Skip to content

impact_model.md

Latest commit

 

History

History
108 lines (80 loc) · 5.5 KB

impact_model.md

File metadata and controls

108 lines (80 loc) · 5.5 KB
Raw

AlphaStream India - Impact Model

Problem Scale

  • 14 Cr+ demat accounts in India (CDSL + NSDL)
  • ~3 Cr active traders
  • Average retail investor spends 2+ hours/day on research
  • 80% rely on WhatsApp tips and unverified sources
  • Average missed opportunity cost: ₹3,000-10,000/quarter

Market Opportunity (TAM / SAM / SOM)

Level Segment Size Notes
TAM Total Indian equity market participants 14 Cr CDSL + NSDL demat accounts (2024)
SAM Active traders who research daily ~3 Cr ~20% of demat accounts trade monthly
SOM Realistic 5-year premium capture (0.5%) 1.5 L Conservative; industry freemium converts at 0.5-1%

Key assumption: SOM uses 0.5% of SAM, not TAM — anchored to paying users, not total accounts.

AlphaStream India Impact

1. Time Saved

Metric Without AlphaStream With AlphaStream
Daily research time 2+ hours 15 minutes
Filing analysis Manual reading (30 min/filing) Instant LLM classification
Insider trade monitoring Manual NSE checks Automated alerts
FII/DII tracking Visit NSDL website daily Real-time flow analysis

Time saved: 1.75 hrs/day × ₹500/hr × 250 trading days = ₹2.19L/user/year

2. Alpha Generated

Signal Type Detection Backtest Win Rate Avg Return (30d)
RSI Divergence Automated scan across Nifty 500 60-78% +3.26%
MACD Crossover Real-time detection 45-57% +0.71%
Insider Cluster Buying 3+ insiders in 30 days Historical strong signal Varies
FII Buying Streak 5+ consecutive sessions ~72% (historical) +3-5%

Conservative estimate: 2-3 actionable signals/month × 60% accuracy × ₹3,000 avg = ₹54,000/user/year

3. Risk Reduction

  • Early warning on adverse signals prevents panic selling
  • Portfolio-aware alerts: "Your IT holdings at risk from FII selling"
  • Backtested confidence: "This pattern has worked 78% of the time on this stock"
  • Estimated 15% reduction in behavioral losses

4. ET Markets Business Impact

Metric Estimate Assumption
Increased engagement +40% time on platform Signal-dependent users return daily
Premium conversion 0.3% free → paid ₹999/month subscription
Revenue (Year 1) ₹59.9 Cr/year 50K early-adopter premium users × ₹999 × 12
Reduced churn 2x retention Signal-dependent users have higher stickiness

Revenue Projection (ET Markets Premium)

Year 1 Year 3 Year 5
Premium users 50,000 2,00,000 5,00,000
Price (₹/month) ₹999 ₹999 ₹999
Annual Revenue ₹59.9 Cr ₹239.8 Cr ₹599.4 Cr

Assumptions: Year 1 = early-adopter base from existing ET Markets free users. Year 3 = viral growth via portfolio alerts sharing. Year 5 = network effects + Tier 2/3 city expansion.

Cost Reduction for ET Markets

Cost Category Current Approach With AlphaStream Estimated Saving
Signal content (est. ₹20 Cr/year) Manual analyst teams + editorial curation LLM-generated, human-reviewed signal summaries ₹5-10 Cr/year (30%)

Automated signal pipelines handle high-frequency, repeatable content (FII flows, insider alerts, technical scans). Humans focus on narrative, investigative, and opinion content — higher value, lower volume.

Competitive Moat

  • India-specific regulatory data: Direct integration with NSE SAST/PIT filings and NSDL FII/DII data — not available via generic global APIs.
  • Backtested signals, not LLM speculation: Every signal carries a win rate grounded in historical NSE/BSE data; pure LLM wrappers cannot provide this without the same data pipeline.
  • Real-time streaming architecture: Pathway-powered incremental ingestion with sub-second updates; static dashboards require full re-computation on each load.
  • Portfolio-aware alerts: Signals are contextualised to a user's actual holdings — generic screeners give the same alert to everyone regardless of exposure.
  • Compound moat over time: Each user interaction (query, alert acknowledgment) improves signal ranking; the system gets more accurate as the user base grows.

Assumptions (Stated Explicitly)

  1. Win rates based on backtesting against Nifty 50 stocks with 3-5 year data
  2. Average return estimates use median, not mean (to reduce outlier effect)
  3. Time savings assumes basic digital literacy and daily market participation
  4. Premium conversion rate of 0.3% is conservative (industry average 0.5-1%)
  5. Back-of-envelope calculations — detailed validation would require A/B testing
  6. SOM (1.5L users) = 0.5% of SAM (3 Cr active traders) — conservative for a 5-year horizon
  7. Revenue projections assume no price increase over 5 years (understates upside)
  8. Cost reduction estimate (₹5-10 Cr/year) assumes research content budget of ~₹20 Cr/year for a platform at ET Markets' scale
  9. Year 3 and Year 5 user projections assume compounding growth of ~100% and ~150% respectively from Year 1 base

What Makes This Different

What 99% of teams build What AlphaStream builds
Call GPT-4 → get text blob Text2SQL pipeline → grounded in real data
No verification Historical backtest with win rates
Mock data Live NSE/BSE/FII data
Streamlit dashboard Production React + WebSocket
No guardrails Input/output guardrails + correction loop
Single ticker lookup Multi-signal fusion (Alpha Score)