correct RSI baseline sizing documentation

Author: dharmanan

research/rsi-divergence-reversal/research-plan-v1.md

@@ -1,6 +1,6 @@
-# RSI Divergence Reversal — Frozen Research Plan v1
+# RSI Divergence Reversal — Frozen Baseline Research Plan v1
 
-Status: frozen before performance results
+Status: baseline completed; current preset failed development and validation performance checks
 
 ## Scope
 
@@ -20,7 +20,7 @@ Research window:
 - validation: 2023-01-01 through 2024-12-31
 - final holdout: 2025-01-01 onward, closed
 
-## Frozen strategy rules
+## Frozen baseline rules
 
 - long and short
 - confirmed candles only
@@ -32,14 +32,17 @@ Research window:
 - short requires RSI <= 60 at the divergence confirmation bar
 - cooldown: more than 5 bars since the previous accepted long or short signal
 - entry: confirmation-bar close
-- position sizing: 100% of strategy equity
+- TradingView preset order quantity: one instrument unit, inherited from the existing generated Strategy declaration
+- normalized reference research: one independent return unit per trade, without compounding
 - pyramiding: disabled
 - opposite accepted signal reverses the position at that bar's close
 - ATR14 frozen at entry
 - initial stop: 2 ATR
 - target: 2:1 reward/risk, equal to 4 ATR from entry
 - stop and target become eligible from the first bar after entry
 
+The TradingView dollar result and the reference engine's normalized return total are therefore not expected to have the same numerical scale. Trade timestamps, directions, stop/target levels and exit reasons are the parity targets.
+
 ## Execution profiles
 
 1. `tradingview_default`
@@ -62,30 +65,16 @@ Research window:
    - 0.10% adverse slippage per side
    - stop first on ambiguous bars
 
-## Report requirements
-
-For each partition, profile and symbol:
-
-- closed trades
-- winning and losing trades
-- win rate
-- total normalized net return
-- average normalized net return
-- profit factor
-- maximum normalized drawdown
-- ambiguous intrabar trade count
-- long/short trade counts
-- exit-reason counts
+## Baseline conclusion
 
-A combined trade ledger must be written for each execution profile.
+The current preset failed in both development and validation. All three symbols were negative, profit factor was below one in every profile, and realistic costs worsened the result. This result is the frozen baseline for the separately specified improvement study; it must not be overwritten or presented as a profitable strategy.
 
-## Correctness gates before interpretation
+## Correctness gates retained
 
-- generated Pine strategy uses 100% equity sizing
 - generated Pine freezes ATR at entry
 - indicator and strategy use the same confirmed regular-divergence events
-- no lookahead or 2025+ data
+- no future-data lookahead and no 2025+ data
 - aggregation uses only complete contiguous groups of three official 5-minute candles
 - synthetic tests cover aggregation, divergence confirmation, next-bar risk activation, ambiguous intrabar ordering and cost arithmetic
 
-No claim of profitability is allowed until the runner and all tests complete successfully.
+No improved rule may replace the preset until it is selected from development data and independently passes the frozen validation gates.