Product · FinNexus Terminal

—

searchable securities

—

tools per analysis

—

automatic checks

—

score components

—

markets covered

04Red flags — seven checks, pure math

Leverage FAIL

Is the company over-borrowed?

Debt/Equity stands at 5.19, breaching the >2 threshold for non-financials

Promoter pledge N/A

Have the owners borrowed against their own shares?

Data unavailable — not published for this market

Auditor change PASS

Did the accountants walk away?

No auditor-change signal in 14 recent headlines

A rule with no data returns N/A, never PASS.

09Signal base rates — honest backtesting

Signal	Fired	Hit rate	Read
MACD crosses above signal	55	58.2% 45.0–70.3	does not beat the 52.3% baseline
RSI(14) crosses up through 30	20	55.0% 34.2–74.2	does not beat the baseline
Golden cross	4	100% 51.0–100	only 4 occurrences — no evidence

Losers shown beside winners, every result with its sample size and 95% interval.

03Score — the arithmetic, printed

Valuation 2.0 × 25% = 0.50 Growth 10.0 × 25% = 2.50 Financial Health 3.9 × 20% = 0.78 Momentum 10.0 × 15% = 1.50 Sector Tailwind 6.5 × 15% = 0.98 ───── 6.26 Red-flag penalty −1.00 (one FAIL: leverage) ───── FINAL SCORE 5.3 / 10 MIXED

Sliders let you re-weight it — the recompute runs server-side through the same function the pipeline uses.

Quality screen — the funnel, not just the survivors

Step	In	Cleared	Failed	No data
P/E 30 or less	1,132	486	371	275
Return on equity 25% or more	486	71	382	33
Profit growth 50% or more	71	9	58	4
Above the 200-day EMA	9	4	5	0

Two survivors with no explanation looks like a broken page. The column that is really doing the work is usually the one on the right.

Fund look-through — what you own, not what you bought

Inside the fund	Weight	Score	Flags
HDFC Bank	8.14%	6.8	leverage rule skipped — lender
Reliance Industries	6.02%	5.3	1 FAIL · 1 WARN
Unlisted / unmatched	4.7%	—	excluded from every weight above, not counted as zero

Two funds, one overlap figure: the weight they hold in common, summed name by name.

07Quality scores — published, not proprietary

	Piotroski test	Reading
pass	Profitable this year	ROA 0.92%
fail	Profit backed by cash	CFO −4.2bn vs PAT 5.6bn
fail	Liquidity adequate	current ratio 0.57x
n/a	Gross margin rising	not reported

Missing inputs shrink the denominator — "5 of 7", never a quietly deflated 5 of 9.

Capabilities

What you actually get.

Eight things, each built so you can check it rather than trust it.

Seven red flags

One comparison each, against a stored number.

Promoter pledge above 20%
Profit that is not becoming cash
Receivables outrunning revenue
Debt/equity above 2 — skipped for lenders
Interest coverage below 2×
Promoters selling three quarters running
Auditor departure in the headlines

missing data returns N/A, never PASS

Score out of 10

Valuation 25% · Growth 25% · Financial health 20% · Momentum 15% · Sector tailwind 15%, minus 0.5 per WARN and 1.0 per FAIL.

Valuation measured against the stock's own five-year history, not a universal threshold
Every weight, formula and input value printed beside the result
Sliders let you re-weight it and watch the score move

a missing input is redistributed, never scored as zero

SWOT, on a short leash

The one place the model genuinely writes — from the sealed pack and nothing else.

Every bullet ends with a source chip
Values not in the pack must be written as "data unavailable"
No buy, sell, hold or target-price language, ever

returned through a JSON schema, so it cannot wander off format

Bull vs Bear

Two personas, three rounds, one shared evidence pack. They argue interpretation, never data — neither can introduce a fact the other lacks.

Each round rebuts the one before it
The scorecard adjudicates; nobody wins on rhetoric

the same fact often supports both cases — that is the point

Every market, one method

The same seven checks and the same scorecard applied to every market, each in its own currency.

Segments ranked within each market, never across them
Peers stay on the same exchange, so sector signals are not blurred
London quotes in pence; prices are converted to pounds at ingest
Watchlist compares across markets by percentage, never by amount

every figure traces back to a stored price row

Quality screen

eight thresholds you set yourself, run across every cached name at once, with a funnel showing which one did the cutting.

P/E, return on equity, earnings per share, profit growth, profit margin, sales growth, debt to equity, promoter holding
A final pass keeps only names trading above their 200-day EMA, smoothed at 2/201 so it matches a charting package
Each step reports how many entered, cleared, failed and could not be tested
Near misses are listed separately — names that failed one or two tests
Twelve more filters behind a disclosure, off until you switch one on: size and liquidity, further valuation and balance-sheet measures, and two that compare a company with its own past

a stock with no stored figure is dropped, never passed

Mutual funds, ETFs and SIPs

The same research → watchlist → portfolio path as equities, plus the thing a factsheet will not tell you: what the fund actually owns.

Look-through to the companies inside, each scored by the terminal's own rules
Overlap between any two funds you hold, as the weight they share
Money-weighted return (XIRR) on your own dated cash flows, not the fund's published return
Monthly SIPs fill in their own instalments; a CAMS or KFintech statement imports folios and transactions directly
Seven checks on the evidence itself — how old the disclosure is, how much of the fund it covers, how much was identified by ISIN rather than by name

the funds module opens the equity database read-only and cannot write to it

Demo Mode

A recorded replay of the real app that runs with the internet switched off.

Frozen tool events, evidence pack and model output
Rendered by the same module as the live page — not a mock-up
Five scenarios, including a labelled market-crash simulation

demos cannot fail: no network, no rate limit, no outage

Every number shows its working.