feat: wavelets simulation

frontend/src/lib/waveletTrade.ts

@@ -481,6 +481,8 @@ export function runOptimizationGrid(
 // lookback x wavelet combo) sequentially — deliberately not parallel, so the
 // backend isn't hammered with concurrent CPU-heavy CWT requests. The cheap part
 // (grid of trigger configs per fetched series) runs entirely in-browser.
+export type WaveletDateRange = { startDate: string; endDate?: string } | null
+
 export async function runFullOptimization(
   instruments: string[], lookbacks: number[], wavelets: string[], period: string, levels: number,
   bandsToTest: number[], kindsToTest: WaveletTriggerKind[], modesToTest: WaveletPositionMode[],
@@ -489,6 +491,7 @@ export async function runFullOptimization(
   onBatchResults: (results: WaveletOptimizationResult[]) => void,
   method: 'cwt' | 'ssq' = 'cwt',
   filterBandIndex: number | null = null,
+  dateRange: WaveletDateRange = null,
 ): Promise<void> {
   const combos: { symbol: string; lookback: number; wavelet: string }[] = []
   for (const symbol of instruments) {
@@ -504,7 +507,10 @@ export async function runFullOptimization(
     onProgress(i, combos.length, `${symbol} @ ${lookback}j (${wavelet})`, failures)
     try {
       const { data: rolling } = await api.get('/wavelet/rolling', {
-        params: { symbol, period, levels, wavelet, lookback, step: 1, method },
+        params: {
+          symbol, levels, wavelet, lookback, step: 1, method,
+          ...(dateRange ? { start_date: dateRange.startDate, end_date: dateRange.endDate } : { period }),
+        },
       })
       onBatchResults(runOptimizationGrid(symbol, lookback, wavelet, rolling, bandsToTest, kindsToTest, modesToTest, exitVariants, filterBandIndex))
     } catch (error) {