().describe('Minimum deviation percentage for ZigZag (default: 5)'),
// Change of Character specific
lookback: z.number().int().min(10).max(200).optional().describe('Lookback period for COC (default: 50)'),
})
export type PatternsInput = z.infer<typeof patternsInputSchema>
export async function calculatePatterns(input: PatternsInput): Promise<any> {
const {
type, closes, highs, lows,
deviation = 5,
lookback = 50
} = input
switch (type) {
case 'fractals': {
const result = calculateFractals(highs, lows)
if (!result) {
throw new Error(`Insufficient data for Fractals`)
}
return { type, ...result }
}
case 'zigzag': {
const result = calculateZigZag(closes, deviation)
if (!result) {
throw new Error(`Insufficient data for ZigZag`)
}
return { type, ...result }
}
case 'change_of_character': {
const historicalData = closes.map((close, i) => ({
time: Date.now() - (closes.length - i) * 60000,
open: closes[i > 0 ? i - 1 : 0],
high: highs[i],
low: lows[i],
close: close,
volume: 0
}))
const result = detectChangeOfCharacter(historicalData, closes[closes.length - 1])
if (!result) {
throw new Error(`Insufficient data for Change of Character`)
}
return { type, ...result }
}
default:
throw new Error(`Unknown pattern type: ${type}`)
}
}
export function registerPatternsTool(server: any) {
server.registerTool(
'patterns',
{
title: 'Patterns (Unified)',
description: `Calculate various pattern recognition indicators. Supported types:
- fractals: Bill Williams Fractals - identifies swing highs and lows
- zigzag: ZigZag Indicator - filters noise to show significant moves
- change_of_character: COC - detects market structure changes`,
inputSchema: patternsInputSchema,
outputSchema: z.object({
type: z.string(),
}).passthrough(),
},
async (input: PatternsInput) => {
try {
const result = await calculatePatterns(input)
return {
content: [
{
type: 'text',
text: JSON.stringify(result, null, 2),
},
],
structuredContent: result,
}
} catch (error: any) {
throw new Error(`Patterns calculation failed: ${error.message}`)
}
}
)
}
