2-Week Backtest Report

FL2 (Fashionably Late V2) -- VWAP Band Pullback Strategy
March 30 - April 10, 2026 -- 9 Trading Days -- 4 Symbols
Backtest Complete Multi-Band Trail GLD / USO / IBIT / NVDA 129 Trades 296 Alert Events
Net P&L (9 days)
-$130.67
Trades
129
Win Rate
51.9%
TP1 Hit Rate
33.9%
Avg Winner
+$44.09
Avg Loser
-$49.93
Best Trade
+$273.28
Worst Trade
-$214.36
Winning Days
3 of 9
Best Day
+$969.28
Worst Day
-$891.65
Alert Events
296

How FL2 Works

FL2 is a VWAP band pullback strategy. It waits for a directional signal, confirms momentum with a push to outer bands, then enters on the pullback to the first band. Exits are managed with a partial take-profit and a trailing stop using VWAP bands.

Step 1
EMA Crosses VWAP
EMA4 on 2-min chart crosses VWAP -- directional signal
▶
Step 2
Push to Band1-Band2
Impulse move to 0.5-1.0 SD confirms momentum
▶
Step 3
Retrace to Band1
Entry zone at +0.5 SD -- buying the pullback
▶
Step 4
TP1 at Band3
Exit 50% at +2.0 SD target
▶
Step 5
Runner Trails
Remaining 50% trails via multi-band system

Multi-Band Trail System NEW

After TP1 fires, the remaining 50% ("the runner") is trailed simultaneously behind 3 VWAP band levels. The tightest trail that gets hit triggers the exit. All 3 are logged so we can analyze which performs best over time via the param surface.

Visual: Long Trade VWAP Band Layout
Band3 (+2.0 SD)
TP1 TARGET (exit 50%)
Band2 (+1.0 SD)
Impulse confirmation zone
Band1 (+0.5 SD)
ENTRY ZONE (pullback buy)
VWAP (0 SD)
Anchor line
Runner Trail Levels (after TP1)
BandTrail_Tight
+0.5 SD -- Aggressive, locks profit fast
BandTrail_Mid
+1.0 SD -- Moderate (Trader Dale standard)
BandTrail_Wide
+2.0 SD -- Loose, max room for runners

Key insight: We don't know which trail width is optimal yet. By running all 3 simultaneously, we collect data on all of them and let the param surface tell us which produces the best risk-adjusted returns.

Daily P&L

Cumulative Equity Curve

Exit Reason Breakdown

P&L by Symbol

P&L by Exit Reason

Wins / Losses by Symbol

Daily Summary

Date Trades Winners Losers Win Rate Net P&L Best Trade Worst Trade Cumulative

Symbol Breakdown

Symbol Trades Winners Losers Win Rate Total P&L Avg Winner Avg Loser Best Trade Worst Trade

Key Insights

Nearly breakeven across 129 trades: -$130.67

The strategy found setups consistently across all 9 days and 4 symbols but oscillated around breakeven. Win rate of 51.9% with avg winner ($44.09) slightly smaller than avg loser (-$49.93) explains the slight net loss. The system is clearly identifying setups -- it needs parameter optimization to tilt the edge positive.

March 30 was the worst day (-$891.65) -- first-day blowup

20 trades on Day 1 with 12 losers. This was the first day of the backtest and the strategy took too many positions. The worst trade was GLD short at -$214.36. Likely a trending day that whipsawed the pullback strategy. Consider: was March 30 a high-VIX day? The number of entries (21 alert entries) suggests the qualification filter was too permissive.

April 9 was the best day (+$969.28) -- 18 winners out of 24 trades

75% win rate on April 9. The best trade was USO short for +$273.28. USO and GLD both had large directional moves that the band trail captured perfectly. IBIT also contributed positive P&L. This was the ideal day for a pullback strategy: strong trends with clean pullbacks to Band1.

BandTrail_Wide is the clear winner among trail types

33 exits via BandTrail_Wide generated +$838.65 total P&L with 28 wins / 5 losses (85% win rate, avg +$25.41). BandTrail_Mid had only 4 exits (+$369.96) and BandTrail_Tight had 3 exits (-$20.46). The wider trail gives runners the room they need while still capturing profits on trend continuation.

Trend_Break was the biggest P&L drag: -$986.00 across 48 exits

48 exits via Trend_Break (47% of all exits) with only 13 wins and 35 losses. These are trades that never reached TP1 and were stopped out when the trend reversed. This is the primary area for optimization: either tighten entry criteria to avoid false setups, or add a faster exit for trades that show no momentum after entry.

NVDA was the worst performer: -$370.05 across 22 trades

Only 9 winners vs 12 losers for NVDA. The stock's high price and wide intraday range created large stop distances that amplified losses. Consider whether NVDA needs different band width parameters or whether its price action is too volatile for the standard FL2 settings.

GLD, USO, and IBIT were all profitable

GLD: +$118.58 (38 trades), USO: +$112.97 (44 trades), IBIT: +$146.32 (25 trades). All three had positive total P&L despite losing days. The strategy works well on liquid ETFs with moderate intraday ranges.

Stop_UpperBand and Stop_LowerBand cost -$1,489.57 combined

12 full-stop exits (8 upper, 4 lower) produced exclusively losses. These are trades that immediately went against the position and hit the stop band. Improving entry timing or adding a confirmation candle could reduce these.

TP1 hit rate of 33.9% (40 out of 118 entries)

Only about 1 in 3 entries reaches the +2.0 SD target for partial profit. The remaining 2/3 are exited before TP1 via stop or trend break. This suggests the TP1 target may be too ambitious for some market conditions, or entries are happening too late in the move.

What's Next

✓

2-Week Backtest -- COMPLETE

March 30 - April 10, 2026 (9 trading days, 129 trades, 296 alerts). Submission hash: 271b7ed5da7c971d40db37442caf9532.

▶

Forward Test -- LIVE on Account 203955

FL2 V2 with multi-band trail is running live in paper trading. Real-time alerts validating backtest findings against live market conditions.

1

Webhook Update for Band4/Band5 Data Collection

Add +3.0 SD and +4.0 SD band targets to the alert webhook. This expands the param surface from the current 3-band trail to a 5-band system for wider target optimization.

2

RVOL Filter Analysis

Evaluate minimum RVOL thresholds per symbol. Low-RVOL entries (especially on IBIT at 0.18x) consistently underperformed. Need to determine optimal RVOL cutoff to filter weak setups.

3

Param Surface Population

Populate the 3x4 param surface grid with the band trail data from this backtest. BandTrail_Wide is the early frontrunner -- now need to quantify its edge across different market regimes and symbols.

4

Trend Trade Classification (Band Slope Detection)

Add band slope detection to classify trending vs rotating days. The -$891 first day was likely a rotation day where pullback entries get whipsawed. Trend classification would allow the strategy to sit out rotation days.