# Method

# Gautam Jha ("Stock Learners"): backtest plan (FINAL, 11 Oct 2026)

Subject: Gautam Jha, YouTube "Stock Learners" (UCg3LNMpW2sgZc3nthsbsfzQ, 143 videos + daily "Crypto Gold Live Trading"
streams) and "Gautam Jha" (UCTS8wjQ3Ivzh4q99QMiZchQ, incl. "Masterclass 1.0", pth7Wl4ZyY8). Identity, money trail, critics and
his public record: agents A/B/D (`projects/.research/gautam/`). Reference method: `projects/stockburner-audit/BACKTEST_PLAN.md`,
`projects/topg-audit/BACKTEST_PLAN.md`. Code: `tools/gautam_backtest.py`. Rules (every rule carries its quote + seconds):
`projects/gautam-audit/gautam_rules.json`; pre-registered copy `prereg/gautam_rules_20261010T1306Z_v1_prerun.json`
(sha1 ec267ab3…, written before any run). Generated tables: `BACKTEST.md`; data: `engine/public/projects/gautam-audit/data/`.
Nothing here is a trade recommendation.

> **Status (final 11 Oct 2026; run 10 Oct 2026, cross-checked, numbers unchanged).** His headline strategy found, coded and run on gold (his market), NIFTY, BANK NIFTY,
> BTC and ETH; Trading With Sidhant's reading and the opposite reading both run; his 2023 swing strategy run as a secondary test.
> Whisper: the "50% at 1:10" claim is verified 3-way; owner listens pending for C2 ("1:6–7" vs caption "1:67") and C11 (2022 "1–2 in 10").
> 39 of his videos read for rules (`projects/.research/gautam/C_strategy/rules_extract_masterclass1.md`, `rules_extract_other_videos.md`;
> KZxJme_W6V0 missing: caption download blocked).

---

## 0. What he claims (his words first)

Captions = YouTube's own Hindi ASR (`hi-orig`, `projects/.research/gautam/C_strategy/subs/`); Whisper = local large-v3-turbo on
short audio sections (`…/C_strategy/whisper_windows/`; Whisper section times drift ~5–7 s, use caption times for clips).

| # | Claim | Where | Check |
|---|---|---|---|
| C1 | «मोस्ट ऑफ द केसेस तुम 10 ट्रेड प्लेस करोगे एक्यूरेसी बहुत अच्छा है कि चार पांच ट्रेड छह ट्रेड अगर तुम समझदारी से लो फोमो में ना लो तो चल जाएगा। आप बोलोगे 50% का एक्यूरेसी 1:2 के रिस्क रिवार्ड पे नहीं है। यह 50% का एक्यूरेसी 1:10 के रिस्क रिवार्ड पे है।» "In most cases you place 10 trades and the accuracy is very good — four, five, six trades will work if you take them sensibly, not in FOMO. You'll say 50% accuracy isn't at 1:2. This 50% accuracy is at 1:10 risk-reward." | "Best Trading Strategy For Beginners - Scalping Strategy" (Stock Learners, 7 Jun 2026, **1.83M views**), [14:22–14:37](https://www.youtube.com/watch?v=wI9b968AvW8&t=862s) | Captions (word times 862.6–877.0 s) + Whisper hi («50% का accuracy 1-10 के risk reward पे है») + Whisper en ("This is 50% accuracy is 1-10 risk") — **verified 3-way** |
| C2 | «एवरेज ट्रेड का रिस्क रिवार्ड 1:67 है मेरा … जितना लोगों का ड्रीम प्रॉफिट होता है वो मेरा फेलियर ट्रेड होता है 1:3» "My average trade's risk-reward is **1:6–7** … people's dream profit is my failure trade, 1:3." | same, [6:21–6:27](https://www.youtube.com/watch?v=wI9b968AvW8&t=380s) | Caption reads "1:67"; **Whisper hi: "1 is 6 to 7" / "1-6-7"; Whisper en: "1-6-7"** → read as 1:6–7. Failure trade 1:3 (caption + Whisper hi "1 is 3"). Owner listen |
| C3 | «1:10 वाला ट्रेड अगर मैं दे रहा हूं तो वो मारो … 1:200 भी दे दिया है» "When I'm giving you a 1:10 trade, take that … I've even given 1:200." | same, [6:07–6:18](https://www.youtube.com/watch?v=wI9b968AvW8&t=367s) | Captions + Whisper hi + en agree |
| C4 | «वही पैसा तुमको 10x या बोल सकते हो कि 1:10 … 1:20, 30 तक का रिस्क रिवॉर्ड दे देगा» "the same money gives you 10x, i.e. 1:10 … up to 1:20, 1:30" | same, [12:25–12:34](https://www.youtube.com/watch?v=wI9b968AvW8&t=745s) | Captions; Whisper en "1 to 10, 1 to 8, 1 to 20, 30" |
| C5 | «दुनिया का बेस्ट स्ट्रेटजी हो या ना हो ये तुम्हारे लिए तुम्हारा दुनिया का सबसे बेस्ट स्ट्रेटजी बनेगा … तुमको रिजल्ट दिखेगा» "World's best or not, it will be the world's best strategy for you … you'll see results" (then: "backtest it two days; if your mind isn't blown, talk to me") | same, [1:22–1:55](https://www.youtube.com/watch?v=wI9b968AvW8&t=82s) | Captions |
| C6 (fairness) | «तो इसका एक्यूरेसी 30 से 35% रहेगा 40% रहेगा» "Its accuracy will be 30–35%, 40%" — about small-timeframe entries on liquidity alone | Masterclass 1.0 (recorded ~15 Nov 2025, uploaded 16 Jun 2026, 781K), [107:08](https://www.youtube.com/watch?v=pth7Wl4ZyY8&t=6428s) | Captions (rules_extract_masterclass1.md N10) |
| C8 | "dono mere Telegram ka accuracy more than 50% hai with 1:10" — about his Telegram calls | live "Crypto Gold Live Trading 14 Sep" (0stizzV7D1Y, 4344.5–4375 s) | from agent B/D (02_critics.md / D_record); verify there before use. Compared in BACKTEST.md with what his public strategy does |
| C9 | "group ka accuracy more than 60–70% hai with 1:10… phenomenal" — his group | live 17 Sep 2026 (e4lZ5QeeZ6E, 37:19) | from agent B/D |
| C10 (2nd strategy) | «जबरदस्त प्रॉफिट बंता है कंसिस्टेंट प्रॉफिट बंता है 89% एक्यूरेसी है» — caption "89%"; **Whisper hi + en: "80–90% accuracy"** | "SWING TRADING STRATEGY" (Stock Learners, 24 Mar 2023, 368K), [20:31–20:36](https://www.youtube.com/watch?v=aBaJD8Bnx6w&t=1231s) | Whisper hi + en agree on 80–90% → use "80–90%", never "89%" |
| C11 (2022, contrast) | «नॉर्मल ही 1:4 1:10 … 1:20 तक का ट्रेड जो है इजीली मिलेगा … उसमें से [एक दो ट्रेड] तो … [1:10 1:20] का मिल जाएगा» — "normally you'll easily get 1:4, 1:10, up to 1:20 … of those, one or two trades will give 1:10 / 1:20" | "BEST INTRADAY STRATEGY FOR BANKNIFTY" (16 Apr 2022), [3:15–3:25](https://www.youtube.com/watch?v=SZdXegZhI7Y&t=195s) | Captions garbled ("1241 स्टेम 120", "12 ट्रैक … 220"); Whisper hi: «एक दो ट्रेड तो वन इस्टू टैन वन इस्टू ट्वेंटी का मिल ही जाएगा»; Whisper en: "If you have 10 trades, one to 10 trades will get 1 to 10, 1 is to 20" → "1–2 in 10 reach 1:10" in 2022 vs "5 in 10" in 2026. **Owner listen** before use |
| C7 | «$800 के कैपिटल से मैंने तुमको $3100 बुक करके दिखाया» "from $800 capital I booked $3,100 and showed you" | Masterclass 1.0, [160:23](https://www.youtube.com/watch?v=pth7Wl4ZyY8&t=9623s) | Captions — record claim for agent D |

## 1. Sources and weight

| Tier | What | Use |
|---|---|---|
| A — his channels, free | wI9b968AvW8 (the strategy, 1.83M), XHOxOGlaAKw (CHoCH/trap, 1.27M), pth7Wl4ZyY8 (Masterclass 1.0), ival0cWJFhA, ~37 further strategy videos (extracts in `projects/.research/gautam/C_strategy/rules_extract_*.md`) | Rules **and** on-screen quotes |
| B — paid | "GOT" / Game of Trading (Premium/Ultimate, ~40 h; Masterclass 149:18–151:35); Telegram/"community" calls | **Not used** (not bought). He says the two-line strategy is complete in the free video ("bas do hi line tumko draw karne aana chahiye, that's it", 22:06–22:36) |
| C — third parties | "leaked course" re-uploads (ceBHYSJXt30, 8XNqCSiRZg4, qesEcW7b9Ps) | Not used for rules; listed for B |

## 2. The strategy tested — his "two-line" / previous-day liquidity strategy (wI9b968AvW8)

He presents it as "the world's best strategy" for people who lost confidence; demonstrated on gold (VANTAGE:XAUUSD on TradingView,
frame at 15:05 shows "Gold US Dollar · 1D · Vantage", spread 0.28, 21 May 2026 17:52 IST). All quotes: `gautam_rules.json` → `twoline.rules_table`.

| Item | Rule as taught (quote · seconds) | Coded as |
|---|---|---|
| Instrument | «वो यूज़ करने वाले हैं गोल्ड में … BTC … इथेरियम … निफ्टी ससेक्स … आपको थोड़ा बैक टेस्ट करना पड़ेगा» (393–404) | gold headline; NIFTY, BANK NIFTY, BTC, ETH as "your own backtest" |
| Levels | «एक वन डे का टाइम फ्रेम चूज़ करो» (458); «जो प्रीवियस डे का कैंडल था … रनिंग … छोड़ देना है … दो लाइन मार्क करना है ऊपर वाले सिरे पे नीचे वाले सिरे पे» (480–490) | previous day's high and low (BID). Day = New York 17:00 day (TradingView daily candle of broker CFD feeds); variant UTC midnight |
| Timeframe | «टाइम फ्रेम चेंज कर देना है वन डे से 1 मिनट पे» (520–522) | 1-minute candles |
| Inside = no trade | «बीच में रह गया तो देयर इज़ नो ट्रेड। ट्रेड कब बनेगा? जब यह कैंडल इसका बाउंड्री क्रॉस करेगा» (553–557) | setups only after the line has been traded through that day |
| Short trigger | «एक फ्रेश जगह से अगर मार्केट किसी भी बॉडी वाले रेड कैंडल या फिर इवन रेड कैंडल का लो ब्रेक करता है … वहां पे शॉर्ट का ट्रेड लेना है» (598–611) | **our reading of "fresh"**: the red candle (or the one before it) printed the most extreme price since the line was crossed. Strict / generous variants |
| Break timing | «ये वाला ब्रेक होना चाहिए। नेक्स्ट कैंडल में।» (616–618) | the next 1-minute candle must trade below the red candle's low (sell stop); generous variant: any time until invalidated |
| Stop | «स्टॉप लॉस इसके ऊपर रहेगा» (618–620) | the red candle's high (short stop executes on ASK) |
| Target 1 + runner | «जहां से भी मार्केट स्विंग में रिवर्स कर रहा है … पहला टारगेट … 20% क्वांटिटी तुम छोड़ दोगे बड़े टारगेट्स के लिए» (621–633) | **our swing**: 1-minute pivot (3 candles each side, visible at entry) — the most recent one beyond the entry; 80% exits there, stop → entry, 20% to the next swing (or day close). Variants k=5, k=10, no runner |
| Long (mirror) | green candle high break below the low line; TG1 = swing high; «मेजॉरिटी क्वांटिटी एग्जिट» (683–706) | mirror |
| Partial → CTC | «20% 30% रिड्यूस किया क्वांटिटी। एसएल तुम्हारा सीटीसी हो गया … एसएल काउंट नहीं होगा» (812–818); «ट्रेल तो वैसे भी नहीं करना है» (795) | stop to entry after any partial; variant "30% at 1:1 first" |
| Daily limits | «टोटल तुमको दो ट्रेड प्लेस करने हैं» (~683); «दो फुल एसएल» (797–799); «अगर दोनों में प्रॉफिट … एक फाइट लो। तीन से ज्यादा नहीं लेना … दो एसएल पूरा चला गया तब नहीं लेना है» (847–860) | max 3 trades/day, stop after 2 full SLs; all-conditions: max 2 |
| Stop size | «70 80 90 पिप्स का स्टॉप लॉस बन रहा है अवॉयड» (934–937); «140 पिप्स का स्टॉप लॉस लेना नहीं है» (1264–1266) | skip if the candle-based stop > 70 pips ($7) at his price ($4,507) = 0.155% of price (headline); literal $7; none. Other markets: 7% of the 20-day median daily range |
| Fresh again after a stop-out | «मार्केट फ्रेश लिक्विडिटी बना रहा है … पुराना होता तो एसएल हिट नहीं होता … फ्रेश लिक्विडिटी पे ट्रेड बनता है» (~1215–1230) | consistent with the "new extreme" reading |

**Not coded (his words say not to count them):** the "creative" trades he shows but excludes (the 1000-pip move on 20 May,
"micro liquidity", 1085–1145, 1266–1280); his CHoCH formula taught on lives (1100–1112, optional).
**Our assumptions (stated, not his):** one position at a time; positions closed at the day's close (17:00 New York); no entry in the
last 15 minutes of the day; a setup whose stop exceeds the cap is skipped and consumed; if no swing exists, TG1 = 2R (never happened).

**Other strategies found and why they are not the headline test:** XHOxOGlaAKw (change of character at a liquidity level with "double
confirmation", entries "a good short candle", targets "liquidity, liquidity and liquidity") and Masterclass S4 (1m/5m trend, retracement
to a level "drawn in advance", panic point, high must not break) depend on hand-drawn levels and pointing at the screen; Masterclass S3
"event trades" have two conflicting level definitions and no exit rule. They are listed with quotes in
`projects/.research/gautam/C_strategy/rules_extract_masterclass1.md` (§7 lists candidate codifications); the free parameters would be ours,
not his. He gives no numeric claim for them except C6 (30–40% accuracy) and "super high accuracy" for event trades.

**Two more rule sets found in his other videos (`rules_extract_other_videos.md`, 37 of 38 videos read):**
- **SZdXegZhI7Y** "BEST INTRADAY STRATEGY FOR BANKNIFTY" (16 Apr 2022): the same 1-minute mechanics ("stop-loss hunting": a swept
  15-minute stop area, then the first opposite-colour candle's break, stop beyond it) with **his own numbers: stop 25–80 points, max 2
  trades, stop after 2 SL** (1:58, 9:11–10:34, 16:03, 21:18). Used to give the two-line rules his own Bank Nifty stop range (labelled
  variant, added after the first pass).
- **aBaJD8Bnx6w** "SWING TRADING STRATEGY" (24 Mar 2023, "80–90% accuracy"): daily swing-low break → first green close, stop below it
  minus ₹2–3, 50% out at +3%, rest at the previous high. Several definitions are ours ("uptrend", "big candle", pivot size); tested as a
  SECONDARY strategy on today's Nifty 100 (survivorship bias in his favour), pre-registered (`prereg/…v2_swing_bnf_prerun.json`)
  before any stock data was downloaded.
- Everything else is filters, heuristics or discretionary (no full options rule set anywhere: no video gives entry time + strike +
  stop + target together).

## 3. Variants (pre-registered, `prereg/…v1_prerun.json`)

Readings for critics: **EXACT** (the video read literally), **ALL CONDITIONS** (strict fresh, bodied candle ≥ 50% of range, 2 trades a day),
**GENEROUS** (any candle beyond the line, break any time, no cap, no daily limit), **HIS CLAIM 1:10** (exact entries, full position to
10R), plus 1:3 ("my failure trade"), 1:1 (most generous to the win rate), his early-partial option, swing size k=5/10, cap literal/none,
UTC days, shorts-only/longs-only. Every headline also runs at zero cost, zero cost + optimistic intrabar (most generous), real Dukascopy
spread + commission, slippage ×2. No parameter was changed after seeing a result. Post-registration edits: documentation in `claims`
only (Whisper readings), no rule change. **Added after the first pass, labelled `_added_after_first_pass` in the rules file and never
replacing a headline:** (1) only his live-stream hours (21:00–01:00 IST); (2) Bank Nifty with his own 2022 stop range 25–80 points
and 2 trades a day (swing targets / fixed 1:3); (3) **Trading With Sidhant's reading** ("the red candle itself must be the highest
candle", UdGXYHp1nYA ~1263–1268, ~1605–1612; his "rule, mistake or oversight?" 1447.7–1452.9) and **his commenter's reading** (any
first red/green candle beyond the line, broken by the immediate next candle; @kr4T0X) — both run with zero-cost / most-generous
twins (`prereg/…v3_readings_sidhant_commenter_prerun.json`); (4) the swing strategy (v2 snapshot).

## 4. Data

- **Gold XAUUSD (his market):** Dukascopy BID+ASK 1-minute, 2015-01-01 → 2026-09-30, 4.16M active minutes (built for topg-audit by
  `tools/fx_data.py`, re-used in place). Second source: HistData BID 1-minute (same period).
- **NIFTY 50 / NIFTY BANK spot 1-minute,** 2015-01-09 → 2026-05-15 (Kaggle debashis74017; NSE is geo-blocked here), re-used from
  stockburner-audit.
- **BTCUSDT / ETHUSDT 1-minute,** Binance spot (data.binance.vision monthly zips), 2020-01 → 2026-09, as the price proxy for the crypto
  perpetuals he trades live (`gautam_backtest.py binance`; ~0.45 GB in `projects/gautam-audit/data_raw/binance/`).

## 5. Execution and costs

Signals on the BID chart (gold) / spot chart. Entries are stop orders at the candle's low/high: triggered when the BID trades through,
filled at the level or a worse open; longs pay the spread (ASK) + slippage, shorts fill at BID − slippage. Exits on the executable side
(long on BID, short on ASK); stop first in a shared minute; a stop touch in the entry minute is a loss; a target in the entry minute counts
only if the minute closed beyond it. **Gold costs (headline, generous to him):** the 0.28 spread **shown on his own screen** + $0.05/oz
slippage on stop fills, no commission. Sensitivity: Dukascopy's real spread (median 0.24–0.69 by year) + $0.07/oz commission; slippage ×2;
zero. **NIFTY/BANK NIFTY:** statutory option charges by date + 0.5/1.0 premium-point slippage per side at delta 0.5 + ₹20/order (stockburner
model). **Crypto:** Delta Exchange taker 0.05% + 18% GST + 0.01% slippage per side. Money: fixed 1% of the start per trade ($100 of
$10,000; ₹1,000 of ₹1,00,000), no compounding.

## 6. What we measure

Per variant/market/year: trades, **win rate (net R > 0) with 95% Wilson CI**, share of trades whose favourable move ever reaches **10R
before the stop** (his "1:10"), first-target hit rate, full-SL rate, avg R (95% CI), PF, total R, longest losing run, green days, money
at fixed risk, results by stop size (quartiles), and the period after he recorded (22 May 2026) / published (7 Jun 2026) the video.

## 7. His premises, measured directly

- **P1 "the previous day's high/low is liquidity: after it is taken, big players reverse it"**: after the first trade beyond the line, does
  price reach the previous day's midpoint before it extends the same distance beyond the line (same day)?
- **P2 "50% at 1:10", "average 1:6–7", "1:3 is my failure trade"**: share of all entries whose move ever reaches 3R / 10R / 67R.
- **P3 his own demo days (19–21 May 2026)**: what the exact rules do on the days he showed.

## 8. Critics' arguments, answered by design

1. *"Wrong market / timeframe"* → gold, 1-minute, his broker's daily candle, exactly as on his screen; plus NIFTY/BANK NIFTY/BTC/ETH.
2. *"You missed his conditions"* → all-conditions variant; list of what isn't coded and why.
3. *"Costs are made up"* → his own screen's spread; zero-cost and most-generous runs everywhere.
4. *"He takes partials / runners"* → coded as he says (80/20, CTC), plus his 30%-at-1:1 option and fixed 1:10 / 1:3 / 1:1 targets.
5. *"Tiny stops are unrealistic"* → results by stop-size quartile and the stop cap variants.
6. *"It worked in 2026"* → per year and the period after his video.
7. *"Trading With Sidhant already backtested it and said the rules are unclear"* (UdGXYHp1nYA, 248K) → his reading and the
   opposite reading both tested; he showed a handful of days without a total; his cost point ("1:7 but 81 pips… brokerage") is
   answered with costs in pips (BACKTEST.md §2).
8. *"Show the data"* → every trade in `data_raw/trades/` and `engine/public/projects/gautam-audit/data/trades/`, rules, code, prereg copy,
   second data source, 5 hand spot checks.

## 9. Spot checks (5 random trades, seed 11, recomputed by hand from the printed minutes — `data_raw/spotcheck_exact.txt`)

2023-09-27 00:11 NY long (PDL 1899.005; setup 00:10 green at the new low 1898.575; entry = 1898.715 high + 0.28 + 0.05 = 1899.045; stop
1898.575 hit 00:12 at 1898.525 → −1.106R); 2023-09-28 10:33 long (previous candle printed the extreme 1869.218; entry 1871.635, stop
1870.505 touched 10:35 → −1.044R); 2023-12-27 20:56 short (setup 20:55 red at the new high 2087.645; entry 2087.005; ASK reached the stop
at 21:01 (2087.415 + 0.28) → −1.078R; max favourable 0.95R); 2024-10-15 01:11 long (stop touched inside the entry minute → loss,
−1.082R); 2025-09-14 21:02 long (stop $5.44 under the $5.65 cap; never traded above entry; −1.009R). All match the engine to the cent.
Swing (2nd strategy, seed 5): SIEMENS 2021 (pivot low 1,620.10 on 29 Oct confirmed 3 Nov; broken 25 Nov; 29 Nov green candle skipped as
"big" (range 109.15 > 1.5 × ATR 64.65); entry 2 Dec close 1,672.90 + slippage = 1,673.74; stop 1,626.50 − 1% = 1,610.24; half at +3%
on 9 Dec, half at the previous high 1,857.25 on 10 Dec → +1.774R) and CANBK 2019 (gap fill of the +3% half at the 26 Aug open 40.72,
rest stopped 3 Sep at 36.57 → −0.034R) recompute exactly.
His 20 May 2026 demo long (his chart: line 4,464.65, stop 4,460.07, target 4,473.52) is reproduced by the engine: entry 4,464.67,
stop 4,459.66, first swing 4,476.81.

## 10. Results (final; full tables in `BACKTEST.md`)

| Test | Trades | His number | Measured |
|---|---|---|---|
| **EXACT, gold 1-min, 2015-01 → 2026-09** (his market, his screen's spread) | 5,041 (8.2/week) | 50% accuracy at 1:10 | **win 19.7%** (18.7–20.9); **only 4.4% of trades ever reach 1:10** before the stop (3.9–5.0); avg −0.476R after costs; **0 of 12 years** profitable; 80% of trades end at the full stop; longest losing run 31; $10k at $100/trade: blown by Jul 2015 |
| same, zero cost | 5,234 | — | win 26.4%, 1:10 reached 9.1%, avg +0.072R (CI +0.009…+0.134) — the pre-cost edge sits in the smallest stops (≤ $0.41: +0.26R); the quarter with the biggest stops is −0.036R |
| same, the quarter of trades with the biggest stops ($1.57–8.16) | 1,256 | — | win 28.6%, 1:10 reached 5.2%, avg −0.178R |
| **HIS CLAIM 1:10** (exact entries, full position to 10R) | 4,610 | 50% | **7.8% win**, 10R hit 4.4%; avg −0.433R; longest losing run **79** |
| ALL CONDITIONS / GENEROUS | 2,406 / 112,450 | — | win 24.4% / 17.6%; 1:10 5.1% / 2.7%; avg −0.407R / −0.698R |
| His early partial (30% at 1:1) / fixed 1:1 / fixed 1:3 | 5,376 / 5,331 / 4,888 | — | win 30.8% / 29.5% / 15.3%; all −0.46…−0.47R |
| Only his live-stream hours, 21:00–01:00 IST (added after the first pass, labelled) | 1,518 | — | win 20.5%, 1:10 reached 1.9%, avg −0.458R |
| **After he recorded the video** (22 May → 30 Sep 2026, out-of-sample) | 169 | 50% at 1:10 | win 26.0%, 1:10 reached 4.1%, avg −0.177R, green days 29.5% |
| NIFTY / BANK NIFTY exact (option costs) 2015 → May 2026 | 4,091 / 3,834 | — | win 20.7% / 22.1%; 1:10 9.1% / 8.7%; avg −0.491R / −0.272R; 0 / 1 of 12 years |
| BTC / ETH exact (Delta fees) 2020 → Sep 2026 | 3,249 / 3,218 | — | win 25.0% / 26.4%; 1:10 10.2% / 9.9%; median stop 0.14% / 0.20% of price, fees ≈ 96% / 69% of the stop; biggest-stop quarter −0.62R / −0.48R; 0 of 7 years |
| Premise P1 "liquidity taken → reversal" | 2,726 gold days (5 markets) | — | after the line is taken, price extends the same distance further **before** returning to the previous day's midpoint more often: gold 52.7% vs 47.3%; NIFTY 56.5/43.5; BANK NIFTY 53.3/46.7; BTC 56.2/43.8; ETH 56.4/43.6 |
| Premise P2 reach 3R / 10R / 67R (gold exact) | 5,041 | 1:10 half the time; avg 1:6–7 | 14.7% / 4.4% / 0.02%; median first target 2.67R; avg winning trade +1.95R |
| His Masterclass admission (C6) | — | 30–40% accuracy for small-TF liquidity entries | measured win 19.7% (26.4% before costs) |
| His demo days 19–21 May 2026 | 6 trades | — | the exact rules took 6 trades: 2 stops, 4 winners (+1.9…+5.6R, two moves ran past 50R) = **+12.6R in 3 days, a top-0.26% three-day stretch of 3,031 since 2015** (top 1.55% in 2026); only 18.7% of all three-day windows are positive (`gautam_backtest.py demorank` → `gautam_demo_rank.json`: daily R of the exact rules over every NY trading day, no-trade days = 0, rolling 3). 2026 as a whole: win 24.5%, avg −0.193R |
| Second source (HistData, zero cost) | 5,226 vs 5,234 | — | win 26.4% vs 26.4%; 86% identical entry minutes |
| **Sidhant's reading / commenter's reading** (gold) | 4,105 / 5,587 | 50% at 1:10 | win 21.0% / 19.3%; 1:10 reached 4.4% / 5.0%; avg −0.447R / −0.467R (zero cost +0.086R / +0.083R) — the disputed rule does not change the answer |
| His Telegram / group claims (C8, C9) | — | >50% / 60–70% at 1:10 | the strategy he teaches publicly reaches 1:10 in 4.4% of trades; his calls themselves → agent D |
| **Bank Nifty with his own 2022 stop range (25–80 pts), 2 trades a day** | 2,405 | — | win 28.5% (26.8–30.4), 1:10 reached 7.1%, avg −0.059R (CI −0.152…+0.035, i.e. not distinguishable from break-even), 4 of 12 years positive; fixed 1:3: win 26.4%, −0.146R |
| **2nd strategy: 2023 daily swing (aBaJD8Bnx6w), exact** — Nifty 100 daily (Kaggle 15-min → daily), Feb 2015 → Aug 2025 | 6,966 | "80–90% accuracy" | **+3% reached before the stop 45.7%** (44.6–46.9); closed in profit 44.0%; previous high reached 40.3%; avg −0.020R (CI −0.050…+0.010) ≈ break-even after costs (+0.103R at zero cost); **0 of 11 years at ≥ 80%** (best year 2020: 52.3%) |
| swing, all conditions (uptrend) / bigger swings / close-break | 3,108 / 4,812 / 5,671 | 80–90% | +3% first 46.3% / 49.0% / 48.1%; avg +0.001R / +0.011R / −0.028R |
| swing on Yahoo (30 stocks incl. his examples, to Sep 2026) | 2,360 | 80–90% | +3% first 42.2%, in profit 43.7%, −0.060R; Exide (his example): 83 trades, +3% first 50.6%, in profit 41.0% |
| Costs in pips (Sidhant's "81 pips" point) | — | — | median stop 9.6 pips; spread 2.8 + slippage ≈ 3.8 pips per round trip = 39.6% of the median stop |

## 11. Not testable / needs the owner

- Owner listen: C2 (6:21 "1:6–7" per Whisper vs caption "1:67"). C1 is verified 3-way.
- His Telegram/"community" calls and paid GOT content (not bought).
- His live trades (agent D: "Crypto Gold Live Trading" streams 14 Sep → 8 Oct 2026).
- Vantage's own price feed (Dukascopy differs by ~$0.3–0.5 at times; the 20 May demo trade still matches).


---

# Results

# Backtest: Gautam Jha's "two-line" liquidity strategy (rules as he taught them)

Generated by `tools/gautam_backtest.py report` from `engine/public/projects/gautam-audit/data/*.json` (2026-10-10 19:53 UTC). Rules, quotes and every assumption: `projects/gautam-audit/BACKTEST_PLAN.md` and `gautam_rules.json` (pre-registered copy in `prereg/`). **FINAL** (marked 11 Oct 2026 after the cross-checks below; no numbers changed).

His claim (wI9b968AvW8, 14:22–14:40): out of 10 trades "four, five, six" work; "this 50% accuracy is at 1:10 risk-reward". If that were true the strategy would make about +4.5R per trade (0.5 × 10 − 0.5 × 1).

### His numbers vs what his rules do (gold, 1-minute, 2015 → Sep 2026)

| His claim | Where | Measured: share of trades that EVER reach 1:10 before the stop | Measured: trades that win at all |
|---|---|---|---|
| "yeh 50% ka accuracy 1:10 ke risk reward pe hai" (this strategy) | wI9b968AvW8 14:33 (verified 3-way) | **4.4%** [3.9–5.0] of 5,041 | 19.7% (full position to 10R: 7.8%) |
| "dono mere Telegram ka accuracy more than 50% hai with 1:10" (his Telegram calls) | live 14 Sep 2026, 0stizzV7D1Y 4344.5–4375 (agent B/D) | 4.4% for the strategy he teaches publicly; his Telegram calls themselves: agent D | — |
| "group ka accuracy more than 60–70% hai with 1:10… phenomenal" (his group) | live 17 Sep 2026, e4lZ5QeeZ6E 37:19 (agent B/D) | same | — |

Reading his 1:10 claim generously: "accuracy at 1:10" = the trade's favourable move reaches ten times the stop distance before the stop is hit, measured on every minute until the day's close, ignoring costs in the fills except his own screen's spread. Even with ZERO costs it is about 9% (cost table below).

## 1. Headline: his exact rules, his market (gold, 1-minute), 2015 → Sep 2026

| Market | Data | Trades | Win rate (net R > 0) [95% CI] | Trades that ever reach 1:10 before the stop [95% CI] | Avg R after costs [95% CI] | Avg R, the quarter of trades with the BIGGEST stops | Avg R at ZERO cost (all / biggest-stop quarter) | Years with a profit (after costs) |
|---|---|---|---|---|---|---|---|---|
| XAUUSD | dukascopy 2015-01-01 .. 2026-09-30 | 5,041 | 19.7% [18.7–20.9] | 4.4% [3.9–5.0] | -0.476R [-0.515, -0.437] | -0.178R | +0.072R / -0.036R | 0 of 12 |
| NIFTY | kaggle 2015-01-09 .. 2026-05-15 | 4,091 | 20.7% [19.5–22.0] | 9.1% [8.2–10.0] | -0.491R [-0.568, -0.414] | -0.258R | +0.055R / +0.018R | 0 of 12 |
| BANKNIFTY | kaggle 2015-01-09 .. 2026-05-15 | 3,834 | 22.1% [20.8–23.4] | 8.7% [7.9–9.6] | -0.272R [-0.351, -0.193] | -0.129R | +0.078R / +0.067R | 1 of 12 |
| BTCUSDT | binance 2020-01-01 .. 2026-09-30 | 3,249 | 25.0% [23.5–26.5] | 10.2% [9.2–11.3] | -10.692R [-23.878, +2.493] | -0.617R | +0.212R / +0.058R | 0 of 7 |
| ETHUSDT | binance 2020-01-01 .. 2026-09-30 | 3,218 | 26.4% [24.9–28.0] | 9.9% [8.9–10.9] | -0.746R [-0.836, -0.657] | -0.478R | +0.151R / +0.009R | 0 of 7 |

Costs: gold = the 0.28 spread on HIS OWN SCREEN + $0.05 slippage on stop fills, no commission; NIFTY/BANK NIFTY = statutory option charges by date + 0.5/1.0 premium-point slippage (delta 0.5); crypto = Delta Exchange taker fee + GST + 0.01% slippage. Exact rules for indices/crypto use a stop cap of 7% of the 20-day median daily range (his 70 pips was 7% of gold's daily range when he said it).

## 2. Gold (XAUUSD) — every reading of his rules

| Variant | Trades | /week | Win rate | Reach TG1 | Reach 1:10 | Full SL | Avg R | PF | Total R | Longest losing run | $10k at $100/trade |
|---|---|---|---|---|---|---|---|---|---|---|---|
| EXACT: fresh candle beyond the line, next-candle break, SL its extreme, 80% at last swing + 20% runner, <=3 trades / 2 full SL, SL <= 70 pips (price-scaled) | 5,041 | 8.23 | 19.7% | 19.3% | 4.4% | 80.2% | -0.476R | 0.447 | -2,399.0 | 31 | blown 2015-07-02 |
| HIS CLAIM 1:10: exact entries, fixed 10R target, full position | 4,610 | 7.52 | 7.8% | 4.4% | 4.4% | 92.1% | -0.433R | 0.561 | -1,998.1 | 79 | blown 2015-10-09 |
| GENEROUS: any candle beyond the line, break any time, no SL cap, no daily limits, swing targets | 112,450 | 183.48 | 17.6% | 17.7% | 2.7% | 82.0% | -0.698R | 0.244 | -78,471.3 | 364 | blown 2015-01-06 |
| ALL CONDITIONS: strict fresh, bodied candle (>=50%), 2 trades a day, 2 full SL, SL <= 70 pips (price-scaled) | 2,406 | 3.93 | 24.4% | 24.0% | 5.1% | 75.4% | -0.407R | 0.495 | -980.0 | 34 | blown 2016-02-15 |
| EXACT + his early partial: 30% at 1:1, 50% at the swing, 20% runner | 5,376 | 8.77 | 30.8% | 19.5% | 4.4% | 69.0% | -0.469R | 0.369 | -2,522.7 | 24 | blown 2015-06-24 |
| EXACT, SL cap literally $7 (70 pips) every year | 5,144 | 8.39 | 20.6% | 20.1% | 4.3% | 79.4% | -0.455R | 0.465 | -2,342.7 | 31 | blown 2015-07-23 |
| EXACT, no SL cap | 5,206 | 8.49 | 20.9% | 20.4% | 4.2% | 79.0% | -0.453R | 0.465 | -2,358.4 | 31 | blown 2015-07-23 |
| EXACT, bigger swings (k=5) | 4,934 | 8.05 | 16.7% | 16.2% | 4.4% | 83.2% | -0.476R | 0.467 | -2,348.7 | 36 | blown 2015-07-05 |
| EXACT, biggest swings (k=10) | 4,842 | 7.9 | 13.9% | 13.0% | 4.4% | 86.0% | -0.448R | 0.514 | -2,169.1 | 41 | blown 2015-07-05 |
| EXACT, all at the first swing (no runner) | 5,041 | 8.23 | 19.8% | 19.3% | 4.4% | 80.2% | -0.479R | 0.443 | -2,415.9 | 31 | blown 2015-07-02 |
| EXACT entries, fixed 1:3 target ('my failure trade') | 4,888 | 7.98 | 15.3% | 14.8% | 4.5% | 84.6% | -0.460R | 0.493 | -2,249.7 | 55 | blown 2015-07-07 |
| EXACT entries, fixed 1:1 target (most generous to the win rate) | 5,331 | 8.7 | 29.5% | 29.5% | 4.5% | 70.5% | -0.465R | 0.388 | -2,476.8 | 28 | blown 2015-06-10 |
| EXACT, UTC-midnight daily lines | 5,645 | 9.21 | 19.0% | 18.5% | 4.1% | 80.8% | -0.514R | 0.411 | -2,898.7 | 40 | blown 2015-05-05 |
| EXACT, shorts only (high line) | 2,852 | 4.65 | 13.7% | 13.4% | 3.6% | 86.2% | -0.616R | 0.344 | -1,756.7 | 70 | blown 2015-10-23 |
| EXACT, longs only (low line) | 2,473 | 4.04 | 27.0% | 26.3% | 5.2% | 72.8% | -0.298R | 0.609 | -737.7 | 21 | blown 2015-12-30 |
| EXACT, only his live-stream hours (21:00-01:00 IST) | 1,518 | 2.48 | 20.5% | 19.4% | 1.9% | 79.4% | -0.458R | 0.46 | -695.2 | 32 | blown 2016-10-03 |
| SIDHANT'S READING: the red/green candle itself must be the highest/lowest candle since the line was taken (rest exact) | 4,105 | 6.7 | 21.0% | 20.6% | 4.4% | 78.8% | -0.447R | 0.471 | -1,836.0 | 45 | blown 2015-11-04 |
| COMMENTER'S READING: any red/green candle beyond the line; only the immediate next candle may break it (rest exact) | 5,587 | 9.12 | 19.3% | 19.1% | 5.0% | 80.6% | -0.467R | 0.462 | -2,611.0 | 34 | blown 2015-06-21 |

Cost sensitivity on the headline readings (same trades logic, different cost / fill assumptions):

| Variant | Costs as on his screen | Zero cost | Zero cost + optimistic intrabar (most generous) | Dukascopy real spread + $0.07 commission | Slippage ×2 |
|---|---|---|---|---|---|
| EXACT: fresh candle beyond the line, next-candle break, SL i | win 19.7%, 1:10 4.4%, -0.476R | win 26.4%, 1:10 9.1%, +0.072R | win 26.5%, 1:10 9.1%, +0.076R | win 18.0%, 1:10 3.7%, -0.637R | win 19.6%, 1:10 4.1%, -0.547R |
| HIS CLAIM 1:10: exact entries, fixed 10R target, full positi | win 7.8%, 1:10 4.4%, -0.433R | win 12.1%, 1:10 9.0%, +0.132R | win 12.2%, 1:10 8.9%, +0.134R | win 7.0%, 1:10 3.7%, -0.609R | win 7.7%, 1:10 4.2%, -0.506R |
| GENEROUS: any candle beyond the line, break any time, no SL  | win 17.6%, 1:10 2.7%, -0.698R | win 33.8%, 1:10 8.4%, +0.049R | win 34.3%, 1:10 8.4%, +0.063R | win 14.6%, 1:10 2.2%, -0.922R | win 17.2%, 1:10 2.6%, -0.787R |
| SIDHANT'S READING: the red/green candle itself must be the h | win 21.0%, 1:10 4.4%, -0.447R | win 28.1%, 1:10 9.2%, +0.086R | win 28.2%, 1:10 9.2%, +0.087R | win 18.8%, 1:10 3.6%, -0.620R | win 20.9%, 1:10 4.2%, -0.517R |
| COMMENTER'S READING: any red/green candle beyond the line; o | win 19.3%, 1:10 5.0%, -0.467R | win 25.8%, 1:10 9.3%, +0.083R | win 26.1%, 1:10 9.3%, +0.090R | win 17.7%, 1:10 4.1%, -0.638R | win 19.2%, 1:10 4.7%, -0.542R |

Headline per year (EXACT: fresh candle beyond the line, nex…):

| Year | Trades | Win rate | Reach 1:10 | Avg R | Total R |
|---|---|---|---|---|---|
| 2015 | 373 | 16.6% | 3.5% | -0.563 | -210.1 |
| 2016 | 406 | 20.7% | 4.9% | -0.423 | -171.9 |
| 2017 | 424 | 16.3% | 2.4% | -0.676 | -286.8 |
| 2018 | 431 | 16.7% | 3.5% | -0.649 | -279.8 |
| 2019 | 410 | 15.9% | 3.2% | -0.673 | -276.1 |
| 2020 | 424 | 19.3% | 3.5% | -0.505 | -214.0 |
| 2021 | 444 | 19.4% | 3.4% | -0.467 | -207.5 |
| 2022 | 438 | 19.4% | 4.6% | -0.458 | -200.7 |
| 2023 | 468 | 24.1% | 5.8% | -0.412 | -192.8 |
| 2024 | 439 | 22.6% | 5.9% | -0.368 | -161.7 |
| 2025 | 466 | 21.5% | 6.7% | -0.292 | -136.1 |
| 2026 | 318 | 24.5% | 5.3% | -0.193 | -61.4 |

Years with win rate ≥ 50%: none; years where ≥ 50% of trades reach 1:10: none; profitable years after costs: none.

**Costs in plain pips (answers Trading With Sidhant's "1:7 bana hai lekin 81 pips hai… brokerage to poora lagega", UdGXYHp1nYA 1743–1752):** gold pip = $0.10. The stops these rules give are small because they sit on one 1-minute candle: median 9.6 pips. Every trade pays the spread (2.8 pips on his own screen) plus slippage on the entry and on a stop exit (0.5 pip each) — about 3.8 pips per round trip, a median 39.6% of the stop. So a 1:7 that is only ~80 pips loses a big slice to costs, and a full stop costs more than 1R. With Dukascopy's real 2025–26 spreads (5.8–6.9 pips) it is worse; with Indian-exchange crypto fees (Delta: 0.05% + GST per side) BTC costs are ~96% of the stop (BTC section).

By stop size (quartiles of the stop distance in $/oz) — answers "nobody uses such tiny stops":

| Stop size | Trades | Win rate | Reach 1:10 | Avg R | Costs as % of the stop (median) | Avg R at zero cost (same quartile cut, zero-cost run) |
|---|---|---|---|---|---|---|
| $0.05–0.59 | 1263 | 6.6% | 2.3% | -0.914 | 88.4% | +0.256R |
| $0.59–0.96 | 1260 | 18.8% | 4.1% | -0.534 | 50.2% | +0.021R |
| $0.96–1.57 | 1262 | 25.0% | 6.0% | -0.276 | 31.3% | +0.046R |
| $1.57–8.16 | 1256 | 28.6% | 5.2% | -0.178 | 16.2% | -0.036R |

After he recorded / published the video (out-of-sample for him):

| Period | Trades | Win rate | Reach 1:10 | Avg R | Total R | Green days |
|---|---|---|---|---|---|---|
| EXACT: fresh candle beyond t… since_he_recorded_2026-05-22 | 169 | 26.0% | 4.1% | -0.177 | -30.0 | 29.5% |
| EXACT: fresh candle beyond t… since_upload_2026-06-07 | 148 | 26.4% | 4.7% | -0.164 | -24.2 | 30.4% |
| EXACT: fresh candle beyond t… last_12_months_2025-10-01 | 432 | 24.1% | 6.5% | -0.183 | -79.2 | 27.5% |
| HIS CLAIM 1:10: exact entrie… since_he_recorded_2026-05-22 | 152 | 6.6% | 3.3% | -0.526 | -80.0 | 10.3% |
| HIS CLAIM 1:10: exact entrie… since_upload_2026-06-07 | 134 | 6.7% | 3.7% | -0.508 | -68.1 | 10.1% |
| HIS CLAIM 1:10: exact entrie… last_12_months_2025-10-01 | 390 | 7.7% | 5.9% | -0.304 | -118.6 | 12.7% |

## NIFTY (kaggle, 2015-01-09 .. 2026-05-15) — he says use it here only after your own backtest

| Variant | Trades | Win rate | Reach 1:10 | Full SL | Avg R | Avg R zero cost | PF | Total R | Profitable years |
|---|---|---|---|---|---|---|---|---|---|
| EXACT (non-gold): as exact, SL cap = 7% of the 20-day median daily range | 4,091 | 20.7% | 9.1% | 78.8% | -0.491R | +0.055R | 0.598 | -2,009.6 | 0/12 |
| HIS CLAIM 1:10: exact entries, fixed 10R target, full position | 3,822 | 11.1% | 9.3% | 88.8% | -0.422R | +0.119R | 0.691 | -1,614.0 | 1/12 |
| GENEROUS: any candle beyond the line, break any time, no SL cap, no daily limits, swing targets | 36,616 | 27.2% | 8.5% | 68.1% | -0.523R | +0.077R | 0.532 | -19,165.4 | 0/12 |
| EXACT entries, fixed 1:3 target ('my failure trade') | 4,215 | 26.3% | 9.2% | 73.7% | -0.497R | — | 0.561 | -2,096.0 | 0/12 |
| EXACT entries, fixed 1:1 target (most generous to the win rate) | 4,741 | 47.7% | 9.4% | 47.9% | -0.513R | — | 0.316 | -2,434.4 | 0/12 |

Exact rules by stop size (median stop = 0.039% of price; median costs = 47.6% of the stop):

| Stop size (price units) | Trades | Win rate | Reach 1:10 | Avg R | Costs as % of the stop | Avg R at zero cost |
|---|---|---|---|---|---|---|
| 0.849–3.550 | 1027 | 21.0% | 11.1% | -0.656 | 84.9% | +0.286R |
| 3.550–5.250 | 1040 | 18.9% | 8.8% | -0.634 | 55.2% | -0.076R |
| 5.250–7.800 | 1002 | 21.9% | 8.3% | -0.412 | 40.3% | -0.005R |
| 7.800–35.450 | 1022 | 21.1% | 8.1% | -0.258 | 28.2% | +0.018R |

- last_12_months_2025-10-01: 234 trades, win 14.5%, reach 1:10 7.3%, avg -0.669R, green days 17.8%

## BANKNIFTY (kaggle, 2015-01-09 .. 2026-05-15) — he says use it here only after your own backtest

| Variant | Trades | Win rate | Reach 1:10 | Full SL | Avg R | Avg R zero cost | PF | Total R | Profitable years |
|---|---|---|---|---|---|---|---|---|---|
| EXACT (non-gold): as exact, SL cap = 7% of the 20-day median daily range | 3,834 | 22.1% | 8.7% | 77.5% | -0.272R | +0.078R | 0.741 | -1,042.2 | 1/12 |
| HIS CLAIM 1:10: exact entries, fixed 10R target, full position | 3,569 | 11.0% | 8.6% | 88.8% | -0.281R | +0.066R | 0.766 | -1,001.8 | 2/12 |
| GENEROUS: any candle beyond the line, break any time, no SL cap, no daily limits, swing targets | 34,731 | 28.9% | 7.3% | 67.6% | -0.344R | +0.032R | 0.637 | -11,940.0 | 0/12 |
| EXACT entries, fixed 1:3 target ('my failure trade') | 3,906 | 25.4% | 8.7% | 74.5% | -0.335R | — | 0.667 | -1,307.7 | 0/12 |
| EXACT entries, fixed 1:1 target (most generous to the win rate) | 4,402 | 50.0% | 8.6% | 49.7% | -0.349R | — | 0.483 | -1,534.5 | 0/12 |
| BANK NIFTY with his own 2022 stop numbers: SL 25-80 points, max 2 trades, stop after 2 SL | 2,405 | 28.5% | 7.1% | 70.7% | -0.059R | — | 0.93 | -141.6 | 4/12 |
| BANK NIFTY 2022 stop numbers (25-80 pts), max 2 trades, fixed 1:3 | 2,406 | 26.4% | 7.1% | 73.4% | -0.146R | — | 0.832 | -350.7 | 1/12 |

Exact rules by stop size (median stop = 0.056% of price; median costs = 31.3% of the stop):

| Stop size (price units) | Trades | Win rate | Reach 1:10 | Avg R | Costs as % of the stop | Avg R at zero cost |
|---|---|---|---|---|---|---|
| 2.899–12.050 | 959 | 22.0% | 9.6% | -0.445 | 52.7% | +0.121R |
| 12.050–17.950 | 960 | 22.1% | 8.6% | -0.314 | 36.1% | +0.049R |
| 17.950–25.837 | 956 | 22.9% | 8.9% | -0.199 | 27.0% | +0.074R |
| 25.837–93.950 | 959 | 21.4% | 7.7% | -0.129 | 19.9% | +0.067R |

- last_12_months_2025-10-01: 206 trades, win 12.1%, reach 1:10 3.9%, avg -0.716R, green days 18.6%

## BTCUSDT (binance, 2020-01-01 .. 2026-09-30) — he says use it here only after your own backtest

| Variant | Trades | Win rate | Reach 1:10 | Full SL | Avg R | Avg R zero cost | PF | Total R | Profitable years |
|---|---|---|---|---|---|---|---|---|---|
| EXACT (non-gold): as exact, SL cap = 7% of the 20-day median daily range | 3,249 | 25.0% | 10.2% | 70.5% | -10.692R | +0.212R | 0.05 | -34,739.8 | 0/7 |
| HIS CLAIM 1:10: exact entries, fixed 10R target, full position | 2,958 | 13.3% | 10.4% | 86.3% | -11.528R | +0.301R | 0.077 | -34,100.7 | 0/7 |
| GENEROUS: any candle beyond the line, break any time, no SL cap, no daily limits, swing targets | 64,921 | 19.4% | 10.2% | 63.5% | -67.146R | +1.105R | 0.006 | -4,359,212.9 | 0/7 |
| EXACT entries, fixed 1:3 target ('my failure trade') | 3,248 | 25.9% | 10.3% | 70.6% | -11.072R | — | 0.042 | -35,963.3 | 0/7 |
| EXACT entries, fixed 1:1 target (most generous to the win rate) | 3,568 | 26.3% | 10.5% | 47.7% | -10.302R | — | 0.008 | -36,758.0 | 0/7 |

Exact rules by stop size (median stop = 0.144% of price; median costs = 95.7% of the stop):

| Stop size (price units) | Trades | Win rate | Reach 1:10 | Avg R | Costs as % of the stop | Avg R at zero cost |
|---|---|---|---|---|---|---|
| 0.009–36.850 | 813 | 22.3% | 11.8% | -40.072 | 138.8% | +0.404R |
| 36.850–68.800 | 812 | 23.9% | 11.5% | -1.143 | 117.1% | +0.267R |
| 68.800–110.050 | 812 | 24.6% | 8.5% | -0.901 | 94.9% | +0.121R |
| 110.050–315.750 | 812 | 29.1% | 9.1% | -0.617 | 64.2% | +0.058R |

- since_he_recorded_2026-05-22: 218 trades, win 25.2%, reach 1:10 9.6%, avg -1.706R, green days 23.0%
- since_upload_2026-06-07: 194 trades, win 25.3%, reach 1:10 9.3%, avg -1.720R, green days 21.6%
- last_12_months_2025-10-01: 573 trades, win 23.2%, reach 1:10 9.9%, avg -1.393R, green days 21.8%

## ETHUSDT (binance, 2020-01-01 .. 2026-09-30) — he says use it here only after your own backtest

| Variant | Trades | Win rate | Reach 1:10 | Full SL | Avg R | Avg R zero cost | PF | Total R | Profitable years |
|---|---|---|---|---|---|---|---|---|---|
| EXACT (non-gold): as exact, SL cap = 7% of the 20-day median daily range | 3,218 | 26.4% | 9.9% | 71.1% | -0.746R | +0.151R | 0.457 | -2,402.1 | 0/7 |
| HIS CLAIM 1:10: exact entries, fixed 10R target, full position | 2,936 | 12.6% | 9.7% | 87.1% | -0.665R | +0.228R | 0.594 | -1,952.7 | 0/7 |
| GENEROUS: any candle beyond the line, break any time, no SL cap, no daily limits, swing targets | 63,516 | 22.5% | 9.7% | 64.1% | -2.251R | +0.194R | 0.155 | -142,948.3 | 0/7 |
| EXACT entries, fixed 1:3 target ('my failure trade') | 3,196 | 26.7% | 9.7% | 72.4% | -0.806R | — | 0.415 | -2,576.5 | 0/7 |
| EXACT entries, fixed 1:1 target (most generous to the win rate) | 3,554 | 37.3% | 9.9% | 48.4% | -0.864R | — | 0.147 | -3,070.6 | 0/7 |

Exact rules by stop size (median stop = 0.199% of price; median costs = 69.4% of the stop):

| Stop size (price units) | Trades | Win rate | Reach 1:10 | Avg R | Costs as % of the stop | Avg R at zero cost |
|---|---|---|---|---|---|---|
| 0.029–2.440 | 806 | 24.3% | 11.8% | -1.184 | 108.4% | +0.293R |
| 2.440–4.290 | 804 | 26.5% | 10.1% | -0.713 | 88.8% | +0.235R |
| 4.290–6.687 | 803 | 26.3% | 9.7% | -0.609 | 66.2% | +0.067R |
| 6.687–24.210 | 805 | 28.7% | 7.8% | -0.478 | 48.0% | +0.009R |

- since_he_recorded_2026-05-22: 174 trades, win 25.9%, reach 1:10 6.9%, avg -0.863R, green days 24.1%
- since_upload_2026-06-07: 153 trades, win 26.8%, reach 1:10 7.2%, avg -0.865R, green days 24.7%
- last_12_months_2025-10-01: 475 trades, win 25.9%, reach 1:10 7.4%, avg -0.806R, green days 26.3%

## Premises (his ideas measured directly)

P1 "the previous day's high/low is liquidity: once it is taken, big players reverse the market". After the first trade beyond the line, which comes first the same day — back to the previous day's midpoint, or the same distance further beyond the line? (A coin flip would be ~50/50.)

| Market | Days with a line taken | Reversed to the midpoint first | Extended first | Neither by day end | Day closed back inside the lines |
|---|---|---|---|---|---|
| XAUUSD | 2726 | 47.3% | 52.7% | 31.6% | 50.4% |
| NIFTY | 1342 | 43.5% | 56.5% | 39.3% | 44.3% |
| BANKNIFTY | 1372 | 46.7% | 53.3% | 37.2% | 47.4% |
| BTCUSDT | 2170 | 43.8% | 56.2% | 29.7% | 51.9% |
| ETHUSDT | 2157 | 43.6% | 56.4% | 32.3% | 51.6% |

P2 "50% accuracy at 1:10" / "my average trade is 1:6–7" / "1:3 is my failure trade": share of ALL entries his exact rules take (gold) whose favourable move EVER reaches the multiple before the stop, same day: 3R 14.7%, 10R 4.4%, 67R 0.02%. Median first target (last swing) = 2.67R.

## His own demo days (19–21 May 2026, gold), what the exact rules do

| Day (NY 17:00 day) | Prev-day high / low | Trades (entry NY time, dir, stop size, result) |
|---|---|---|
| 2026-05-19 | 4584.02 / 4480.06 | 19:17 short risk $1.65 → -1.03R (stop, max 0.5R); 19:23 short risk $1.48 → +2.33R (breakeven_after_partial, max 5.6R); 20:03 short risk $2.02 → +4.83R (target, max 60.6R) |
| 2026-05-20 | 4589.30 / 4464.61 | 21:38 long risk $5.01 → +1.94R (breakeven_after_partial, max 4.7R) |
| 2026-05-21 | 4552.59 / 4453.15 | 21:32 short risk $2.64 → -1.02R (stop, max 1.5R); 21:43 short risk $1.46 → +5.57R (target, max 54.8R) |

How unusual were his demo days? Under the exact rules the three days total +12.6R. Of all 3,031 three-consecutive-day windows 2015 → Sep 2026 only 0.26% were as good or better (1.55% of 2026 windows); only 18.7% of three-day windows made any profit at all (`gautam_backtest.py demorank`; no-trade days count as 0R).

## Second data source (gold, zero cost): Dukascopy BID vs HistData BID

| Variant | Dukascopy trades / win / avg R | HistData trades / win / avg R | Same entry minute |
|---|---|---|---|
| EXACT: fresh candle beyond the line, next-candle break, SL i | 5234 / 26.4% / +0.072 | 5226 / 26.4% / +0.091 | 4478 |
| HIS CLAIM 1:10: exact entries, fixed 10R target, full positi | 4726 / 12.1% / +0.132 | 4719 / 11.9% / +0.115 | 4041 |
| GENEROUS: any candle beyond the line, break any time, no SL  | 87041 / 33.8% / +0.049 | 86085 / 33.7% / +0.041 | 73431 |
| SIDHANT'S READING: the red/green candle itself must be the h | 4227 / 28.1% / +0.086 | 4230 / 27.5% / +0.090 | 3550 |
| COMMENTER'S READING: any red/green candle beyond the line; o | 5803 / 25.8% / +0.083 | 5811 / 25.9% / +0.090 | 4905 |

## The disputed rule: Trading With Sidhant's reading vs his commenter's (both tested)

Sidhant ("I Backtested Gautam Jha's Gold Strategy", UdGXYHp1nYA, 31 Aug 2026, 248K) reads the rule as "the red candle has to be THE highest candle" and asks whether the long side breaking it is "a rule, a mistake or an oversight?" (1447.7–1452.9). A commenter (@kr4T0X) says that is a misreading: any first red/green candle, broken by the immediate next candle. Sidhant shows days (10–11 Aug, 17–18 Aug, his re-run of 19–21 May) but no total win rate. All three readings, same everything else:

| Reading | Trades | Win rate | Reach 1:10 | Avg R | Avg R zero cost |
|---|---|---|---|---|---|
| Ours (exact): the candle or the one before it printed the extreme | 5,041 | 19.7% | 4.4% | -0.476 | +0.072R |
| Sidhant: the candle itself is the extreme | 4,105 | 21.0% | 4.4% | -0.447 | +0.086R |
| Commenter: any candle beyond the line, immediate break | 5,587 | 19.3% | 5.0% | -0.467 | +0.083R |

The disputed rule does not change the answer. Where Sidhant and we differ: he exits at the previous swing or day level by eye on a few days; we code every day from 2015 and keep Gautam's 80/20 split.

## Second strategy: his 2023 daily swing setup (aBaJD8Bnx6w, "80–90% accuracy") — SECONDARY test

Daily bars built from the Kaggle Nifty 100 15-minute set (debashis74017), Feb 2015 → Aug 2025, 100 stocks = today's Nifty 50 + Next 50 (two of his watchlists; survivorship bias in his favour). Several definitions are ours (pivot size, 'big candle', buffer, 60-day time exit): see `gautam_rules.json` → `swing.interpreted_by_us`. Costs: 0.25% round trip (delivery) + 0.05% slippage per side.

| Variant | Trades | Reached +3% before the stop [95% CI] | Closed in profit [95% CI] | Reached the previous high | Avg R [95% CI] | Avg return per trade | Years where +3% came first ≥ 80% |
|---|---|---|---|---|---|---|---|
| SWING EXACT: pivot k=3, wick break, first green close (no big candle) within 5, stop low - 1%, 50% at +3%, rest at previous high | 6,966 | 45.7% [44.6–46.9] | 44.0% [42.9–45.2] | 40.3% | -0.020 [-0.050, +0.010] | -0.038% | 0 of 11 |
| SWING EXACT: pivot k=3, wick break, first green close (no big candle) within 5, stop low - 1%, 50% at +3%, rest at previous high [zero_cost] | 6,969 | 46.3% [45.1–47.5] | 46.9% [45.8–48.1] | 40.3% | +0.103 [+0.072, +0.133] | +0.289% | 0 of 11 |
| SWING EXACT: pivot k=3, wick break, first green close (no big candle) within 5, stop low - 1%, 50% at +3%, rest at previous high [most_generous] | 6,967 | 47.0% [45.9–48.2] | 47.7% [46.5–48.9] | 40.9% | +0.122 [+0.092, +0.152] | +0.344% | 0 of 11 |
| SWING ALL CONDITIONS: + uptrend (close above a rising 50-day average) | 3,108 | 46.3% [44.5–48.0] | 46.0% [44.3–47.8] | 42.3% | +0.001 [-0.043, +0.044] | +0.040% | 0 of 11 |
| SWING: close below the low (not just a wick) | 5,671 | 48.1% [46.8–49.4] | 41.0% [39.7–42.3] | 36.4% | -0.028 [-0.062, +0.007] | -0.035% | 0 of 11 |
| SWING: bigger swings (k=5) | 4,812 | 49.0% [47.6–50.5] | 41.2% [39.8–42.6] | 35.7% | +0.011 [-0.027, +0.050] | +0.095% | 0 of 11 |
| SWING: ₹2.5 flat buffer | 7,135 | 42.8% [41.7–44.0] | 42.2% [41.1–43.4] | 38.2% | -0.051 [-0.086, -0.015] | -0.038% | 0 of 11 |
| SWING: entry next open | 6,829 | 43.9% [42.8–45.1] | 43.2% [42.0–44.3] | 40.3% | -0.133 [-0.282, +0.016] | -0.150% | 0 of 11 |

Cross-check on a second source (Yahoo daily, 30 stocks incl. his named Exide, Aptus, M&M Finance; to Sep 2026): exact rules 2360 trades, +3% first 42.2%, in profit 43.7%, avg -0.060R.
His own example stocks: APTUS: 50 trades, +3% first 42.0%, in profit 44.0%, avg -0.186R; EXIDEIND: 83 trades, +3% first 50.6%, in profit 41.0%, avg -0.042R; INFY: 77 trades, +3% first 44.2%, in profit 51.9%, avg +0.125R; M_MFIN: 105 trades, +3% first 46.7%, in profit 42.9%, avg -0.050R; BHARTIARTL: 79 trades, +3% first 39.2%, in profit 46.8%, avg -0.047R.

Notes: the Kaggle 15-minute set ends 6 Aug 2025 for most stocks; daily bars built from it use the last 15-minute close (official closes differ by a median ₹0.95 on Infosys). Prices are split-adjusted, so the ₹2.5 flat-buffer variant is approximate for split stocks; the 1% buffer is scale-free.

## Files

- BTCUSDT: `engine/public/projects/gautam-audit/data/gautam_results_BTCUSDT_ETHUSDT.json` (rules sha1 16f6dfa5be, generated 2026-10-10T15:48:16)
- ETHUSDT: `engine/public/projects/gautam-audit/data/gautam_results_BTCUSDT_ETHUSDT.json` (rules sha1 16f6dfa5be, generated 2026-10-10T15:48:16)
- NIFTY: `engine/public/projects/gautam-audit/data/gautam_results_NIFTY_BANKNIFTY.json` (rules sha1 16f6dfa5be, generated 2026-10-10T15:48:01)
- BANKNIFTY: `engine/public/projects/gautam-audit/data/gautam_results_NIFTY_BANKNIFTY.json` (rules sha1 16f6dfa5be, generated 2026-10-10T15:48:01)
- XAUUSD: `engine/public/projects/gautam-audit/data/gautam_results_XAUUSD.json` (rules sha1 16f6dfa5be, generated 2026-10-10T15:47:31)
- Trades (every trade, every variant): `projects/gautam-audit/data_raw/trades/<tag>/*.csv.gz`; headline copies for the proof page: `engine/public/projects/gautam-audit/data/trades/*.csv`.