PF probe kanuck-probe-03-max-bars-back-500

// This Pine Script® code is licensed under Apache-2.0
// SPDX-License-Identifier: Apache-2.0
// © PineForge contributors 2026
//
// PF probe kanuck-probe-03-max-bars-back-500 — deep-history reference chain
// Clean-room: written from validation goal, not from source.
//
// Purpose: isolates `max_bars_back=500`. The engine surface under test is
// the runtime's deep-history buffer: a chain that combines `ta.sma(close,
// 400)`, `ta.rsi(close[200], 14)`, and a raw `close[450]` reference forces
// the runtime to retain at least 500 bars of every series touched. Verifies
// that the engine handles the deep buffer without truncation artefacts and
// matches TradingView's broker emulator on the bars where these terms first
// become non-na.
//
// Validation goal: bit-exact entries on `close > close[450]` (long) and
// `close < close[450]` (short) — a state-flipping condition off the
// deep-history buffer that fires many times across the 36k-bar window.
// `ta.sma(close, 400)` and `ta.rsi(close[200], 14)` are computed on
// every bar (and would also be plotted in TV) so the engine still has
// to retain the 500-bar buffer for those terms — they are the deep-
// history terms under test even though they no longer gate entries.
// (The earlier triple-AND form is statistically far too rare on 15m
// crypto to produce a meaningful trade-list parity check.)
// Pass = identical trade list vs TradingView.
//
// TV setup: 15m chart, ETH-USDT-USDT (Binance USDT-M Perpetual). Full window
// matching corpus/data/ohlcv_ETH-USDT-USDT_15m.csv. Strategy properties:
// leave defaults from script header. Export "List of trades" → tv_trades.csv.
//@version=6
strategy("PF probe kanuck-probe-03-max-bars-back-500", shorttitle="PF_kan03_MBB",
     overlay=true, initial_capital=1000000,
     commission_type=strategy.commission.percent, commission_value=0,
     slippage=0, default_qty_type=strategy.fixed, default_qty_value=1,
     pyramiding=1, process_orders_on_close=false,
     max_bars_back=500)

// ---- inputs --------------------------------------------------------
int i_sma_len    = input.int(400, "Long SMA length",         minval=50)
int i_rsi_offset = input.int(200, "RSI source offset",       minval=1)
int i_deep_lag   = input.int(450, "Close-reference lookback", minval=1)

// ---- deep-history terms -------------------------------------------
// All three are computed every bar and force the runtime to retain
// the 500-bar buffer. `long_sma` and `deep_rsi` are the corroborating
// deep-buffer terms; `deep_close` is the entry trigger.
float long_sma   = ta.sma(close, i_sma_len)
float deep_rsi   = ta.rsi(close[i_rsi_offset], 14)
float deep_close = close[i_deep_lag]

// Plot the corroborating deep-history terms so the engine has to keep
// them resolved even though they no longer gate entries.
plot(long_sma, "deep SMA", color=color.new(color.blue,   0))
plot(deep_rsi, "deep RSI", color=color.new(color.purple, 0), display=display.data_window)

// ---- composite condition ------------------------------------------
bool deep_ready = not na(long_sma) and not na(deep_rsi) and not na(deep_close)

bool long_trigger  = deep_ready and close > deep_close
bool short_trigger = deep_ready and close < deep_close

// ---- order routing -------------------------------------------------
if long_trigger and strategy.position_size <= 0
    strategy.entry("L", strategy.long, qty=1, comment="deep-history long")

if short_trigger and strategy.position_size >= 0
    strategy.entry("S", strategy.short, qty=1, comment="deep-history short")