Panduan Portfolio Manager v3

Tata kelola portofolio top-down ala Goldman Sachs / JP Morgan, di-translate ke konteks IDX dengan engine HPS. v3 update: Konsolidasi recommendation, Keep Score ranking, dan workflow 5-layer.

1Filosofi: Top-Down PM

Bottom-up stock picking saja tidak cukup. PM institusi memulai dari pertanyaan jauh lebih atas:

Berapa saham yang ideal untuk portfolio ini? (sizing)
Berapa risiko yang boleh saya ambil? (risk budget)
Sektor mana yang punya tailwind? (sector allocation)
Saham mana di sektor itu yang pantas? (stock selection)
Berapa porsi per saham? (position sizing)

Lima pertanyaan ini adalah 5 layer top-down. Setiap layer menentukan layer berikutnya. Tidak boleh skip.

💡 Brinson & Bagger Insights

Brinson (1986, 1991): ~90% variasi return portofolio dijelaskan oleh asset allocation (regime + sektor), bukan stock picking. Statman (1987): diversifikasi benefit plateau di 12-15 saham. Beyond 20 = diworsification (Lynch).

↓

SIZING → REGIME → SECTOR → STOCK → SIZE → TRADE

−1Layer −1 — Margin Check

Pertanyaan: "Apakah portofolio masih punya buying power yang aman?"

Layer paling atas — gating sebelum analisa apa pun. Kalau margin sudah stretched, semua analisa BUY/top-up jadi tidak relevan; yang penting kurangi exposure dulu.

Formula Margin Check

Inputs Equity = portfolio NAV (atau override manual) Leverage factor = 1× (cash account) / 2× (marginable) Buying Power = Equity × Leverage HPS haircut per ticker = dari Haircut Engine (regulator-aligned)
Margin Used Margin Used = Σ (position_value × HPS_haircut%)
Util Ratio & Status Util = Margin Used / Buying Power < 60% → HEALTHY Buying power ample, BUY allowed 60–80% → WATCH Margin filling up, monitor 80–95% → STRETCHED Block all new BUY > 95% → AT_RISK Force TRIM verdict

⚠ STRETCHED & AT_RISK gate

Saat status STRETCHED atau AT_RISK, Layer 4 Trade Plan otomatis suspend semua BUY (top-up KEEPERS dan NEW IDEAS). Hanya SELL/TRIM yang muncul. Cara recover: liquidate posisi high-haircut dulu (Tier 3-4 stocks), atau tambahkan equity buffer di setting Layer −1.

Setting yang bisa di-adjust

Leverage: 1× cash account (default, paling konservatif), 2× marginable (kalau client sudah register margin facility)
Equity override: kalau client punya cash buffer di luar portfolio yg di-upload (misal cash di RDN belum dimasukkan ke XLSX), bisa input manual. Default 0 = pakai total market value sebagai equity proxy.

💡 Kenapa HPS bukan KBVS sebagai input?

HPS haircut sudah account untuk regime (Crash Radar) + tier composite + adaptive Level 3 (stock stress). Saat market stress, HPS auto-naik → margin used auto-naik → util ratio naik → otomatis di-block BUY. Pakai KBVS quarterly statis akan miss perubahan kondisi market intra-quarter.

0Layer 0 — Portfolio Sizing & Konsolidasi

Pertanyaan: "Berapa saham yang ideal untuk portfolio ini?"

Layer ini adalah filter pertama sebelum analisa lain. Banyak retail investor punya 30-50 saham — over-diversified. Default HPQuant pakai Concentrated Philosophy (5-12 nama tergantung NAV) ala Buffett/Fisher/Munger, bukan textbook diversification (12-25).

💡 Filosofi: Concentrated vs Diversified

Diversified (Statman 1987, Evans-Archer 1968): basis statistical variance reduction, output 12-25 saham. Cocok kalau pengelolaan AUM besar dengan tim research banyak.

Concentrated (Fisher 1958, Buffett 1996, Munger 1995, Greenblatt, Pabrai, Klarman, Ackman): basis manageable conviction. Output 5-12 saham. Filosofi: "Wide diversification is only required when investors do not understand what they are doing" (Buffett). Cocok untuk PM/RM yang handle multiple accounts — operational reality > textbook.

HPQuant default = Concentrated karena (a) PM HP Sekuritas typically handle 20-50 accounts paralel, monitoring 15-20 saham per account = unmanageable; (b) IDX cuma punya ~50-80 saham dengan likuiditas decent (LQ45 + IDX80 fringe), sehingga spread terlalu lebar = forced to pick mid-quality.

Sizing Buckets (Concentrated, default)

Bucket	NAV Range	Ideal # Saham	Rationale
Mikro	< Rp 100 jt	2 – 4	Concentrated bets, 1-2 lot per posisi (Rp 5-15 jt min meaningful)
Kecil	Rp 100 – 500 jt	3 – 5	High conviction only, learn before scale
Menengah	Rp 500 jt – 2 B	5 – 7	Sweet spot retail (Buffett-Fisher), Rp 70-300 jt per posisi
Besar	Rp 2 – 10 B	6 – 8	HNW range — manageable monitoring, deep research per name
Premium	Rp 10 – 50 B	8 – 12	Multi-sektor exposure tapi tetap focused
Institusional	> Rp 50 B	12 – 18	Reg + liquidity constraints, beyond = diworsification (Lynch)

Min Position Size Rule

NAV	Min Position % NAV	Approx Min IDR
< Rp 100 jt	≥ 10%	≥ Rp 10 jt
Rp 100 – 500 jt	≥ 6%	≥ Rp 6-30 jt
Rp 500 jt – 2 B	≥ 4%	≥ Rp 20-80 jt
Rp 2 – 10 B	≥ 3%	≥ Rp 60-300 jt
> Rp 10 B	≥ 2%	≥ Rp 200 jt+

Posisi di bawah threshold = noise → kandidat EXIT. Tidak meaningful untuk move portfolio + memboroskan mental bandwidth RM.

Sumber referensi

Phil Fisher — Common Stocks and Uncommon Profits (1958): "Beyond 10-12 stocks, an individual investor cannot adequately follow them."
Warren Buffett — Berkshire 1996 letter: "Wide diversification is only required when investors do not understand what they are doing."
Charlie Munger — USC Law speech (1995): "Punch card investor" — bayangkan kamu hanya boleh punya 20 keputusan invest seumur hidup.
Mohnish Pabrai — The Dhandho Investor: "Few bets, big bets, infrequent bets."
Mauboussin (2018) — Looking for Easy Games: empirical evidence concentrated active managers (10-20 positions) outperform diversified peers in active-share-adjusted returns.
Seth Klarman (Baupost), Bill Ackman (Pershing Square): biasanya run 8-15 names.
Statman (1987) — counterpoint untuk filosofi Diversified, "12-15 plateau" untuk variance reduction.

Konsolidasi Recommendation Box

Di paling atas halaman, navy gradient box menampilkan headline:

🎯 Headline Format

    36 → 12 saham (drop 24 noise positions)

    SELL PROCEEDS Rp 1.31 B  ·  KEEP VALUE Rp 0.86 B  ·  QUALITY SCORE 53 → 84

KEY METRICS — Pre vs Post Konsolidasi

Metric	Tujuan	Improvement Goal
# Holdings	Jumlah posisi	↓ ke ideal size
HHI Concentration	Herfindahl Index Σw²×10000	↑ (lebih fokus, target 800-1500)
Top-5 Weight	% bobot top 5 posisi	~50-60%
Avg Keep Score	Quality rata-rata	↑ (target ≥ 75)
Wt-avg P/L%	Performance gabungan	—
Min Position Wt	Bobot terkecil	≥ 5% (no noise)

△ Tipikal: Wt-avg P/L% Post bisa lebih buruk

Konsolidasi memilih saham dengan future conviction (signal kuat, sektor OW), bukan retrospective gain. Banyak winners (gain tinggi) yang signal-nya lemah → di-keep tapi tidak ditambah; banyak losers (drawdown) yang signal Buy → di-add karena potential reversal. Jadi P/L% post bisa lebih rendah — bukan masalah, ini forward-looking.

2Keep Score — Composite Priority Ranking

Setiap saham di-skor 0-100. Higher score = higher conviction = priority untuk di-keep.

Formula Keep Score

Base 50 (neutral)
Winner Protection +30 Bagger 🚀 (P/L ≥ +100%) +15 Winner 🏆 (P/L +50% to +100%)
HPQuant Signal +25 Buy +15 Spec Buy 0 Hold -25 Sell -8 No signal data
Tier (HPS Haircut Engine) +15 Premier (Tier 0) +10 Strong (Tier 1) +5 Sound (Tier 2) -10 Heightened (Tier 3) -20 Very High (Tier 4)
Sektor Stance (Layer 2) +15 OW (overweight) 0 Neutral -10 UW (underweight)
Liquidity Tier (ADTV-based) — NEW Phase 1 +8 L1 Premier (ADTV ≥ Rp 50 B/day) +4 L2 Strong (Rp 10 – 50 B/day) 0 L3 Sound (Rp 3 – 10 B/day) -8 L4 Moderate (Rp 1 – 3 B/day) -15 L5 Tight (< Rp 1 B/day)
Haircut Tier (HPS-based) — NEW Phase 2 +8 HPS ≤ 15% (Premier — safe collateral) +5 HPS 15–25% +2 HPS 25–35% -5 HPS 35–50% -12 HPS > 50% (Very High — regulator flagged)
Index member bonus +3 LQ45 0 IDX80 -3 Non-Index
Trend (HPQuant) +5 Trend up (tr=+1) -5 Trend down (tr=-1)
Noise Penalty -8 Position < 1% AND not winner/bagger

Cara baca Keep Score di halaman

Rank circle warna: 🟢 KS ≥ 70 (high conviction) · 🟡 50-69 (moderate) · 🔴 < 50 (low)
Top-N (idealN) keepers dengan skor tertinggi → MASUK list view
Sisanya (drops) → ditampilkan compact chips di bawah, di-realokasi via Trade Plan

💡 Mengapa multi-factor?

Single-factor scoring (cuma signal, atau cuma P/L) bisa misleading. PTRO +135% gain, signal Sell — apakah jual? Multi-factor: bagger bonus (+30) + Sell signal (-25) + tier Sound (+5) + LQ-bonus (+3) = KS ~83, tetap KEEP karena bagger protection dominan. Tanpa multi-factor, akan kelewatan winner besar.

2.5Liquidity Tier — ADTV Gating

Pertanyaan: "Kalau saya beli saham ini sekarang dan tiba-tiba market crash, berapa lama saya butuh untuk exit tanpa move price?"

Sebelum Phase 1, Keep Score & trade plan tidak melihat likuiditas — bisa rekomendasi top-up ke saham yg ADTV-nya cuma Rp 1-2 B/day, dan saat regime berubah, posisi tidak bisa di-exit cepat. Phase 1 fix ini dengan 5-tier liquidity ranking.

Tier definition (berdasarkan ADTV 20 hari)

Tier	ADTV (Rp/hari)	KS adjustment	Behaviour
L1 Premier	≥ 50 B	+8	Big4 banks, TLKM, BBRI — sangat liquid, bisa exit besar
L2 Strong	10 – 50 B	+4	Mid-cap LQ45 — manageable
L3 Sound	3 – 10 B	0	IDX80 fringe — neutral
L4 Moderate	1 – 3 B	−8	BLOCK ADD — top-up tidak diizinkan
L5 Tight	< 1 B	−15	BLOCK ADD — kandidat EXIT

Days-to-Exit (position-aware)

Selain tier static, ada gate dinamis: kalau target weight × NAV terlalu besar relatif ADTV, target di-cap supaya days-to-exit ≤ 7 hari (asumsi 25% partisipasi volume harian).

days_to_exit = position_value / (ADTV × 0.25) max_position_safe = ADTV × 0.25 × 7 // 7-day exit safety target_pct_capped = min(target_pct, max_position_safe / NAV × 100)

⚠ Kasus ACST (real example)

ACST: ADTV ~Rp 2 B/day, target weight 8% pada NAV Rp 100 B = posisi Rp 8 B. Days-to-exit = 8B / (2B × 0.25) = 16 hari. Saat regime berubah dari Aman ke Krisis, akan terjebak. Phase 1 fix: ADTV 2 B → L4 → KS −8, ADD di-block, target di-cap ke 1.75% (= 2B × 0.25 × 7 / 100B).

NEW IDEAS filter

Selain blok ADD untuk holding existing, NEW IDEAS dari sektor OW juga di-filter: saham L4-L5 tidak akan muncul sebagai rekomendasi BUY. Liquidity adalah hard gate, bukan soft penalty.

3Layer 1 — Regime Briefing

Pertanyaan: "Berapa exposure equity yang seharusnya?"

Pakai HPS Crash Radar sebagai compass. Score 0-100 dari 6 macro indicators (volatilitas, drawdown, USD/IDR, yield SUN 10Y, gold, gold/brent).

RISK ON

95–110% equity

CR < 25 · leverage diizinkan · growth + cyclicals

VOLATILITY

75–85% equity

CR 25–45 · cash buffer 15-25% · balanced

BEARISH

50–70% equity

CR 45–65 · defensive only · gold hedge

PANIC

30–50% equity

CR ≥ 65 · cash-heavy + Big4 + gold

Cara baca verdict

OVER-DEPLOYED: equity di atas target band → trim ke cash
ON TARGET: maintain
UNDER-DEPLOYED: deploy sesuai sektor playbook

4Layer 2 — Sector Playbook

Pertanyaan: "Sektor mana yang outperform di regime ini?"

Kelompokkan saham IDX ke 10 sektor IDX-IC:

Sektor	Contoh	Karakter
Energy	ADRO, ITMG, MEDC, PTRO	Cyclical, oil/coal beta
Basic Materials	ANTM, INCO, MDKA, BRMS	Commodity-linked
Industrials	UNTR, ASII, ADHI	Cyclical, capex-driven
Cons. Cyclicals	ACES, MAPI, ERAA	Discretionary spending
Cons. Non-Cyclicals	ICBP, INDF, UNVR	Defensive, staples
Healthcare	KLBF, KAEF, HEAL	Defensive, growth
Financials	BBCA, BBRI, BMRI	Rate-sensitive blue-chip
Properties	BSDE, CTRA, PWON	Rate-sensitive cyclical
Technology	GOTO, BUKA, EMTK	Growth, high beta
Infrastructure	TLKM, JSMR, PGAS	Defensive, dividend

Sector Score & Stance

score = avg(stk_score) × 50% + (% Buy signal) × 30% + (% trend up) × 20%

OW (Overweight): score ≥ 55 (60 di Bearish/Panic) → tambah exposure
Neutral: 40 ≤ score < 55
UW (Underweight): score ≤ 40 → kurangi

Target Weight per Sektor

Distribusikan target equity ke 10 sektor:

OW sektor: 1.5× equal-weight
Neutral: 1.0×
UW: 0.3×

Cara baca CUR vs TGT bars

📊 Visual interpretation

CUR bar (solid color) = % bobot anda saat ini di sektor itu.
TGT bar (diagonal stripes) = % rekomendasi berdasar regime.
Delta > +5% = perlu add exposure ke sektor ini.
Delta < -5% = perlu reduce exposure.

5Layer 3 — Position Review (4-Quadrant + Top-N)

Pertanyaan: "Saham mana yang dipertahankan dan apa actionnya?"

Layer ini punya 3 view yang saling melengkapi:

5.1 — 4-Quadrant Matrix

Klasifikasi konseptual setiap holding di matrix Signal × Tier:

✓ ADD — Beli/Top-up Buy/Spec Buy + Tier ≤ Sound. High conviction: signal positif + risiko terkontrol.

Aksi: BUY shares ke target weight

= HOLD — Pertahankan Buy + Tier ≥ Heightened, atau winner/bagger protected.

Aksi: maintain position

△ TRIM — Reduce 50% Hold/Sell + Tier ≤ Sound. Quality tapi signal lemah.

Aksi: SELL 50% shares, rotasi

× EXIT — Jual Hold/Sell + Tier ≥ Heightened. Signal lemah + risiko tinggi.

Aksi: SELL all shares

5.2 — Top-N Keepers (List View)

Hanya top N saham (dari Keep Score ranking) yang muncul sebagai detail rows. Setiap row punya:

PTRO 🚀 IDX80

Energy · Buy · Tier Sound

9.4%

WEIGHT

+135%

P/L

ADD

Rank circle: warna by KS (green ≥70, yellow 50-69, red <50)
Ticker: + 🏆/🚀 badge + LQ45/IDX80 pill
Subline: Sektor · Signal · Tier (one liner context)
Weight + P/L: prominent numbers
Action pill: ADD/HOLD/TRIM/EXIT (color coded)

5.3 — Dropped Positions (Compact Chips)

Saham di luar Top-N ditampilkan compact chips dengan KS + P/L:

  • BREN KS 65 -19%   • BUMI KS 65 -32%   • BBCA KS 25 -19%   • KEEN KS 5 +38% · 🏆 AMMS KS 47 +67% (winner protected)

6Winner Protection Rule

Filosofi: Let your winners run.

Status	Kondisi	Override Rule
🏆 WINNER	P/L +50% to +99%	Tidak masuk EXIT bucket. Hanya proportional trim kalau bobot >10% NAV.
🚀 BAGGER	P/L ≥ +100%	Tidak masuk EXIT/TRIM. Hanya minimal trim kalau bobot >15% NAV.

💡 Why protect winners?

Stocks yang sudah +100% biasanya fundamental + momentum kuat. Jual karena "signal Sell" sering salah — saham mungkin koreksi sebentar lalu lanjut. Lebih bijak: trim sebagian kalau overweight (concentration risk), tapi biarkan core position. Filosofi Peter Lynch: "the trick is not to learn to trust your gut feelings, but rather to discipline yourself to ignore them."

Praktis

Bagger overweight (>15%): trim ke 12% (let it ride)
Winner overweight (>10%): trim ke 10% (cap concentration)
Winner moderate (≤10%): full hold, no action
Loser dengan Sell signal: full exit (zero toleransi)

△ Winner di drop list?

Bisa terjadi: misal AMMS +67% 🏆 dengan no signal data dan tier high → KS rendah → drop list. Tapi karena winner protected, dia TIDAK dijual penuh. Hanya ditrim kalau overweight, atau kalau weight sudah ≤10% → fully held meskipun rank rendah.

7Layer 4 — Trade Plan

Pertanyaan: "Apa yang harus di-eksekusi?"

Trade plan tersequence dalam 3 step, driven by Keep Score top-N:

Step 1 — SELL (free up capital)

Non-keepers (rank > idealN) yang BUKAN winner → full sell
Winner overweight (>10% wt) → proportional trim ke 10%
Bagger severely overweight (>15% wt) → minimal trim ke 12%
Bagger ≤15% wt → full hold (zero sell)

Step 2 — BUY (top-up keepers)

Keepers di-top-up ke equal weight target = min(15%, 100/N)
Untuk N=12 → target 8.3% per nama
Untuk N=10 → target 10% per nama
Hard cap 15% per single name (concentration limit)
Min 5% per nama (no noise positions)

Step 3 — NEW IDEAS

Saham yang belum ada di portfolio
Filter: Buy/Spec Buy signal + sektor OW + Tier ≤ Sound
Top 10 berdasar HPQuant technical score
Berfungsi sebagai "shopping list" — pilih beberapa untuk diversifikasi

Position Sizing Constraints

Tier	+ LQ45	+ IDX80	Non-Index
Premier (0)	12%	10%	8%
Strong (1)	8%	7%	6%
Sound (2)	6%	6%	5%
Heightened (3)	3% (only if Buy/Spec Buy)
Very High (4)	2% max (only if Buy)

Plus modifier:

Buy: 1.0× · Spec Buy: 0.75× · Hold: 0.6× · Sell: 0%
Risk On: 1.1× · Bearish/Panic: 0.7×
Hard cap: 15% per nama

7.5Reconstruction Mode — Phased Transition

Pertanyaan: "Bagaimana kalau klien punya portofolio existing 25-30 saham dan banyak yang lagi rugi — masa langsung di-cutloss massal?"

Trade Plan biasa cocok untuk akun baru (clean slate). Untuk akun existing dengan banyak laggard, forced cutloss day-1 = klien resisten + RM kena pukul. Reconstruction Mode menyelesaikan ini dengan phased transition yang menghormati psikologi loss aversion.

Cara aktif

Toggle ⏵ RECON ON di header Portfolio Manager (sebelah kanan info nasabah). State persist di localStorage — setiap account bisa beda mode. Saat ON, Layer 4 di-replace dengan multi-phase Reconstruction Plan.

Categorization (otomatis per posisi)

Kategori	Kriteria (any of)	Aksi
🛡 PROTECTED	Bagger / Winner	Hold (winner protection rules)
× BROKEN_THESIS	KS < 35 · OR Sell signal · OR (Tier 4 + sektor UW) · OR (P/L <−40% + Sell) · OR (P/L <−30% + KS < 40)	Cut phased (lihat tranche rules)
△ WAIT_TRIM	KS 35-55 · P/L −5% s/d −40% · no Sell signal	Trim 40% sekarang + 60% trigger di +10% rebound
= HOLD_WATCH	KS 55-70 · monitor	Hold, no action
✓ KEEPER	KS ≥ 70	Hold + receive new money allocation

Broken Thesis tranching (soften by liquidity & size)

Kondisi posisi	Tranche	Span	Why
L4 / L5 (illiquid)	4 tranches · 25%/week	Week 1, 2, 3, 4	Hindari move price karena ADTV rendah
> 7% NAV (large)	3 tranches · 33% each	Week 1, 3, 5	Soften shock book + psikologis
3-7% NAV (medium)	2 tranches · 50/50	Week 1, 3	Split 2 untuk soften psikologis
< 3% + liquid	1 tranche · 100%	Week 1	Posisi kecil + liquid, boleh full sell

Filosofi: "Even broken thesis tidak harus di-dump dalam satu hari". Klien yg lihat 5 saham keluar portfolio sekaligus akan trauma; kalau spread 4-6 minggu dengan rationale jelas per saham, lebih palatable.

4-Phase Plan

PHASE 1 — Week 1-2 (Cut Broken Thesis tranche 1) - 50%-100% dari posisi broken thesis di-sell - Output: capital freed dari "hopeless" positions - Klien message: "ini upgrade quality, bukan loss locking" PHASE 2 — Week 3-6 (Continue cuts + Wait&Trim 40%) - Sisa tranche broken thesis selesai - Wait & Trim positions: trim 40% dari setiap posisi sekarang - Sisa 60% → biarkan untuk recovery PHASE 3 — Month 2-4 (Trigger-based exit) - Set price alert per Wait & Trim sisa: jual saat +10% rebound dari current - Kalau rebound tidak datang dalam 4 bulan, force-exit (don't hold forever) - Klien rasa "diberi kesempatan recovery, bukan dipaksa loss" PHASE 4 — Ongoing (New Money Rule) - Setiap setoran baru / dividend / sell proceeds → 100% ke top KS keepers - Bypass laggards entirely - Over 12-24 bulan, weight % laggards naturally turun karena denominator naik - Klien tidak perlu rasa "cutloss" sama sekali untuk transisi terakhir

Frameworks & sumber psikologis

Kahneman & Tversky (1979) Prospect Theory — loss aversion 2× kekuatan gain. Realisasi loss psikologis terasa 2× lebih sakit daripada gain equivalent.
Shefrin & Statman (1985) Disposition Effect — investor cenderung hold loser (refusal to realize loss) sambil cepat sell winner. Anti-pattern yg perlu di-counter.
Odean (1998) Are Investors Reluctant to Realize Their Losses? — empirical: retail investors hold loser 30-50% lebih lama daripada winner di periode ekuivalen.
Ross Stevens (Stone Ridge, ex-Goldman) — drip method untuk wealthy account transitions. Concept: "transition over months, not days."
Mauboussin (2014) Reflections on the Ten Attributes of Great Investors — concentration + conviction triage = differentiator antara great vs average PM.

Reframing dialogue ke klien

💡 Anchor reset technique

Bukan: "Pak, saya saran cutloss ABCD, sekarang rugi −35%."

Tapi: "Pak, kalau hari ini portfolio kita Rp X cash, akan beli ABCD lagi tidak? Kalau tidak, mari kita rotasi ke saham yang Bapak akan beli sekarang."

Reframing mengubah anchor dari entry price (sunk cost) ke opportunity cost. Banyak nasabah yg tadinya menolak cutloss jadi setuju setelah disadari bahwa mereka sebenarnya tidak mau beli saham itu lagi di harga sekarang.

💡 Pair trading framing (preserve narrative)

Bukan: "Jual ABCD lalu beli BBRI" (terasa seperti dua aksi terpisah, satu di antaranya cutloss).

Tapi: "Rotasi sektor financial dari ABCD ke BBRI" (terasa seperti satu aksi: upgrade quality dalam sektor yang sama, bukan loss).

Klien lebih reseptif ke "sector rotation" daripada "exit position."

Diagnostic table di UI

Saat Reconstruction Mode aktif, di atas phase blocks akan muncul kartu diagnostic — 5 metric cards (count per kategori) plus tabel detail per posisi dengan tag visual:

🔬 Contoh output

× ABCD  BROKEN THESIS  KS 22 · −45% · 4.2% wt

× XYZW  BROKEN THESIS  KS 18 · −38% · 2.8% wt

△ POSC  WAIT TRIM      KS 48 · −25% · 6.3% wt

△ ASII  WAIT TRIM      KS 52 · −12% · 5.1% wt

= MIKA  HOLD WATCH     KS 62 · −8%  · 7.2% wt

🛡 GOTO  PROTECTED      KS 88 · +135% · 22.3% wt

✓ BBCA  KEEPER         KS 82 · +12% · 8.5% wt

Tuning thresholds (kalau perlu adjust)

Threshold default tuned untuk IDX context. Kalau setelah trial dengan portfolio real terasa terlalu agresif/lambat, edit di hpquant/portfolio/index.html:

categorizeReconstruction() — adjust KS thresholds (35, 55), P/L window (−5% s/d −40%)
brokenTranches() — adjust tranche count (4, 3, 2, 1) dan weekly % (25, 33, 50, 100)
Trigger price untuk Wait&Trim: default +10%, ubah di buildReconstructionPlan() Phase 3

8Risk Monitor

Pre-rebalance vs Post-rebalance metrics:

Metric	Limit	Tujuan
Max position weight	< 12%	Single-stock blow-up protection
Max sector weight	< 25%	Sector diversifikasi
HHI concentration	< 1500	Well-diversified threshold
# Positions	10-15	Sweet spot diversifikasi

HHI (Herfindahl-Hirschman Index)

HHI = Σ (weight)² × 10000

Contoh: 5 saham equal-weight (20% each) → HHI = 5 × (0.20)² × 10000 = 2000 (concentrated)
12 saham equal-weight (8.3% each) → HHI = 12 × (0.083)² × 10000 = 826 (well-diversified)
20 saham equal-weight (5% each) → HHI = 20 × (0.05)² × 10000 = 500

9Good PM Practices

1. Sweet spot 12 saham

Min 10 untuk diversifikasi proper
Max 15 untuk meaningful position size + manageable monitoring
Default 12 (Statman, Brinson, Lynch convergence)

2. Position sizing discipline

Max single name: 12% (panic-proof)
Min position: 5% (anything less = noise)
Average position: 8-10% di 10-12 names
Equal-weight default, kecuali high conviction (sampai max cap)

3. Risk management

Stop loss disiplin: -20% per name (atau breach guard line dari HPQuant chart)
Cash buffer: jangan all-in 100% — selalu sisakan optionality sesuai regime
Sector cap: max 25% per sector
Winner protection: jangan jual winner cuma karena signal Sell

4. Review frequency

Daily: monitor Crash Radar score, signal changes
Weekly: review sector playbook, scan new ideas
Monthly/Quarterly: full rebalance — execute trade plan
Setiap regime change: full re-evaluation

5. Behavioral discipline

⚠ Bias yang harus dihindari

Over-diversification: punya 30+ saham karena "lebih aman" — sebenarnya membenamkan return
Loss aversion: tahan loser terlalu lama → cut losses fast (-20% rule)
Disposition effect: jual winner terlalu cepat → let winners run (winner protection)
Anchoring: terikat ke harga beli → fokus signal saat ini
Recency bias: over-react ke berita kemarin → trust framework

10FAQ & Common Mistakes

Q: Mengapa harus konsolidasi 36 → 12 saham? Bukankah 36 lebih aman?

A: Bukan. Riset Statman (1987): diversifikasi benefit plateau di 12-15. Beyond 20 = diworsification — return jadi flat (mendekati index) tapi management cost (research + monitoring + transaction) tinggi. 36 saham juga membuat impossible untuk fokus pada quality. PM institusi typically 20-40 names DENGAN team analyst. Solo investor cocok 10-15.

Q: Apa hubungan Keep Score dengan 4-quadrant?

A: 4-quadrant adalah klasifikasi konseptual (matrix Signal × Tier). Keep Score adalah composite priority numerik (0-100, multi-factor). 4-quadrant menjawab "what kind of action?" (ADD/HOLD/TRIM/EXIT). Keep Score menjawab "which gets cut first?" untuk konsolidasi. Keduanya komplemen.

Q: Kenapa winner saya (AMMS +67%) ada di drop list?

A: AMMS punya gain bagus tapi tidak ada signal HPQuant data → KS rendah (47). Tapi karena winner protected, dia tidak akan dijual penuh. Cek panel "Winners protected (no full sell)" — AMMS ada di sana dengan badge 🏆. Hanya ditrim kalau overweight (>10% wt), kalau weight saat ini ≤10%, fully held.

Q: Trade plan menyarankan jual 60%+ NAV, terlalu drastis?

A: Output framework brutal honest. Anda tidak harus eksekusi 100%. Anggap sebagai direction signal: heavy SELL = banyak posisi anda noise/weak. Eksekusi gradual (mis. 30% week-1, 30% week-2, 40% week-3) lebih realistis. Pertimbangkan tax/fees Indonesia (Buy 0.18% + Sell 0.28%).

Q: Saya cuma punya modal kecil (Rp 500 M), bisa ikuti framework?

A: Bisa. Untuk NAV < Rp 1B, ideal size = 10 saham (target weight 10% per nama). Skip step yang membutuhkan minimum lot terlalu besar. Mulai dari 5-7 saham di 3-4 sektor berbeda lalu naik gradual ke 10. Rebalance bulanan, bukan harian.

Q: New Ideas terlalu sedikit / kosong?

A: Filter strict: Buy signal + sektor OW + Tier ≤ Sound + bukan dimiliki. Kalau kosong, market environment current memang sulit untuk add new positions. Better hold cash daripada force buy. Atau lihat sektor Neutral untuk options yang lebih banyak.

Q: Bagaimana kalau Crash Radar masih AMAN tapi saya merasa pasar mau turun?

A: Trust the framework. Crash Radar agregat 6 indikator macro objektif. Gut feeling sering salah. Tapi kalau anda kuat-kuat ingin lebih conservative, set personal cap mis. selalu sisakan 20% cash regardless of regime. Yang penting konsisten.

Q: Quality Score 53 → 84, P/L -3% → -5%. Kenapa P/L jadi worse?

A: Konsolidasi memilih saham dengan future conviction