Section ini ngejawab pertanyaan kunci: setup mana yang Anda profitable, dan account mana yang nge-drive P&L Anda?

Per-model breakdown

Tabel yang nge-list setiap model Anda dengan:

Trades, jumlah trade di-tag model ini
WR, win rate
P&L, total P&L
R, average R-multiple
PF, profit factor

Cara baca:

Sort by P&L descending, model terbaik di atas.
Sort by R-Multiple descending, model dengan edge per-trade tertinggi.

Realita vs ekspektasi

Ini yang paling insightful. Sering kali, model yang Anda kira terbaik ternyata tidak profit, dan model yang Anda anggap "hanya iseng" ternyata top performer.

Contoh realita:

Anda berpikir "Big Trade Acceptance" itu best Anda. WR 70%, perfect execution.
Tapi setelah 50 trade, "Stacked Imbalance Continuation" yang P&L-nya 2x lebih tinggi.

Kenapa? Bisa jadi:

Big Trade Acceptance muncul lebih jarang, jadi volume-nya rendah
Anda over-trade Stacked Imbalance karena Anda ngerasa confident
Stacked Imbalance punya edge yang lebih jelas di market condition tertentu

[!WARNING] Jangan langsung retire model yang underperform setelah 10-15 trade. Butuh minimal 30 trade, idealnya 50, sebelum signifikansi statistik.

Kapan retire model

Retire model kalau:

❌ Win rate < 40% setelah 30+ trade
❌ R-multiple < 0 (negative expectancy)
❌ P&L negatif setelah 50+ trade
❌ Anda udah 2x coba adjust (tweak entry, SL, TP), masih tidak profit
❌ Anda tidak lagi excited untuk trade model ini

[!TIP] Kalau model punya WR 50%+ tapi R-multiple negatif, biasanya masalah di risk management (SL terlalu jauh atau TP terlalu dekat). Tune ulang, jangan langsung retire.

Per-account breakdown

Tabel yang nge-list setiap akun dengan:

Trades, jumlah trade
WR, win rate akun
P&L, total P&L
Growth %, berapa persen akun Anda tumbuh dari initial balance

Cara baca:

Multi-account setup yang ideal

Anda punya 3 akun:

Forward Test, Stacked Imbalance: 50 trade, WR 60%, +$800 (validation)
Forward Test, Momentum Scalp: 30 trade, WR 45%, -$300 (tidak working)
Live Apex 50K: 80 trade, WR 55%, +$2,400 (proven edge)

Insight: Double down on Stacked Imbalance (proven di Forward Test, bisa di-push ke Live). Retire Momentum Scalp (tidak working). Keep Apex 50K (proven edge).

Live vs Forward Test comparison

Kalau Anda punya akun Live + Forward Test untuk setup yang sama, idealnya:

Forward Test WR ≥ Live WR (Live = harder karena real money + emotions)
Forward Test P&L ≤ Live P&L per trade (Live = fee + slippage)
Forward Test lebih banyak trade (Anda lagi nge-test agresif)

Kalau Live WR > Forward Test, Anda mungkin under-trading di Forward Test (skipping setup yang valid) atau over-trading di Live (tidak match plan). Re-evaluate.

Kalau Live P&L per trade < Forward Test, biasanya normal karena fee + slippage. Gap < 30% acceptable.

Performance score radar

Radar 6-dimensi:

Win Rate (skala 20-85%)
Profit Factor (skala 0.5-3.0)
Expectancy (skala -$300 to $500)
Sharpe (skala -2 to 3)
Consistency (1 - CV, skala 0-1)
Payoff (avgWin/avgLoss, skala 0.5-4.0)

Cara baca:

Luas area besar = strong overall trader.
Salah satu dimensi kecil = ada specific weakness. Misal Consistency 30% = profit Anda tidak repeatable.
Balance shape = ideal. Kalau ada satu spike tinggi dan others low = over-relying on satu metric.

[!INFO] Radar ini auto-computed dari stats Anda. Tidak perlu input manual. Kalau Anda punya lebih dari 50 trade, ini reliable.

Decision framework: what to do with each model

Setelah Anda lihat per-model breakdown, berikut decision tree:

Decision tree per model

STEP 1: Cek trade count (n)
├─ n < 20  → INSUFFICIENT DATA. Continue testing.
├─ 20-50  → EARLY SIGNAL. Pattern emerging, tapi butuh confirm.
└─ n > 50  → STATISTICALLY MEANINGFUL. Trust the numbers.

STEP 2: Cek profit (P&L)
├─ P&L > 0 DAN avg R > 0.5  → EDGE CONFIRMED. Continue.
├─ P&L > 0 tapi avg R < 0.5  → MARGINAL. Cek risk management.
├─ P&L < 0 DAN avg R < 0  → LOSING. Consider retire.
└─ P&L < 0 tapi WR > 50%  → EDGE ada, TAPI risk management issue. Cek avg win vs avg loss.

STEP 3: Cek consistency (variance of returns)
├─ Std dev / mean < 1.0  → Consistent. Good.
├─ Std dev / mean 1.0-2.0  → Moderate variance. Acceptable.
└─ Std dev / mean > 2.0  → High variance. Cek outliers. Mungkin ada big winner/loser inflating/deflating stats.

STEP 4: Cek correlation dengan models lain
├─ Correlation < 0.3  → Independent. Good diversification.
├─ Correlation 0.3-0.7  → Moderate. Bisa diversify atau overlap.
└─ Correlation > 0.7  → High. Model lo mungkin duplicate. Consider retire salah satu.

STEP 5: Action per quadrant

Quadrant analysis

Plot setiap model di 2D (Profit vs Variance):

	Low Variance	High Variance
High Profit	Star (keep, allocate more)	Speculative (keep but smaller size)
Low Profit	Grinder (keep, low priority)	Leaky (retire or fix)

Star = best. Keep, allocate more risk, learn from this setup.

Speculative = high win rate tapi high variance. Keep, but size down. Misal: 70% WR tapi variance tinggi = hot streak. Reduce size to manage variance.

Grinder = low profit, low variance. Steady but unspectacular. Keep as portfolio stabilizer, low allocation.

Leaky = low profit, high variance. Retire or fix. Likely cost: more than benefit.

Model lifecycle

Setiap model punya lifecycle. Understand where you are:

Phase 1: Discovery (0-10 trades)

"Apakah setup ini punya edge? Test."
Forward Test only, jangan live.
Track: win rate, avg R, frequency.

Phase 2: Validation (10-30 trades)

"Apakah edge real atau variance?"
Forward Test atau live dengan size 0.5%.
Decision criteria: 60%+ win rate DAN avg R > 0.5. If yes, promote to Phase 3.

Phase 3: Production (30-100 trades)

"Apakah profitable secara statistical?"
Live dengan size normal (1-2%).
Track: P&L, max drawdown, consistency.
Quarterly review: masih profitable? Trend up or down?

Phase 4: Maturity (100+ trades)

"Apakah masih punya edge di market saat ini?"
Live dengan size penuh.
Decision: terus, modify, atau retire.
Modify: market regime bisa berubah. Anda bisa retune entry, SL, atau TP tanpa changing fundamental setup.

Phase 5: Sunset (jika underperform)

"Apakah masih worth keeping?"
Criteria: 50+ trade terakhir menunjukkan decline → sunset.
Transition: move capital ke model yang masih productive.

[!TIP] Jangan attach emosional ke model. Model adalah tool. Kalau tool tidak berguna, ganti. "Big Trade Acceptance" punya nama bagus, tapi kalau losing 30 trade, retire. Ego vs edge.

Model correlation & portfolio construction

Anda bisa punya multiple models. Realitanya, mereka bisa correlated. Kalau semuanya correlated, Anda punya single point of failure.

Correlation matrix

Hitung daily returns correlation antar model:

              Stacked Imbalance  Big Trade Acceptance  Momentum Scalp
Big Trade         0.42              1.00
Momentum Scalp    0.31              0.27                  1.00

Artinya: Stacked Imbalance dan Big Trade Acceptance moderately correlated (0.42). Momentum Scalp relatively independent.

Portfolio recommendation

Max 3-5 models aktif simultaneously
Max 2 models yang correlated > 0.5 (otherwise redundant)
Mix time horizons: scalper (5s) + day (1h) + swing (3d) → natural diversification
Mix market conditions: trending + mean-reverting + breakout → robust to regime

Bad portfolio: 3 momentum models semua di 1 time frame, 1 market condition. P&L correlated.

❌ 50+ trade dengan R-multiple < 0: Strategy losing. Retire. ❌ Win rate < 40% setelah 30+ trade: Setup Anda tidak match market. Retire. ❌ Profit factor < 1.0 setelah 50+ trade: Total loss > total win. Retire. ❌ Trend of declining performance (last 30 trade << average): Market regime shift. Pause + re-tune atau retire. ❌ Anda tidak excited trade model ini: Trust your gut. Kalau Anda udah malas atau tidak percaya, retire.

Maybe-retire signals (re-evaluate, don't auto-retire)

⚠️ Profit factor < 1.2: Marginal. Tune dulu (SL, TP, atau entry criteria) sebelum retire. ⚠️ 20-30 trade dengan negative P&L: Maybe variance. Continue 20 trade lagi. ⚠️ High variance (std dev > 2× mean): Maybe outlier inflating. Tune.

Don't retire signals

✅ Single losing trade > 2R: Just variance. Continue. ✅ Drawdown 20%: Normal. Continue. ✅ 5 consecutive losses: Just a streak. Continue (with rule: stop trading untuk 1 day).

Kapan modify, bukan retire

Sometimes model underperform bukan karena setup salah, tapi karena tuning outdated.

Re-tune checklist

Kalau model underperform, tanyakan:

SL placement: Terlalu tight (banyak wick)? Terlalu wide (loss besar)?
TP target: Realistis dengan current market? Misal: 5R target di 2022 high-vol vs 2R target di 2024 low-vol.
Entry criteria: Setup quality turun? Banyak entry borderline?
Frequency: Setup terlalu sering (overtrading) atau terlalu jarang (missing real setups)?

Misal: Big Trade Acceptance originally worked di 2024 dengan target 2R dan 18-point SL. 2026, market regime berbeda. Volatilitas turun, target 2R jadi kurang realistic. Tune ke 1.5R target, 12-point SL. Same setup, different parameters.

Don't change setup definition. Tune parameters only. Setup definition = what makes it a valid setup. Parameters = how to execute.

Model portfolio size allocation

Setelah Anda punya multiple profitable models, alokasi modal Anda:

Model quality	Allocation	Risk per trade
Star (high profit, low variance)	50% of capital	1.5-2%
Speculative (high profit, high variance)	20% of capital	0.5-1%
Grinder (low profit, low variance)	20% of capital	1%
New test (Forward Test)	10% of capital	0.5%

Ini guideline. Adjust per personality. Anda yang paling kenal risk appetite Anda.

[!INFO] Modal allocation = modal yang siap di-deploy (bukan total net worth). Kalau Anda punya $100K tapi hanya aktif trade $25K, alokasi 100% = $25K.

Per-account breakdown deeper

Tab Analytics → section "Your trading systems" → "Account performance" table.

Reading the breakdown

Sort by Growth % descending → best performing account di atas.

Inter-account comparison:

Live best performer: account dengan growth % tertinggi. Allocate more capital ke sini (kalau multiple Live accounts).
Forward Test winners: model yang profitable di FT tapi belum di Live. Strong signal to promote to Live.
Underperformers: FT atau Live dengan P&L negatif. Identify why, wrong setup? Wrong sizing? Wrong market?

Account-level decisions

Kalau 1 Live account underperform 30+ trade: cek apakah setup, sizing, atau market. Consider reduce size atau pause.
Kalau 1 FT account outperform Live 30+ trade: signal untuk bring ke Live. Kalau Live underperform setelah migration, ada execution gap (psychological, fee, slippage).
Kalau semua accounts losing: macro issue. Market regime, strategy fundamental, atau sizing. Re-evaluate.

Tactical exercises

Exercise 1: Model portfolio audit

List all your models
Categorize: Star / Speculative / Grinder / Leaky
Identify top 2 performers (Star + Speculative)
Identify bottom 1 (Leaky if applicable)
Decide: keep, modify, retire

Exercise 2: Correlation map

Get daily returns per model
Calculate correlation matrix (Excel =CORREL())
Identify pairs dengan correlation > 0.5
Decide: keep both (diversification benefit) atau retire satu (redundancy)

Exercise 3: Setup quality stratification

Group trades by setup quality bucket (1-3, 4-5, 6-7, 8, 9-10)
Calculate P&L per bucket
Find the "cutoff", quality level di mana P&L flips from positive to negative
Set rule: "No entry below this quality level"

Real-world model evolution

Case study: 3 model portfolio

Month 1-3: Discovery

Stacked Imbalance Continuation: 22 trade, WR 64%, +$1,800 → looks good
Momentum Scalp 5m: 18 trade, WR 45%, -$200 → marginal
Big Trade Acceptance: 8 trade, WR 50%, +$100 → too early

Decision: Continue all. SI strong but small sample. MS borderline. BTA needs more.

Month 4-6: Validation

SI: 60 trade total, WR 60%, +$5,400 → STATISTICALLY MEANINGFUL, promote to Live
MS: 50 trade total, WR 44%, -$300 → STILL MARGINAL. Tune: lower size to 0.5%.
BTA: 35 trade total, WR 51%, +$800 → OK but not great. Continue.

Decision: SI = star. MS = grind. BTA = moderate. Promote SI to Live with full size. MS keep with reduced size. BTA continue.

Month 7-12: Production

SI: 120 trade total, WR 58%, +$11,200 → solid edge. Increase allocation.
MS: 95 trade total, WR 42%, -$500 → borderline. Last 30 trade still negative. RETIRE after 100 trade.
BTA: 80 trade total, WR 50%, +$1,200 → grinder. Stable. Keep.

Decision: SI = primary. MS retired. BTA = secondary stabilizer.

Year 1 result: 2 models active, focused capital, sustainable returns. Better than 5 models with diluted capital.

Common anti-patterns

❌ More models = more diversification = false if models correlated. Quality > quantity.

❌ Tune terlalu sering = over-fitting. If 50 trade still profitable, jangan tweak.

❌ Switch to model baru pas losing = chasing. Stick to proven setups.

❌ Compare dengan orang lain = envy. Your edge ≠ their edge. Track your own progress.

❌ Retire model setelah 1 losing trade = terlalu sensitive. Variance is real.

FAQ

Q: Kapan harus punya model baru vs tune yang ada?

Tune existing kalau:

Setup definition masih match market
Underperformance baru (was profitable, now losing)
Specific parameter (SL, TP) perlu adjustment

Buat model baru kalau:

Anda identify a genuinely different setup (different trigger, different entry)
Existing models tidak cover opportunity yang Anda lihat
Anda pengen diversify ke different market condition

Q: Berapa model yang ideal?

3-5 aktif. 1-2 star + 1-2 grinder + maybe 1 speculative. Lebih dari 5 = diluted focus. Less dari 2 = concentrated risk.

Q: Kapan retire model, berapa trade minimum?

Minimum 30 trade. 50+ better. 100+ ideal. Less dari 30, you're trading variance, not edge.

Tapi kalau:

30 trade dengan P&L < 0 DAN R-multiple < 0 → retire
50 trade dengan P&L < 0 → definitely retire
30+ trade dengan declining performance (last 30 << average) → consider modify atau retire

Q: Saya punya model yang profitable tapi boring. Keep?

Yes! Boring profitable > exciting losing. Grinder models adalah portfolio stabilizers. Keep them even kalau Anda prefer trading stars.

Q: Model profitable di FT tapi losing di Live. Why?

Common causes:

Execution gap: FT bisa cut loss instantly, Live butuh waktu → wider actual loss
Fee + slippage: FT $0, Live $2-5 per round trip → reduce effective R
Psychology: Live trading lebih stressful → emotional decisions override plan
Different market conditions: FT di backtest period A, Live di period B

Debugging: compare FT and Live trade-by-trade. Cek R-multiple distribution, avg trade, dan exit reason. Identifikasi gap.

Q: Bagaimana cara test model baru tanpa lose money?

Forward Test:

Create new account (type: Forward Test)
Trade setup dengan normal size calculation
Track for 30+ trade
Decision: profitable → promote to Live. Losing → modify atau retire.

100% safe. Anda learn tanpa risk.

Q: Saya punya model yang profitable di 1 time frame. Apakah profitable di semua time frame?

Probably not. Time frame matters karena:

Volatility different (1m vs daily)
Liquidity different (RTH vs ETH)
Speed different (1m needs fast execution, daily doesn't)
Setup frequency different (more setups in 1m)

Test per time frame. Jangan asumsikan.

Q: Saya punya model yang profitable di ES. Apakah profitable di NQ?

Probably not directly. Similar tapi beda. Test per instrument. Setup Anda mungkin match ES microstructure tapi tidak NQ.