The few today's action — top-tier buys, sells due, and your open book

To BUY in your broker (the shop just added these - mirror them, then tick)

To SELL in your broker (the shop exited these - sell, then tick)

Exit likely soon (flagged - a sell is coming; heads-up, no action yet)

Add to your winners (proven compounders on a pullback, thesis intact - the greats add here)

Trim to target (re-graded down - reduce in your broker to the target size)

Purchase order

Seven gates. All must pass, exactly like no piece reaches the display case without cost, ticket and appraisal.

Watchlist

Import a holding you already own

Skips the gates (it is already bought). Complete its card after.

Aged inventory, the markdown queue

The time stop protects against dead capital, not volatility. It never triggers on a price drop, only on nothing happening.

Sold, realized

New shop ideas Recalculate all

Ranked by Kaya's own engine, all math computed in the shop. An idea still has to pass the gates AND show a BUY signal.

Runs / Relay catch the sharp LEG, scale out, wait, re-enter · 100× in parts · Refresh

Igniting now fresh legs to catch · ELITE = strong fundamentals (a sustainer, not a pop)

Re-entry watch former runners (≥3×) now BASING - wait for the next ignition

Live track the board grading ITSELF - real, out-of-sample ·

Every fresh ignition is logged and ridden on a trailing stop (no capital - paper proof). The honest record as legs open, scale out at 2×/5×, and close.

Open legs

Closed legs

THE LADDER · 1000× in 10 years double every year = 2¹⁰ = 1024×

THE COPPER SUPERCYCLE 15 years of electrification demand vs constrained mine supply — the disciplined read (research/99)

MARKETS — RATES & RISK the risk-free rate, country risk premiums & sector margins every valuation runs on

VALUATION STUDIO type any ticker — the full Damodaran story-DCF, the Monte-Carlo value distribution, and a plain call

RE-ENTRY proven runners (≥3×) parked & re-armed — never sold, never chased: BUY on the next leg, WAIT for ignition

PRE-IGNITION coiled before the rocket — a volatility squeeze in an uptrend near a pivot, volume drying (quiet accumulation), growth + cash behind it, and valuation runway still ahead

THE LAUNCHPAD multi-year base → breakout — the honest, measured read (research/100)

HYPERGROWTH small/mid caps with MEGA-CAP stats, caught EARLY — the NVIDIA/PLTR pattern before the run

GIANT RE-RATE ≥$30B giants that CRASHED & are re-accelerating on hypergrowth — the NVDA / PLTR shape

5-DAY TRADE REPORT accounting for the cash-week strategy — realized P&L, win rate, full ledger

RIDING your big-run holds — live price, trailing-stop exit, ACTION alerts

THE WINNER LIBRARY measured case studies of the base → breakout → multibag pattern, all markets (research/101)

RED FLAGS fraud & blow-up screen — every name in the system carrying an issue (Manpasand / PC Jeweller / DHFL type)

3× · 5× · 10× the multibagger hunt (INVEST, not trade) — small-caps bucketed by upside runway. Descriptive; hold for years, most never multiply.

Multi-month green names riding a sustained monthly-green run right now (the PLTR/MU setup) — fueled = real engine, momentum = cyclical/hype. Ride, don't predict.

Accel ride the Revenue-Acceleration Ride — above a rising 200d + revenue growth accelerating; hold 3-6mo, exit on the chandelier stop. OOS ~68-75% win, PF 4.4-6.8.

NEXT WINNERS VERY HIGH CONVICTION only — top-tier 100x-DNA, clean verified books, no commodity price-takers or funds (the capstone)

Long Run finder Scan universe

The multibagger doctrine: names that historically ran 5-10x on monthly Heikin-Ashi candles without a confirmed break, graded on today's setup - technicals, fundamentals, the ticker's own run history and the market regime must ALL be green at once. No setup, no bet. The crash guard is the standing catastrophe exit: a COVID/GFC-type event trips the MARKET alarm and clears the shelf before the damage compounds.

Great compounders - ranked by the earned-multibagger signature: margin EXPANSION + operating leverage + revenue growth + earned-not-hype (from the 33-winner study). QUALITY signal - descriptive of what winners shared, NOT a backtested return predictor (survivorship makes it untestable on free data). The few, not the many.

Best value finder

Multibagger finder

The PLTR-at-$10 setup: a stock that crashed ≥40% from its top, whose monthly Heikin-Ashi just flipped green, backed by real growth (≥15% revenue CAGR). Backtested on 11,007 flips over ~50 years, this deep-reversal-with-growth is the ~77% winner (vs ~64% for a bare flip, ~48% for a loss-making bounce). Ranked by multibagger conviction = crash depth × growth × profit/FCF × valuation. Buy as a basket, ride the green, exit on a confirmed red month.

Doubler finder

Small & micro-caps ($100M-$2B) in the doubler setup: deep crash off the highs + volume ignition + proven multibagger DNA + small base + a fresh turn, NOT extended, and it must survive (cash runway & sane debt). Big companies cannot double - this is where 2×/10× actually lives.

HONEST ODDS - there is no 90% doubling: across 64 years, the best deep-reversal + growth setups end GREEN ~77-81% of the time, and the ride keeps ~90-100% of runners green once they move - but only ~12-16% ever double (over years), and ~0% double in a month. The GREATEST tier below is a short list of survivors (not distressed), turning and growing. Size tiny - a few carry the basket, most will not.

Forecast accelerators

Names in this universe whose forward earnings estimates analysts are actively RAISING - the leading signal behind moves like MU's: a rising estimate + a beat streak front-runs the price. Ranked by forecast momentum.

HONEST - a forecast is not a fact: analyst estimates revise HARD and are least reliable for cyclicals near a peak (memory / semis / commodities) - those get slashed when the cycle turns. This catches the turn EARLY (rising estimates off a base, NOT extended); a cyclical or extended tag means it is a trade with a hard exit, never buy-and-hold. Chasing after the run buys the top.

Value 10×

The same 10× cores, ranked by valuation (Damodaran DCF) instead of ride timing. GARP = growth DNA and a fair price — both engines agree (rare, highest conviction). Momentum = great growth but frothy — ride only, where the DCF is a risk gauge, not a buy signal. Tap any name for its full DCF workbook.

10× Holdings

The names you own from the 10×/100× shop, as a live ride table. Add a ticker (+ your entry price) and it tracks the price, your P&L, the quarterly ride state (RIDING / RIDE OVER · SELL when the quarter flips red), and the DCF froth. Stored on this device.

Picks — right-price buys, tracked live by quarter

Every right-price BUY the engine flags (great business + trend up + still early, not already run) is logged here the moment it fires — with live entry price, P&L, and exit when the quarterly trend breaks.

Multi-fold 10×

The shop — pure long-term multi-fold, not trading. CORE = fundamental: hyper sales growth (≥20%) + BIG gross margin (≥50% — the unit economics; net profit is allowed to come later, like ETON 67% while still losing) — that is what makes the 10× (MU 65→1200, PLTR, NVDA all did it on fundamentals in ~12–14 months). The technical is only the ride: monthly & weekly Heikin-Ashi green tells you when the trend supports entering, never the thesis. Global (US + India). Buy small, hold 12–24 months, ride until the monthly flips red.

HA Stack

Multi-timeframe Heikin-Ashi alignment. The higher frames set the trend, the lower frames time the trade: monthly & weekly must be STRONG green (body up, tiny lower wick = buyers in control) — the trend gate. Once gated, BUY when 1-day & 4-hour are green too; EXIT the moment 1-day or 4-hour flips red.

Honest note: requiring all four frames to align is a strong trend filter — it cuts whipsaw and keeps you on the right side of the move, but it is not “win almost all”: you enter later and give a little back on the exit flip. It wins by riding aligned trends and cutting fast when a lower frame turns, not by a high hit-rate.

Trend sleeve

The second leg of the book (research/68). A name is IN TREND when its price sits above its own 10-month average, OUT when it closes below. Long-only, equal-weight, reviewed monthly. This leg is only ~0.4 correlated to the dip-swing book, so running the two together is real diversification: the blend lifted return (CAGR 9% to ~16%) while cutting the worst drawdown from -71% to -45% - shallower than either leg alone.

HOW IT WINS (the opposite of the dip book): a trend leg has a LOW hit-rate (~39% of trades) - many small losses, a few big rides. It makes money by exiting at the line (cut the loser) and letting winners run, not by being right often. Its job here is drawdown reduction for the whole book, not a high-win pick. Size it as a sleeve and equal-weight the names; never concentrate.

THE RACE · your CAPSTONE plays

100× DNA

Small-caps carrying the DNA every 100-bagger shared at its start (Chris Mayer’s study of 365): small base, high sustained growth, high reinvestment (ROE), owner-operators (insider skin), long runway. DESCRIPTIVE, not a prediction - fundamentals are survivorship-tinted and nothing can foresee a 100× (10-25 years + patience + luck). A coffee-can watchlist to research and hold for years, never a promise that any will 100×.

This week - the plan concrete orders, sized to your capital

Mirror these in your broker. Sizes are the desk's suggested per-name amount; Exp $ is the mean expected dollar (Cash Week) or the premium collected (income) - an expectation with variance, never a promise.

Closed trades the public track · in, out, invested & made · each name sized at 5% of base

Every closed trade from the same forward-log the desk grades on: the stock (with the price it went IN → OUT underneath), how much was invested in it, how much it MADE, and the return. Real and graded - the losers are here too, no cherry-picking. Tap a row for the full chart.

By week · summary

Supporting sleeves extra upside · NOT in the base $

Week desk WEEKLY TRADING · SEPARATE STRATEGY Refresh

Panic Bounce STOCKS + CRYPTO · CASH · LONG Refresh

Holding — the recovery tracker your open bounces, riding to RSI-2 ≥ 70

How to read: each row is a bounce you already bought. SELL when RSI-2 fully recovers (≥70) or at the 20-bar cap — the "Action" column tells you. Green = up. HOLD = keep riding.

Snapback Swing S2 · US STOCKS · CASH · LONG Refresh

Holding — the recovery tracker open snapbacks, riding to RSI-2 ≥ 60 / ~day 9

How to read: each row is a snapback you bought. SELL when RSI-2 recovers (≥60) or at the ~9-bar cap — the "Action" column tells you. Green = up. HOLD = keep riding. No stop — the flat $1,500 size is the risk control.

Moonshot LOTTERY SLEEVE · lose-able $ ONLY Refresh

Riding — the moonshot tracker cut at the stop, let the winners run

How to read: "Peak" is the best it has reached; the Stop trails 20% below that (never below −8% from entry). SELL the instant it closes below the stop — the "Action" column tells you. HOLD = let it run.

India Moonshot NSE BREAKOUTS · lose-able ₹ ONLY Refresh

Riding — the India moonshot tracker cut at the stop, let the multibaggers run

How to read: "Peak" is the best it has reached; the Stop trails 20% below that (never below −8% from entry). SELL the instant it closes below the stop. HOLD = let it run.

Phoenix DEEP-BASE TURNAROUNDS · US + INDIA Refresh

Riding — the recovery tracker armed ride: no stop until a 5× peak, then sell −60% off the peak

How to read: "Peak" is the best it has reached. The trail is ARMED: below a 5× gain there is no stop — deep bases chop >60% WHILE basing (TSSI dipped to −63% off an interim peak, then ran to 66×), so a blanket stop shakes you out right before the moonshot. Once it has run 5×, the Stop arms at −60% below the peak. (ExitOpt backtest: the only exit that beats hold-to-now on median, +12%, and still rescues the round-trips.)

Strength BREAKOUT · EARNINGS EXPLOSION Refresh

Riding — the breakout tracker armed ride: no stop until a 5× peak, then sell −60% off the peak

Record what we run, and how it is actually doing Refresh

The honest ledger. Every desk below is graded in PUBLIC on real forward trades — not the backtest. Validated = the measured backtest it earned its place with; Live = the record accruing since it went live (Open still running, Closed graded, Win% / Avg on the closed trades). Open the shareable customer page →

Build log what we changed, when

Playbook swing-strategy research · session 2026-10-10 · where we are, so we resume cleanly

Everything measured below is from backtests (15-agent swarm, 30-year daily data, 234 US names through the 2000 / 2008 / 2020 / 2022 crashes). Survivorship-inflated ~5–10 pts → live is lower. Nothing here is wired into the live desk yet — see Next steps.

The winning engine — “S2 + fast exit + gate”

Research Lab — Validated Edge Stack Kelly Sharpe 1.47

The real edge is not one magic signal — it is MANY small uncorrelated edges, Kelly-sized (the Renaissance principle). Avg pairwise correlation -0.00 (~zero), so losses land on different days and the book's drawdown is far smaller than any single edge. VALUE & TSMOM pay when the others don't (momentum crashes / bear markets). Every edge is walk-forward + cost + survivorship tested; trust grade + caveat shown, never hidden. Over 4 research waves, 20 families hunted, 8 edges + a risk overlay kept, 11 rejected — the discipline is the product.

+ Risk overlay VolTarget: scales whole-book exposure inversely to realized vol; not a return stream (corr 0.9 to SPY) but halves tail drawdowns (2008 −28% vs −55%) at ~zero bull cost.

Combined Kelly Sharpe ~1.47 from 8 edges that individually top out at 0.69 — the √N diversification win (4 edges → 0.97, 6 → 1.20, 7 → 1.41, 8 → 1.47; gains shrink as edges get scarcer — honest diminishing returns). Honest: backtested (survivorship-haircut, some bull-tilted); live will be lower. Scoreboard is expectancy, never win rate. 90% does not exist. Proof accrues live via edge_book_live.mjs.

Net economics — per $1,500 ticket, losses already subtracted

The real constraint: THROUGHPUT, not edge

S2 fires only ~38 signals/year across 234 names — not enough to keep even one ticket continuously busy (a ticket wants ~65 trades/yr). So on this universe the total net output is capped at ~$1,000–2,000/yr no matter how much capital you add; extra tickets sit empty. The edge is solved; the $25k/mo goal is now a throughput problem. Fix = a much bigger scan universe (2,000–4,000 liquid US names) or the looser S1 filter (RSI-2<5 + steep-trend only: ~82 signals/yr, ~$62/ticket net).

Doctrine (locked this session)

• Goal = net $/month on recycled capital. Win rate is a vanity metric — proven: a +0.5% target wins 97% but makes $5/mo and loses after costs. No 90%-win-that-makes-money exists.
• Long only — shorts fight the market's upward drift (~53% win) and carry unbounded risk.
• $1,500 flat per bet, recycled as trades close. Size is the risk control.
• No price stop — measured 3 ways: even a perfect-fill stop has PF 1.65 < 1.91 no-stop (it severs the mean-reversion recovery), and gaps blow through stops anyway.
• Exit FAST — the master lever: makes the MOST money AND the SMALLEST losses (a 9-day loss is −5.7%; the same trade held 60 days rots to −17% / −82%).
• Bigger timeframe (weekly) raises win% to ~69% but makes the SAME money (slower recycle cancels it) and starves throughput — a comfort choice, not a money one.

15-expert swarm verdicts (30y)

Also shipped this session (live now)

• Gold + Silver (Metals sleeve) wired into Panic Bounce — +3% hold, fire on own oversold, uncorrelated. Scan now 416 instruments.
• $1,500 per bet set everywhere (was $500) — both desks, all UI, alert email.
• Real-cash recycling accounting on Record + the public /share page (peak cash, realized P&L, return on capital).

▶ Next steps (when you resume)

1. Wire the S2 engine (deep+steep+hivol+momentum entry, fast RSI-recovery exit, golden-cross gate) into the desk as a new board, graded live on TRACK so the real survivorship-shaved number proves itself.
2. Expand the scan universe to a few thousand liquid US names — the single thing that unlocks the $25k/mo path (feeds throughput).
3. Pick the operating point: S2 (selective, ~$87/ticket net) vs S1 (2× the signals, ~$62/ticket net).
4. Re-point the accounting to the S2 engine once live.

Where is the sun? it is always day somewhere

Measured over 308 straight months: there was never a month without doublers (worst ever, Dec 2008: still 0.3%, with the top-1% stock +73%). This board names the CURRENT daylight - for each market × sector: the share of members that DOUBLED in the trailing 12 months, the share still green this month, and the median best 12-month gain. Read it with the cream rule: high doubled% + fading green% = a spent sun (the leftover zone - never chase); high green% + low doubled% = a rising sun (early - the pond to watch); both high = high noon (harvest, via the funnel only).

Paper LEAPS PAPER ONLY - no capital

A validation sleeve: the machine "buys" (on paper) an in-the-money LEAPS call on the shop's top-2 conviction cream names, marks it every cycle, and closes on the underlying's confirmed structural break or the ~60-day roll point. Max loss = the premium (defined risk, the 1:10 geometry). If a quarter of paper P&L proves the convexity math, it earns a small real sleeve - not before.

Simple options · bet a direction max loss = the premium

The tool reads the trend (price vs the index's own 10-month line) and recommends ONE side - a CALL if the trend is up, a PUT if it is down - at a weekly and a monthly expiry. No betting both ways. Index timing is the hardest call there is, so treat the read as a low-confidence lean and size tiny. The most you can lose is the premium (a contract can expire worthless). NOT backtested; indicative pricing, confirm live. India = INDA (US-listed MSCI India proxy; native NIFTY/BANKNIFTY need an Indian broker feed).

Check any ticker any market · call or put across horizons

Type any symbol - USA (AAPL, NVDA, SPY), India (add .NS: RELIANCE.NS, TCS.NS), or an ETF. The tool reads its trend and shows the recommended side (CALL if up, PUT if down) across 1 day / 1 week / 1 month / 45 / 90 days. The read works for every market; live option pricing is US-listed only (India/native NIFTY need a broker feed - trade the underlying or use INDA). Low-confidence lean, not a prediction; short expiries bleed to theta; max loss = the premium.

Options · LEAPS leverage

Long-dated (1yr+) in-the-money call options on your highest-conviction MULTIBAGGER picks - leverage the best ideas with hard-capped downside (you can only lose the premium). HONEST: options are NOT backtested (no historical data - a structural tool, not a validated edge), can expire worthless (100% loss), and pricing is indicative (last trade - confirm live). A small deliberate sleeve only, never the core.

Income · covered calls on your holdings

Sell an out-of-money call on a name you own -> collect premium as income; caps upside at the strike. Indicative pricing, not backtested.

Income · cash-secured puts at your dip targets

Sell a put near a safe-swing dip target -> get paid to wait; if assigned you buy the dip at (strike − premium). You must hold the cash. Indicative pricing, not backtested.

Hedge · index put-spreads AI-systemic tail protection

Cheap, defined-cost crash insurance for the AI-financing-complex concentration (research/25): buy a ~7% OTM put, sell a ~18% OTM put. Small fixed cost, pays off if the index falls into the −7%..−18% band. A hedge, not an edge; indicative pricing. India is hedged via INDA (US-listed MSCI India ETF) - the only India proxy with a liquid US options chain; native NIFTY/BANKNIFTY and the pure Nifty-50 ETF (INDY) have no usable options on this data feed.

Cycle - position protection

Weinstein stage analysis on every holding - where each sits in its market cycle. TOPPING = Stage 3/4 roll-over (backtested: caught 25/26 major peaks, gave back −21% vs −74% riding) - watch the exit. PARABOLIC = blowoff early-warning (fired 11-26 weeks before 6/8 blowoffs, zero false fires across 2,294 calm weeks) - book into strength & tighten. launch = Stage 1->2 turn. Exits still need a confirmed red month; this is timing context, never a forced order.

Real opportunities growing · moving · moat · at a discount — Refresh

Undervalued only counts when the business is growing and the stock is moving — never a cheap dead value trap. These clear all four: a plausible discount (15–100%), revenue growth ≥15%, an uptrend (Heikin-Ashi green), and a durable moat. The shop's best live ideas — most sit in WATCH because the strict buy gate wants full confluence; this surfaces them so they don't rot.

Today's picks

The daily driver - the two highest-conviction, backtested setups across the whole universe. SAFE SWING = quality large-caps (won't go to zero) on a dip, uptrend intact - ~70-80% recover in ~6mo. MULTIBAGGER = not-extended strength + proven-DNA (tier 3+) - ~2× doubling lift. Both HONEST: not certainty; size small, most fail, winners run.

Master swing - durable dips (durable + calm + up-close, market not frothy; +10% stopless; ~82% out-of-sample honest, not 100%)

Multibagger tier 3+ (buy strength, size by tier, hard cap 15%)

Swing trades ~82% out-of-sample · NOT 100%

THE MASTER FORMULA (research/54,60): a DURABLE (never crashed >35%) + CALM (annualized vol <30%) name on a −8 to −25% dip, long uptrend intact, that CLOSED THE MONTH UP with 6-month buying accumulation, bought only when the market is NOT frothy (index <+15% over its 40-month line). Target +10%, STOPLESS (bank half, let half run); spread 1/N across 12-15 names, half-size if VIX>25. Backtest 473 instruments 2001-2026: honest ~82% deployable (purged walk-forward out-of-sample; 90% CI 74-89%; ~96% in-sample hold-to-target), NOT 100% - recent decade ~82-88%, ~1-2% never recover. Retires the old 47%-win swing.

Valuation MOS overlay (research/52,54): a Damodaran DCF margin-of-safety check now sizes each pick and excludes names priced past their DCF value. strong = at/below the pessimistic case (size up); ok = at/below base fair value; stretched = above base, no margin of safety (size down); veto = above even the +25% optimistic case (dropped). Replaces the old crude P/E>60 bypass. DESCRIPTIVE valuation context - the DCF is a reference price, not a win-validated predictor.

All master-formula swings (broader, ~82% out-of-sample honest)

Email record every recommendation the shop MAILED you, logged at its price and graded in public

Track record

The live proof - every high-conviction call logged at its entry price and graded to market. This is the ONE thing no backtest gives: are the edges real in the wild? Grows automatically as new picks fire. Swings grade to target/stop; multibaggers ride till a confirmed red month.

The shop floor

The engine trading its own recommendations, live - and sized by the opportunity: 2% for an ordinary qualified flip, 5% for a great setup (PRIME or a long-fall reversal), 10% only for the exceptional (SUREFIRE, or PRIME ending a 12+ month fall). One pot of capital backs both markets - a great opportunity draws from the whole shop wherever it appears; Indian pieces stay priced in rupees, capacity converts at the live rate, never more than 80% deployed overall. The doctrine's exits close every position. Green boxes are inventory on the shelf; a blinking red box is the engine calling the exit. Open this page instead of the email - it IS the email, alive. This book is the machine's public track record trading itself; your real account replicates it.

Signal ledger

Every alert the engine ever emailed, written down the moment it flashed - date, call, and the price on the tape that day - then marked against today's price forever. This is the shop's book: the recommendations cannot be edited, forgotten, or quietly deleted. The ledger starts recording from the day it was installed; it fills with every alert run (twice daily, after NSE and US closes).

Conditions barometer

Live 3-month trend of the factors that drive every regime - the dollar, hard assets, commodities, bonds, yields. Green = rising, red = falling.

Aligned with this regime (what expresses it - and which are in your screen)

The canonical vehicles for the current backdrop, live 3-month trend. Names tagged in your screen are in your universe (click to open). DESCRIPTIVE context - buys still come only through the gates.

The playbook - what grows in each condition

DESCRIPTIVE positioning context, NOT buy signals - the current regime is highlighted NOW. Actual buys still come only through the gates (SHOP FLOOR / TO DO). The debasement trade is crowded and rate-sensitive - it whipsaws (gold ran +65% in 2025, then fell hard in 2026).

New here? Every badge and short form used across the shop, in plain words. You can also tap any coloured badge anywhere in the app and a little box pops up with its meaning.

How this shop works

One line: hunt the few great compounders early, buy ONLY what the shop books, ride till the trend flips, cut losers by rule, never churn. The shop is a signal book - it never touches your real money. You buy/sell in your own broker, mirroring the TO DO list. Every tool below is tagged VALIDATED (a real buy/exit engine), DESCRIPTIVE (research only - never buy straight from it), or SPECULATIVE / CONTEXT.

The only buy rule

You buy in your broker only when a name books to SHOP FLOOR and shows in TO DO (you also get an email). Watchlists and screens (100X DNA, COMPOUNDERS, the market table) are not orders - a name there becomes a buy only if it later clears a price gate and books. If it is not in TO DO, you do nothing. Today the list is empty (market froth) -> the right action is hold cash.

• +10% target on a swing -> bank half, let half run (stopless on the runner).
• −20% hard stop from entry (daily close) -> auto-sell. The disaster backstop; caps every single loss.
• Confirmed full-body red month (monthly candle must close red) -> the trend flipped, exit. Intramonth wiggles do NOT count.
• Crash guard (index regime = CRASH) -> sell all, protect capital.
• Abort line on a fresh reversal -> a monthly close below the turn's low cancels the thesis.

Honest limits

Not a crystal ball. VALIDATED engines have a real, tested edge but the honest out-of-sample win is ~82% on the swing (473-name walk-forward, 74-89% CI; ~96% in-sample) and ~73-77% on the ride - never 100%, and ~1 in 10 years is still down. DESCRIPTIVE tools (100X DNA, COMPOUNDERS, scores) are survivorship-tinted and not proven predictors. The real edge is discipline + expectancy + survival over time: buy only what qualifies, cut losers by rule, don't churn, don't overpay, obey the crash guard. Anything promising certainty is lying.