Synthesis · All nine tracks
The synthesis: five laws, base rates, risk math
The distilled core of ~36,000 words of research across eight tracks plus the defence extension — the laws, the numbers, and the practice path. Detail and full citations live in the nine lessons. Anything marked UNVERIFIED in the lessons stays unverified here.
The five laws (apply in this order, every time)
- Legality first. If the route isn't legal for the person's residency, nothing else matters. Offshore forex margin trading is unauthorised for Indian residents (RBI Alert List, FEMA, LRS prohibition). NSE's Terms of Use ban scraping. SEBI's retail algo framework is mandatory since 1 Apr 2026.
- Costs second. Costs are the most reliable predictor of net outcome, and they dominate short horizons completely. A scalper's 0.1% target leaves ₹17 on ₹1,00,000 before slippage.
- Base rates third. Most people who try a style lose money at it. Any plan that ignores the base rate is fantasy.
- Risk math fourth. Position sizing, expectancy, and drawdown asymmetry are the only parts with hard math. Everything else is judgment.
- Evidence hierarchy last. Regulator studies > peer-reviewed research > backtests > backtested marketing > guru claims. A backtest is not evidence until it survives costs, out-of-sample data, and walk-forward testing.
The base rates — the single most important section
| Arena | Finding | Source |
|---|---|---|
| Individual F&O traders, FY26 | 87.7% lost money; ₹91,685 cr aggregate losses; avg loss ₹1,16,654; only ~1% netted >₹1 lakh after costs | Lesson 02 (SEBI) |
| Individual F&O traders, FY25 / FY22–24 | 91% / 93% lost | Lesson 02 (SEBI) |
| Intraday cash-equity traders, FY23 | 71% lost; 80% of those doing >500 trades/yr lost; costs added 57% on top of loss-makers' losses | Lesson 02 (SEBI) |
| Experience effect | ~Nil: FY26 loss rates 87.81% (new) vs 87.69% (regular); 54% with 3 yrs' experience still lost intraday | Lesson 02 |
| Retail FX/CFD accounts | 74–89% lose (ESMA 2018), 82% (FCA 2016), 68.4% (ASIC FY2024) — at regulated brokers | Lesson 03 |
| Active large-cap funds, 10 yrs | 76.3% underperformed (SPIVA India); ELSS 82.9%; mid/small-cap 79% | Lesson 01 |
| Active traders vs market (Barber & Odean) | 11.4% vs 17.9% market return; disposition effect costs ~3–5 pts/yr | Lesson 01 |
Note: FY25 F&O aggregate loss totals disagree between SEBI studies (₹1,05,603 cr vs ₹1,11,788 cr — sample differences). Treat the exact total as approximate; the loss rate is consistent.
The India rulebook (binding constraints)
- Hours: NSE/BSE equity 09:15–15:30 IST = 06:45–13:00 Bahrain, Mon–Fri, no lunch break. Currency derivatives 09:00–17:00 IST (cross-currency close time conflicts across sources — UNVERIFIED). MCX bullion/metals/energy to ~23:30 IST; agri to 17:00 IST.
- Settlement: T+1 for equities since Jan 2023. Demat via NSDL/CDSL through a broker DP.
- Leverage: peak-margin rules cap equity intraday leverage at ~5×; F&O MIS margin benefit largely removed. Bracket orders effectively dead; cover orders broker-selective. Currency derivatives: SPAN + exposure margin ~2–2.5% (~40:1 at minimum margin — still enough to wipe an account on a 2.5% move).
- Forex/FEMA: Indian residents may trade ONLY exchange-traded currency derivatives in 7 pairs (USD/EUR/GBP/JPY-INR + EUR/USD, GBP/USD, USD/JPY) via SEBI-registered brokers. Since 3 May 2024, rupee-pair derivatives require genuine underlying FX exposure; RBI Circular No. 25 (10 Oct 2026) cut the no-documentation threshold from $100m to $5m and banned rebooking cancelled contracts. Offshore margin trading: RBI Alert List (95 entries), LRS remittances for margin calls prohibited, FEMA penalties up to 3× the sum, enforced by the ED. Cross-currency pairs carry no exposure condition — the only exchange-traded directional route — but are far less liquid.
- Bahrain/NRI angle: FEMA's 182-day residency test decides which rules apply; NRIs are widely reported as barred from currency/commodity derivatives on Indian exchanges. Needs professional FEMA/CA advice — do not advise definitively.
- Algos: SEBI retail algo framework mandatory since 1 Apr 2026: registered static IP, exchange Algo ID, 10 orders/sec registration threshold, broker accountable, 5-year audit trail.
- Data: NSE Terms of Use (clause 9) prohibit scraping/harvesting. Safe paths: free bhavcopy downloads, NSE's official MCP servers (25+ tools, 5 years of history, 1–3 min delayed live data), licensed vendors.
- Tax (verify with a CA before acting): equity STCG 20%, LTCG 12.5% above ₹1.25 lakh/yr (post-23 Jul 2024); STT 0.1% each side on delivery; intraday equity = speculative business income; F&O = non-speculative; currency derivatives = non-speculative; commodity derivatives = business income + 0.01% CTT sell-side; gold ETF LTCG 12.5% after 12 months; SGBs: no new tranche since Feb 2024, Budget 2026 restricted tax-free exit to original subscribers only. ITR-3 for business income; 44AD excluded for speculative.
What works, by horizon (evidence-weighted)
- Long-term (years): the strongest evidence in all of finance. Broad index exposure, low costs (direct plans avg 0.96% vs 1.89% regular — a 1% fee costs ~₹10.3 lakh on a ₹10k/month 20-yr SIP), asset allocation, and behaviour beat stock-picking for almost everyone. Nifty's ~11–14% long-run CAGR came with −60% (2008), −54% (dot-com), −40% (COVID) drawdowns and 8–25 month recoveries. SIPs help behaviour, not returns; SIP date is irrelevant (<0.1% XIRR). Factors (momentum/value/quality) are real but rotate — never chase last year's factor. What predicts returns: earnings growth + starting valuation, loosely, over long horizons only.
- Swing/positional (days–weeks): the best-supported active horizon. Multi-month momentum has the strongest academic support in trend trading (Jegadeesh & Titman 1993, ~1%/month gross). Fewer decisions, lower cost drag, overnight gap risk is the price.
- Day trading / scalping (minutes–hours): the weakest evidence and the highest cost drag. No robust standalone post-cost evidence for RSI/MACD or classic chart patterns (Bulkowski's "81% head-and-shoulders" is a 5%-move threshold before costs, not a win rate). Scalping is a spread-and-speed game professionals dominate. Auction/short-delivery risk: an unsquared intraday short can be closed out up to 20% above settlement.
- Forex: carry and momentum are real, documented risk premia with crash risk (carry) — no robust evidence of a retail chart-pattern edge. Leverage maths: at 500:1 a 0.2% move wipes the account; gap risk defeats stops around RBI/FOMC events.
- Gold: diversifier, not a return engine. −0.03 to −0.05 equity correlation over 15–20 yrs; works over weeks in crises, falls with everything in acute liquidity panics (Mar 2020). MCX gold = dollar gold × USD–INR, so much of the rupee return is rupee depreciation. Futures ≠ spot: ~1%/month contango shrinks rolled exposure ~11.4% over 12 flat months. Access ranking for Indians: SGBs were best (2.5% interest + tax-free maturity) but no new tranches and restricted exit; ETFs are the clean default now; digital gold is unregulated (SEBI 2021 broker ban, Nov 2025 caution) with 3% GST + 2–6% spreads; physical has making charges + storage/theft risk. Crude lesson: 20 Apr 2020 WTI settled −$37.63, MCX −₹2,884/bbl (~₹435 cr payouts, upheld by Bombay HC) — never hold an expiring energy contract without an exit plan. Dabba trading is illegal with no investor protection.
The risk math (the transferable skill)
Expectancy: E = (win% × avg win) − (loss% × avg loss), in R-multiples, AFTER all costs. If you can't state your expectancy, you don't have a strategy, you have a habit.
Position sizing: fixed-fractional, risked as % of capital per trade (1% is the sane default for active trading). Size from the stop distance, never from conviction. Shares = (Capital × Risk%) ÷ (Entry − Stop-loss). Example: ₹5,00,000 capital, 1% risk = ₹5,000; entry ₹1,000, stop ₹980 → 250 shares.
Drawdown asymmetry: a 50% loss needs a 100% gain to recover. This one fact kills most aggressive sizing.
Kelly: a ceiling, not a target — full Kelly is unusable in practice (estimation error; ~1-in-3 chance of a 50% drawdown). Fractional Kelly or fixed-fractional sizing is the usable form.
Leverage wipeout: at margin m, an adverse move of m wipes the deposit. Gold ~6% margin, crude ~10% — overnight gaps go straight through stops.
Journal metric: rule-following rate, not win rate. Review weekly.
The practice path (no capital required to start)
- Weekend project (Lesson 05): free NSE bhavcopy data + ONE strategy written down BEFORE coding + next-open fills with real costs + a journal. Primary metric: rule-following rate.
- Honest backtesting checklist: no look-ahead bias; point-in-time universes with delisted names kept; capped parameter searches; all-in costs modelled; out-of-sample + walk-forward validation. Bhavcopy prices are UNADJUSTED for splits/bonuses — never mix adjusted and unadjusted series in one backtest.
- Paper trading in your own code. No Zerodha sandbox/paper trading exists; only Dhan (real sandbox) and Upstox (mock-only, no fills) offer broker test environments. Build fills yourself.
- Tiny size, then scale only on evidence. Scale on verified expectancy over 50+ live trades, never on a good week.
- Get professional advice before: NRI derivative eligibility, any tax filing position, and any use of leverage beyond exchange minimums.
Methods & tools — the evidence inversion
Technical analysis. The least glamorous methods have the strongest evidence; the most YouTube-popular have the least. Multi-month momentum is the one strong result (replicated across 58 markets and a century of data — Moskowitz, Ooi & Pedersen 2012; Hurst et al. 2017), but it is a months-scale phenomenon with crash risk, not an intraday edge. Classical chart patterns carry weak information and no post-cost edge (Lo, Mamaysky & Wang 2000; Osler 1998: head-and-shoulders averaged −0.24% per round trip before costs). Candlestick evidence is mixed and sample-dependent; location at a defended level matters more than shape. RSI/MACD effects are market- and parameter-specific; ADX, Stochastic, CCI, OBV, Parabolic SAR have no rigorous profitability evidence. ATR's sizing use is arithmetic, not a claim — the most defensible indicator use. Round-number support/resistance has the best mechanism evidence (Osler's 9,667-order bank study: 8.7% of orders at rates ending in 00 vs ~1% expected). Fibonacci levels were indistinguishable from random levels in a G10 FX backtest (~6.6M bars). Elliott Wave is unfalsifiable in real time; Wyckoff is a reading framework clearest in hindsight. SMC/ICT got its first large independent audit in 2026 (Tangirala preprint: 44 detectors, 496 S&P 500 stocks, 17.6M simulated trades) — none beat random entry against a matched null; the 80–90% FVG fill rates quoted online are unsourced lore. Full scorecard in Lesson 06.
Fundamental analysis — cash is truth. Read all three statements together on consolidated figures: profit that never arrives as cash shows up as receivables/inventory and as an absence in operating cash flow. Sloan (1996): lowest-accrual stocks beat highest-accrual by ~10.4pp/year — the 5-year CFO-vs-profit tracking test is the best first screen. Decompose ROE (leverage can fake it); ROCE must beat the ~9–12% Indian cost of borrowing. India governance checks: promoter pledging (SAST Reg 31 disclosure; DHFL had 80%+ pledged pre-collapse), related-party transactions, mid-tenure auditor resignation (a fire alarm — Manpasand/Deloitte, ~95% fall), mandatory auditor rotation. DCF is a range, never a point: terminal value is 60–80% of value, and ±1% in WACC/growth swings value 79→138 on a base of 100. P/E is useless for loss-makers and inverted for cyclicals (lowest P/E at cycle peaks). The 60-minute routine: business in one paragraph → 10 years of numbers → ratios/peers → governance sweep → valuation sanity → one-paragraph verdict; "I don't understand it" is a valid output. Full method in Lesson 07.
Tools — the free stack is real. Free standalone TradingView NSE data went 15-min delayed in late 2026 — free TradingView is no longer a live intraday tool for NSE; broker-integrated TradingView (Dhan's tv.dhan.co, Zerodha Kite) restores real-time data free. TradingView paid tiers (verified 30 Oct 2026): Essential $12.95/mo, Plus $29.95, Premium $59.95 annual billing. Pine Script v6 is on all plans including free but cannot place orders alone; webhook automation = algo trading under SEBI's framework — broker GTT orders are the compliant free alternative. Screener.in Premium ₹4,999/yr (verified). Recommended stacks: free ₹0/mo (Dhan + Screener.in free + NSE official + GTTs); sensible paid ~$12.95/mo + ~₹417/mo (TradingView Essential annual + Screener.in Premium). Pricing date-stamped 31 Oct 2026; INR TradingView pricing conflicts across sources — UNVERIFIED, check at checkout. NRI correction (31 Oct 2026): Dhan does not offer NRI accounts — for Bahrain-based NRIs the free-stack broker is Zerodha Kite (NRI onboarding, ₹500 opening, ₹500 + GST AMC, non-PIS equity 0.5% capped ₹50/order). Full comparison in Lesson 08; the setup guide is on the Tools page.
Defence — the extension in brief
The demand story is real and policy-built: FY27 defence allocation ₹7.85 lakh crore (+15.2%), capital head ₹2.19 lakh crore, 75% earmarked for domestic industry; exports ₹38,424 crore in FY26 (+62.7% YoY); 26 listed defence companies earned ₹23,111 crore in FY26, +208% vs FY21. But approvals ≠ contracts — the ₹9.3 lakh crore AoN pipeline and the ₹2.12 lakh crore DAC clearance of 26 Oct 2026 are not revenue until signed, and conversion is the binding constraint. The stocks already ran and are correcting: Mazagon +1,232% over 5 years but −38% over 1 year and −41% from its high; shipyards and BDL down 24–38% over the year; HAL/BEL sideways. Order books: HAL ₹2.55 lakh crore (~8× sales), BEL ₹73,882 crore, BDL ₹26,176 crore (10× sales yet FY26 sales fell 27% — delivery phasing); Mazagon's book is declining (~₹20,000 crore). Cash conversion diverges: HAL ~125% of EBITDA, Mazagon under 30% with −₹2,654 crore FY26 operating cash flow. Valuation spread is extreme: Solar P/E ~87 vs Mazagon ~28; BDL ~70× with the worst ROCE (7–14%). The binary catalyst: ~₹1.5 lakh crore awaiting CCS clearance (QRSAM, P75I submarines, next-gen corvettes). Full watchlist with kill switches in Lesson 09 and the Ideas page.
Open questions — confirm before acting
- NRI eligibility for currency/commodity derivatives (FEMA/CA advice).
- Dividend TDS threshold; exact F&O close time; cross-currency close time; client position-limit dollar figure; Zerodha/Angel/Fyers API rate limits (UNVERIFIED in Lesson 05 — re-check official docs).
- SGB taxation for secondary-market buyers post-Budget 2026 (CA).
- Whether SEBI's FY26 F&O figures are final or interim.