TARGET 20 v2 · PROFIT LAB

Earn the right
to trade live.

A 20% objective, evaluated with unseen data, realistic costs, adaptive exposure, and automatic graduation gates.

CURRENT DECISIONRESEARCH ONLYAutomated model promotion is blocked
Walk-forward return+27.0%All unseen windows compounded
Robust quarters5/860% quarterly threshold passed
Worst unseen window-8.4%30-day test segment
Maximum exposure1.00×Volatility scaled · spot only

EVIDENCE GATES

Six checks before promotion

!

3-year point-in-time data1095 days verified; historical universe membership still required

LOCKED
✓

Walk-forward robustness5/8 quarterly windows positive

PASS
✓

Kraken costs modeled0.26% fee + 0.10% slippage

PASS
✓

Spot-only exposureMaximum 1.0×; no shorting

PASS
!

Forward paper evidence0/60 distinct days recorded

LOCKED
!

Automated model promotionRequires 60 days, higher net return, and drawdown no worse than −15%

LOCKED

ADAPTIVE ENGINE

How v2 seeks better returns

01

dual momentum

02

50-day trend confirmation

03

market breadth regime filter

04

45% annualized volatility target

05

cash fallback

Frozen head-to-head · 0/60 days

Target20 v2 and Diversified v1 are locked. Each check captures completed Kraken daily closes, both portfolios’ weights, costs, equity, and drawdown.

Target20+0.0%Diversified+0.0%LeaderTied
Promotion requires 60 distinct days, a higher net return, and no worse than −15% drawdown.

CAPITAL GRADUATION

Scale evidence, not confidence.

STAGE 0Research$0 live
STAGE 1Pilot$25–$100 orders
STAGE 2Proven25% capital ceiling
STAGE 3MatureRisk-budgeted capital

ANNUAL PROFIT CHALLENGER · PAPER ONLY

Annual return first. Still fragile.

Each quarter selected from 1,728 cost-aware configurations using a rolling one-year training window. It failed out of sample.

Compounded return-10.1%Six unseen quarters
Annualized return-6.9%20% objective missed
Positive quarters2/633.3% hit rate
Sequence drawdown-30.1%No loss clipping · costs included
REJECTED

DAILY 1 CHALLENGER · CORRECTED

Optimistic result corrected.

The earlier test clipped losses at −1% using daily bars, which could not prove those fills. With full close-to-close losses restored, the holdout result is negative.

Holdout return-16.6%Close-to-close · costs included
Average daily-0.0%1.00% target missed
1% days1.4%5 of 364 days
Maximum drawdown-9.5%5% limit breached
REJECTED

NEW LEADING CHALLENGER · REUSED HISTORY

Diversification beat the matched benchmark.

Volatility-weighted positions across the top two to four eligible assets improved compounded return on shared windows. It still lacks enough positive quarters for promotion.

Matched challenger+55.8%22 shared monthly windows
Matched Target20+50.6%Apples-to-apples dates
Positive quarters3/760% robustness gate missed
Worst month-7.2%Below 10% loss limit
UNVALIDATED

INTRADAY + SENTIMENT RESEARCH

More ideas. Better rejection.

13,440 verified hourly observations per asset were tested in rolling unseen windows. Lagged sentiment was also added to Target20; none improved the leader after costs.

Hourly breakout-20.7%1/6 windows positive
Hourly momentum-43.8%53 costed trades
Mean reversion-26.6%2/6 windows positive
Target20 + sentiment+5.1%8 folds selected the filter
REJECTED

Target20 remains the promoted baseline at +27.0% on its full evaluation. The diversified challenger won the matched-window return comparison, but has not passed robustness or fresh-data validation. Further tuning on the same history would increase overfitting risk, not independent evidence. Automated promotion remains locked until point-in-time data and the forward-paper gate pass. User-authorized manual Kraken execution is a separate control.