| Version | Area | Change | Why (evidence) | Judged by | Status |
|---|---|---|---|---|---|
| v155 2026-11-01 | research | Research-only job: /bt/fetch (admin key) downloads Screener.in pages for every NSE stock listed by 2023 at one page per ~2.5 s and stores parsed annual financials and shareholding in bt_screener; /bt/status, /bt/dump. No effect on any trading book. | Kushal, 31 Oct: back-test the hidden-gems rubric point-in-time (score as of Oct 2023, outcome to Oct 2026) before building the engine. | bt_screener holds ~1,800 companies; back-test report delivered. | live |
| v154.5 2026-10-29 | llm | Gemini fallback: thinking disabled (thinkingBudget 0) for short JSON verdicts, timeout >= 45 s; retried without thinkingConfig if a model rejects it. | 29 Oct 11:17: after v154.4 Gemini was reachable but every call hit 'read operation timed out' (flash models think by default). | Desk banner shows calls answered by Gemini; judge verdicts appear again while Anthropic credits are empty. | live |
| v154.4 2026-10-29 | llm | Gemini fallback no longer depends on one model id: tries the last working id, the configured one, gemini-flash-latest / 2.5-flash / 2.0-flash, then any flash model the account lists; remembers the winner. Fyers daily call counter persisted in the DB so restarts no longer reset it. | 29 Oct 10:01: Anthropic credits exhausted and Gemini answered HTTP 404 (model id retired) -> all LLM calls failed, judge names released. 27 Oct: Fyers hit its 5,000/day cap while our counter showed 2,808 (reset by 3 deploys). | Desk banner shows 'answered by Gemini' instead of LLM DOWN; /feed-status calls_today survives a restart. | live |
| v154.3 2026-10-27 | feed | Fyers pacer also enforces the 50 requests/minute cap (max 44 in any rolling 60 s); a request beyond that waits for the window instead of failing. | 27 Oct 10:20: after the v154.2 restart Fyers returned code -360 '50 requests/minute' (empty cache -> universe pull + every engine's small quote calls in the same minute). | /feed-status shows no -360 errors for the rest of the session. | live |
| v154.2 2026-10-27 | feed | Between the 3-minute Fyers universe pulls, a Yahoo refresh is used only if most rows are FRESH; when Yahoo is weak the Fyers snapshot (< 3 min old) is served and Yahoo is paused 3 min. | 27 Oct: Fyers healthy after v154.1 but breadth stayed STALE -- Yahoo returned ~37 of 723 names, the rest were last-good copies flagged stale. | Desk header breadth fresh (>= 300 quotes) all session; no STALE banner while Fyers is up. | live |
| v154.1 2026-10-27 | feed | Fyers per-second limit: all Fyers requests now go through one process-wide pacer (>= 0.3 s apart, ~3.3/s max, under the 5/s Standard cap) and a rate-limited batch is retried once after 1.5 s instead of being dropped. | 27 Oct 10:00: Fyers code -360 'Standard plan limit of 5 requests/second' -- engines on separate clocks overlapped; universe pulls came back with gaps, breadth went STALE (37 fresh quotes) and breadth-gated engines (dr, ws) could not act. | /feed-status shows no -360 errors; desk header breadth fresh with >= 300 quotes. | live |
| v154 2026-10-24 | weak desk | New paper book ws: on fresh breadth < 40, 10:00-13:30, names <= -3% in a red sector (turnover >= 5 Cr, no positive filing, not at lower band) are blind-ranked by the PM model (move % hidden) and SHORTED on the first failed 5-min bounce; stop 1.5%, trail 1% after +2%, square-off 15:10, max 3 open / 5 a day. | Kushal 23-24 Oct: five red sessions with no day-desk trade; results season (PASALO, SKYGOLD) missed; 'our decision is not suggesting stocks to buy and ranking of that'. | Replay: +0.14%/trade on breadth<40 days. Promote to live if the shadow book is positive after 10 fills; drop if avg < 0 after 15. | shadow |
| v154 2026-10-24 | results desk | filings.py now records EVERY results classification (beat or miss) into results_events. New paper book rd: BEAT names go LONG 09:30-10:30 only if above the open and previous close (< +8%, not at band; large caps half, big banks skipped), stop 1.5 / target 3 / square-off 15:10. MISS names go SHORT 09:30-11:00 if below the open and previous close, for up to 5 sessions, stop 2.5 / target 5 / trail 1.5 after 3. | Q2 FY27 study (89 reporters): median reaction -1.5%, move is open->close not the gap, up names fade -2.6% after, down names fall another -3.1%, mid-caps punished most. | Shadow through results season; promote the side that is positive after 10 fills. | shadow |
| v154 2026-10-24 | portal | /recos: one ranked page -- today longs, today shorts, 7-30 day ideas by win score (with SHORT column), Owner's List (populated by the Sunday Themes run). /rd-status and /ws-status JSON. Desk page shows the two shadow books separately. | Kushal 23-24 Oct: five red sessions with no day-desk trade; results season (PASALO, SKYGOLD) missed; 'our decision is not suggesting stocks to buy and ranking of that'. | Kushal can see what would be bought/shorted and why without reading logs. | live |
| v153.3 2026-10-24 | feed | Fyers budget: daily cap 4,000 requests (plan limit 5,000); outside 09:15-09:45 the universe is pulled from Fyers at most every 3 min, Yahoo refresh in between; 20 s shared cache. | 23 Oct: Fyers 429/-360 then 5,000/day cap; Yahoo fallback returned 2 quotes -> breadth 0%; zone exits of -3.9L invisible on /desk-view. | Fyers calls_today stays <4,000; breadth never prints 0%/50% artefacts on a stale feed; exits visible same day. | live |
| v153.3 2026-10-24 | feed | Yahoo fallback now pulls per-symbol with a 12-way concurrency limit (no burst 429s) and merges last-good quotes (<10 min old) flagged stale. | 23 Oct: Fyers 429/-360 then 5,000/day cap; Yahoo fallback returned 2 quotes -> breadth 0%; zone exits of -3.9L invisible on /desk-view. | Fyers calls_today stays <4,000; breadth never prints 0%/50% artefacts on a stale feed; exits visible same day. | live |
| v153.3 2026-10-24 | regime | Breadth is computed only when >=300 fresh quotes are available; otherwise REGIME is marked STALE and engines do not act on it (fixes 0%/2.5%/50% breadth artefacts of 23 Oct). | 23 Oct: Fyers 429/-360 then 5,000/day cap; Yahoo fallback returned 2 quotes -> breadth 0%; zone exits of -3.9L invisible on /desk-view. | Fyers calls_today stays <4,000; breadth never prints 0%/50% artefacts on a stale feed; exits visible same day. | live |
| v153.3 2026-10-24 | portal | /desk-view trades now include positions exited today and all open positions; /zpm trades show last 5 days incl. exits, newest activity first. | 23 Oct: Fyers 429/-360 then 5,000/day cap; Yahoo fallback returned 2 quotes -> breadth 0%; zone exits of -3.9L invisible on /desk-view. | Fyers calls_today stays <4,000; breadth never prints 0%/50% artefacts on a stale feed; exits visible same day. | live |
| v153.3 2026-10-24 | portal | Header banner shows stale-breadth state, Fyers last ok/err and whether the daily Fyers login is missing. | 23 Oct: Fyers 429/-360 then 5,000/day cap; Yahoo fallback returned 2 quotes -> breadth 0%; zone exits of -3.9L invisible on /desk-view. | Fyers calls_today stays <4,000; breadth never prints 0%/50% artefacts on a stale feed; exits visible same day. | live |
| v153.2 2026-10-23 | Pages | All status pages (/desk-view, /pm, /zpm, /ideas, /changes) update in place every 5 seconds (background fetch + swap, scroll position kept) instead of a 60s full reload. | Kushal, 23 Oct: 'auto refresh every 5 seconds'. | Page heading time stays within 5s of the clock. | live |
| v153.1 2026-10-23 | Compliance + feed diagnostics + portfolio risk layer | (1) Jain: SIGMAADV excluded; the filing TEXT and live-event news text are screened (ammunition/artillery/ordnance/missile/NATO/armed forces/liquor/meat terms), not only the issuer name. (2) Fyers: every failed batch records HTTP status + message (shown on /feed-status and the desk banner, with 'daily login not done' when the token is missing); one shared universe quote per 20s for all engines (~4x fewer Fyers calls). (3) guard.py: per-book daily circuit breaker (-2% today -> no new entries), sector cap (max 2 open per sector per book), drawdown de-risk (book < -5% since start -> half size) applied to sc/dr/ol/pm/zpm/1d entries. (4) Ideas: catalyst date must fall inside the horizon (+5d) else WATCH; +25% in 20 sessions without a dated in-horizon catalyst and score>=70 -> WATCH; book threshold 60 -> 65. (5) /pm, /zpm, /ideas show run times. | 21 Oct SIGMAADV (155mm ammunition order) armed by name-only screen; Fyers dark midday 3 sessions running with the cause swallowed; old portal had book-level risk controls the server lacked; SKYGOLD bought 15d with a catalyst 85 days out after +32%/month. | GUARD lines on /desk-view when a breaker fires; Fyers error text visible next outage; no defence/excluded name armed; WATCH ideas tracked separately from WAITING on /ideas. | live |
| v153 2026-10-19 | Resilience + judgment fixes (no new setups) | (1) LLM: Gemini as last-rung fallback when every Anthropic model fails (credits/outage); HEALTH tracker; LLM-DOWN banner on /desk-view + Telegram alert after 3 failures/10 min. (2) Live desk: a failed judge call no longer marks the name as judged or counts against the 150 cap. (3) Short gate prompt rewritten as a short seller's checklist: unexplained spike = act, concrete event/circuit/results-due = stand down; valuation never a reason. (4) sc WHYNOT log: every name up >=6% gets one line per reason saying which filter stopped it; shown on /desk-view. (5) Quote feed: Fyers rows with no price fall through to Yahoo (IXIGO); feed-source banner. (6) Jain screen: explicit defence/aerospace ticker list (MTARTECH, AXISCADES, APOLLO, DYNAMATECH, CENTUM...) and removed from the universe (728 -> 723). (7) Pages auto-refresh every 60s. (8) Zone PM no-chase: a name up >=12% in 5 sessions on >=3x volume is listed as WATCH (score -15), not entered, for a 10-session cool-off. (9) Intraday VWAP/opening-range data falls back to Yahoo 5-min bars when Fyers history is dark (JSWSTEEL 19 Oct: armed 12:04, above trigger 12:05-13:15, never filled -- 'no minute data'). | 19 Oct: credits ran out at the open -- 80 min dark, 42 names burned as 'judged', no exits on open longs; gate stood down JUSTDIAL (+9.9%, no news) citing 'no news'; MTARTECH armed by the catalyst scan; IXIGO unpriceable; ORIENTELEC/HBLENGINE bought the morning after 14-17% spike days. Replays (727 names, 60 sessions): spiked names -1.2% next 5 sessions / -2.2% sessions 6-10, pullback entry worse (-1.7%); short entry on the exhaustion bar beats waiting for a failed bounce (+0.24% vs +0.14%) so entry timing is unchanged; weak-regime breakdown shorts +0.14%/trade at 45% -- parked for the v154 PM-led design, not added mechanically. | Next 10 sessions: sc fills/day and gate stand-down reasons on /desk-view; zero 'judge call failed' names lost; WATCH ideas' 10-day forward return vs WAITING ideas; no defence name in any book. | live |
| v152 2026-10-16 | Short engine + dip-reclaim (breakage + rule) | (1) Warm start: after a restart during market hours sc and dr rebuild today's 5-minute buckets from the day's bar history instead of being blind for 15 minutes. (2) Short-climax trigger lowered from +8% to +6% on the day; climax 4x / dry-up 50% / gate / manager / regime floors unchanged. DR breadth gate (45) kept. | 15 Oct: four climax signatures (SHOPERSTOP +3.7, BAJAJELEC +10.4, JAIBALAJI -1.5, GALLANTT -1.5 in slot terms) all missed because three deploys left the engine without bars. 60-session replay (727 names, 5-min bars, mechanical exits): 8%/4x n=304 avg +0.33% hit 45% per-day +1.75%; 6%/4x n=701 avg +0.24% hit 45% per-day +2.88%; 5%/3x dilutes to +0.11%. DR by true breadth at 10:15: <45 is -0.4%/-0.0%, >=50 is +0.34%/+0.58% -- gate stays. | sc: signals/day and per-trade avg over the next 10 sessions vs the +0.24%/45% replay; WARM log lines after any restart. dr: unchanged (gate kept at 45 on evidence). | live |
| v151.2 2026-10-15 | Long desk (breakage) | Patience floors: long manager does not review a position before 15 min unless >= +1.5% or <= -1.0%; dip-reclaim needs a real reclaim (close >= 0.3% above prior bar high). (Daily cap / spacing removed at Kushal's direction -- 3 slots + 15-min patience are the limiters.) | 15 Oct 09:25-10:09: 32 dip-reclaim round trips in 45 min, each booked by the manager within 1-6 min at +/-0.1%. | median hold of long positions >= 20 min; manager BOOK before 15 min only on +1.5%/-1%. | live |
| v151 2026-10-14 | Long day desk | Position MANAGER for every long day-desk position (catalyst 1d, dip-reclaim, opening-leaders, day PM): reviewed every 5 min (every minute once >= +1.5% or held 40 min) with last ten bars, breadth + slope, sector, news, time, today's record -> HOLD / BOOK / TIGHTEN(floor), logged as MGR and Ledger-scored. Replaces the fixed lock ladder and 3-bar trails; stop and square-off stay as hard floors. | Kushal 14 Oct: 'how are you handling profit booking.. look at SHYAMMETL' -- peaked +2.1%, only a mechanical floor ~1% under the peak was watching it. | Ledger MGR_BOOK / MGR_TIGHTEN on longs vs subsequent move; give-back from peak on winning longs. | live |
| v150 2026-10-14 | Short engine | Intelligent short desk: (1) entry gate DECIDES for shorts and sees the tape (breadth + 30-min slope, sector move today, time, today's record, news/story on the name in last 2h) not only the business; (2) a position MANAGER (Haiku) reviews every open short every 5 min (every minute once >= +1.5% or held 40 min) and returns HOLD / BOOK / TIGHTEN(floor) with a reason, logged as MGR and scored on the Ledger; (3) mechanical layer reduced to hard floors: stop 1.5%, 1-minute squeeze exit, 60-min no-progress, square-off, regime floor (strong tape: no entry before 10:15, half size, stand down after 2 stops). | Kushal 14 Oct: 'logic is right but engine needs to be smart... alert about when to book profit... more agile intelligent'; DEN shorted twice while up on a story. 13-14 Oct: 0/6, -Rs 41k; HATHWAY +3.1% peak closed +0.4%. | Ledger MGR_BOOK/MGR_TIGHTEN vs subsequent move; FILL_SHORT by breadth bucket; give-back from peak; shorts refused by gate vs their close. | live |
| v149 2026-10-13 | Long desk + Ideas | (1) Decision gate DECIDES for the long desks (1d/ol/pm): stand-down blocks the fill, multiplier sizes it; sc/dr stay advisory. (2) Surprise check on every filing: 5-day run, same-event filings in 30 days -> FRESH / PARTLY / ALREADY_PRICED, fed to the gate. (3) /ideas: 7-30 day suggestions ranked by win score, catalyst + date first, why-now, surprise, entry zone/stop/target, with a running track record; research daily 16:15; the zpm book trades only the top 3 scoring >=60. (4) Healthcare (hospitals/diagnostics, 23 names) removed from universe and added to the Jain screen. | Kushal 13 Oct: 'catalyst more intelligent and smart to pick and choose'; 'one section with 7-30 day suggestions with winning score and clear catalyst, not mechanical'; Fortis/hospitals not allowed. | Ledger GATE_ACT vs GATE_STANDDOWN on 1d; /ideas track record (target vs stopped by score band); zero healthcare names in any book. | live |
| v148.2 2026-10-09 | Feeds (breakage) | Desk keeps its own close-of-day snapshot (closes table, 15:31; seedable from Yahoo) and uses it as the reference previous close for every quote. | 9 Oct: Fyers returned Wednesday's close as prev_close on Friday -> breadth read 7% on a 67% day, dip-reclaim stayed off, PM called a strong tape neutral, short threshold mis-measured. | Server breadth vs independent Yahoo breadth each morning (should agree within 3 points). | live |
| v148 2026-10-08 | Zone books | Mechanical 7/15/30 entries paused (shadow + A7/B15/C30). Zone PM: candidates from filings/dossiers/RS/sectors get the full scanner workup; strong model writes a zone note Mon/Thu 16:15 (3-5 positions: horizon, business thesis, dated catalyst, entry ZONE, volatility stop, target from own range, invalidation; do-not-touch list); daily 15:40 thesis review (HOLD/ADD/TRAIL/CUT); machine only fills in-zone and enforces limits. Book zpm, Rs 1 Cr, /zpm. | Kushal 8 Oct: zone decisions 'too mechanical… needs to be more intelligent and fundamental'. Record: 47 closed, 8 wins, 0 targets. | Ledger ZPM_IDEA / ZPM_REFUSED forward returns over 7/15/30 days; book zpm vs legacy zone books. | live (research needs API credit) |
| v147 2026-10-07 | PM desk | The model is the trader: strong model (Sonnet) writes a morning note at 08:35 from overnight filings/wire, movers, sectors, global tape, regime, dossiers, RS leaders and its own scorecard -> market view, 3-6 LONG/SHORT ideas each with a checkable entry condition, stop, target range, size, invalidation, plus refusals. Reviews at 09:40/11:30/14:00 (HOLD/CUT/TIGHTEN/ADD/CANCEL). Mechanical layer only executes its conditions and hard limits (Jain, liquidity, max 5, size<=25%, stop<=3%). Book pm; /pm page. | Kushal 7 Oct: 'mechanical with super intelligence is a failed model' -- intelligence must pick the stock and the trade, mechanics only execute. | Ledger PM_LONG_IDEA / PM_SHORT_IDEA / PM_REFUSED forward returns; book pm expectancy vs sc/dr/ol/1d on the same days. | live from 8 Oct 08:35 |
| v146 2026-10-07 | Engine | Opening Leaders engine (book ol): only names with an overnight/pre-open filing or wire reason; gap>=3% and still above open at 09:30; not at band; gate reads dossier+event; max 3, stop 2%, trail 1.5% after +3%. Filing reader's first morning pass now covers since yesterday's close. | Study 7 Oct, 57 sessions: mechanical 09:30 volume leaders -0.09%/trade (284 trades, 32/57 days negative); gap>=5% + extended past gap +0.93%/trade, 59% hit (n=27) -- continuation of real events is the only opening pattern with a number. | Ledger OPENING_LEADER FILL_LONG expectancy vs study +0.93%; GATE ACT vs STAND DOWN forward returns. | live from 8 Oct 09:30 (gate shadow) |
| v145 2026-10-07 | Intelligence | Stock dossier table (quality, 4-quarter trajectory, ROCE, pledge, promoter, P/E) built nightly in rotation; decision gate (Haiku) on every short-climax, dip-reclaim and filing-long signal returns act/size/reason, logged as GATE and scored by the Ledger by quality bucket. | Kushal 7 Oct: 'engine is too mechanical, add fundamentals and intelligence'. | Ledger GATE_ACT vs GATE_STANDDOWN forward returns; mechanical vs gated expectancy. Gate goes live on new entries only if it beats mechanical. | live (gate in shadow mode) |
| v144.1 2026-10-07 | Feeds | Fyers retries missing symbols as BE series (new listings). | SWASTIKAIN (Rs 616 Cr MSEDCL order, listed 30 Sep) armed at 09:46 but quote came back 0.0 - could never fill. | Count of ARMED plans that log 'px 0.0' after this date; should be zero. | live |
| v144 2026-10-07 | Engine | Dip-reclaim LONG engine (book dr): dip >=3% by 10:15, reclaim prior-bar high, only when breadth >=45. | Study 25 Aug-25 Sep, 704 names: +0.95%/trade, 70% hit on breadth>=45 days; ~0 on weak days. | Ledger DIP_RECLAIM FILL_LONG expectancy vs study +0.95%. | live |
| v144 2026-10-07 | Day desk | E5 failed-to-progress exit (45 min, peak <1%); fixed +3% target cap removed for 1d longs, lock ladder trails instead. | Day-desk longs that had not shown +1% in 45 min averaged -0.4% by close; winners were being capped. | 1d avg loss should shrink below -1.09%; avg win above +1.07%. | live |
| v144 2026-10-07 | Zone books | B15 paused for new entries; A7/C30 new entries at half size; no new zone long when breadth <40; pace rule cuts only losers. | A7 -6.71L, B15 -11.02L, C30 -4.25L realised; 0 of 44 shadow trades ever reached target. | Zone realised P&L per week; count of new entries opened below breadth 40 (should be 0). | live |
| v144 2026-10-07 | Measurement | Reaction Ledger: every verdict (LIVE GO/NO, FILING, ARMED, FILL, EXIT, SKIP) scored at +15m/+1h/+3h/close with MFE/MAE; nightly 16:30; /ledger and /desk-view. | Kushal 6 Oct: results 'very usual'; nothing measured whether verdicts were right. | 1,527 rows back-filled from 23 Sep. All long verdicts negative to close; shorts positive. | live |
| v144 2026-10-07 | Sources | News-wire RSS (ET, Moneycontrol, Business Standard) feeds the filing materiality pipeline, tagged WIRE. | Blue Cloud / Whirlpool moved on news the system never saw. | Ledger FILING rows with src WIRE: expectancy vs BSE/NSE rows. | live |
| v144 2026-10-07 | Ops | /admin/deploy: server commits a bundle to GitHub itself using GITHUB_DEPLOY_TOKEN from env. | 10-day outage 26 Sep-6 Oct: shipping depended on Kushal's laptop + browser. | Deploys without browser once token is set. | live, token pending |
| v144 2026-10-07 | Universe | Universe rebuilt to 751 names from NSE 500/Midcap150/Smallcap250/Microcap250, Jain-screened. | Ola, Suzlon, Idea, Blue Cloud were outside or filtered. | Movers >5% not in universe per day (target <3). | live |
| v143 2026-09-25 | Bug | Long monitor no longer touches book sc. | SHOPERSTOP short closed instantly as 'E4 STOP' by the long monitor. | No sc exits with E-codes. | live |
| v142 2026-09-25 | Engine | SHORT climax-exhaustion engine (book sc): up>=8%, climax vol>=4x, exhaustion bar <50% & no new high; stop 1.5, FTP 45m, trail after +3, square-off 15:10. | Study: +0.61%/trade n=91, 53% hit, W/L 2.1, positive in every breadth bucket. | Live: 9 trades, 7 wins, +Rs 52,261 (+1.49%/trade) to 6 Oct. | live |
| v141 2026-09-25 | Sources | BSE/NSE RSS filing reader every minute 08:45-14:45 -> Haiku materiality -> LIVE_EVENT plans. | Ola/Blue Cloud: moves came from filings the system did not read. | Filings read/day (~80); ARMED from filings; Ledger FILING ARMED expectancy. | live |
| v135-v140 2026-09-24 | Portal | /scan stock scanner: tape, Screener fundamentals, quarterly trajectory, Perplexity news, desk history, committee JSON with direction and ranges. | Kushal: 'put any stock and get full details'. | Known gaps: TATAELXSI quarter parse; ts label. | live |
| v134 2026-09-24 | Universe | Rs 50 price floor replaced by liquidity guard (px>=10, turnover>=Rs 5 Cr). | Ola 41->44 missed; Suzlon, Idea invisible. | Sub-Rs50 names judged per day. | live |
| v133/v138 2026-09-24 | Live desk | Judge cap 60->150/day; judged set hydrated from DB on restart. | Cap exhausted by 11:48; every deploy re-judged the same names. | LLM spend/day; duplicate judgments (should be 0). | live |
| v132 2026-09-24 | Regime | Breadth/Nifty regime gate at fill: >=55 normal; 40-55 first-order half size; <40 first-order conf>=0.75 at 10%. | Four 24-Sep morning longs all entered 1-3 bars after a volume climax on 25% breadth. | Ledger GO_WITHDRAWN rows: what the gate refused and what it did afterwards. | live |
| planned | Regime | Regime controller: per-minute breadth/slope/Nifty -> per-engine size; sizes re-derived weekly from the Ledger by regime bucket. | Ledger: long fills -0.41% below breadth 40, +0.21% at 40-55; shorts positive in all buckets. | Per-regime expectancy of each engine. | awaiting go |
| planned | Study | Strong-breadth long study: leader continuation / first VWAP pullback with filing, on breadth>=55 days only. | No proven long engine for strong tapes. | Expectancy, hit rate, W/L before anything is enabled. | tonight |
| planned | Zone books | Stop all new zone entries; redesign entries (regime constructive, above 20dma, verified first-order filing), targets from 10-day range, volatility stops. | 0 of 44 zone trades reached target; -Rs 22L realised. | Two weeks paper on Ledger before capital. | awaiting decision |