Market Data Warehouse

yfinance → dbt → a dimensional (star-schema) warehouse: dim_date, dim_security, dim_exchange, and a fact-table family sharing that grain (fct_security_price, fct_security_technical_daily, fct_security_risk_daily, fct_corporate_action). This page reads straight off the built warehouse -- a live DuckDB query, not canned numbers. Runs on DuckDB or MotherDuck's free tier, and the same dbt code has been run end to end on Snowflake too (a trial account, with key-pair auth, a least-privilege role and a transactional load).

Securities Tracked

Exchanges

1962-01-02

Earliest Trading Day

2026-09-10

Latest Trading Day

94,346

Price Rows (fct_security_price)

1,071

Dividends Paid

Stock Splits

Relative performance

Adjusted close, indexed to 100 at each security's earliest point in the selected window, so tickers at very different price levels are comparable on one chart. Straight off fct_security_price, joined to dim_security.

From To

Latest snapshot

Most recent trading day on record for each security, from the same fact table.

Ticker	Name	Sector	Exchange	As of	Close	Day change	Volume
AAPL	Apple Inc.	Technology	Nasdaq Stock Market	2026-09-10	326.57	+3.56%	69,925,100
AMZN	Amazon.com, Inc.	Consumer Cyclical	Nasdaq Stock Market	2026-09-10	251.89	-0.20%	25,443,300
GOOGL	Alphabet Inc.	Communication Services	Nasdaq Stock Market	2026-09-10	332.60	+0.59%	23,508,500
JPM	JPMorgan Chase & Co.	Financial Services	New York Stock Exchange	2026-09-10	353.56	-0.32%	4,184,200
KO	The Coca-Cola Company	Consumer Defensive	New York Stock Exchange	2026-09-10	87.83	+0.32%	12,019,900
MSFT	Microsoft Corporation	Technology	Nasdaq Stock Market	2026-09-10	492.44	+0.16%	16,018,200
NVDA	NVIDIA Corporation	Technology	Nasdaq Stock Market	2026-09-10	218.36	-2.26%	105,482,000
SPY	State Street SPDR S&P 500 ETF Trust	--	NYSE Arca	2026-09-10	757.83	-0.60%	42,679,700
XOM	ExxonMobil Holdings Corporation	Energy	New York Stock Exchange	2026-09-10	165.23	+0.61%	16,615,300

Technical & risk indicators

From fct_security_technical_daily and fct_security_risk_daily -- two more fact tables at the same (security, trading day) grain as the price fact, kept separate because they serve a different question and change at a different rate. RSI here is a simplified, SMA-based approximation, not the canonical Wilder-smoothed formula. Beta is versus SPY. Sharpe/Sortino assume a 0% risk-free rate (no real T-bill data source in this project) -- directionally useful, not precise.

Ticker	SMA-20	SMA-50	SMA-200	RSI-14*	52w range	Volatility (ann.)	Sharpe	Sortino	Beta (vs SPY)	Max drawdown
AAPL	315.28	317.22	284.65	64.7	226.79–340.08	25.2%	1.46	2.15	0.68	-13.8%
AMZN	259.38	254.95	239.62	40.5	198.79–284.02	34.4%	0.33	0.54	1.41	-21.7%
GOOGL	341.45	347.63	336.62	42.5	236.57–402.62	31.5%	1.21	1.96	1.36	-21.1%
JPM	357.08	351.30	319.15	53.2	282.84–365.18	22.1%	0.97	1.38	0.78	-15.5%
KO	89.04	86.41	78.70	36.1	65.67–91.99	18.7%	1.62	2.77	-0.26	-11.1%
MSFT	494.50	451.44	431.14	56.9	352.83–542.07	32.4%	0.15	0.23	0.95	-34.9%
NVDA	220.51	212.17	197.28	51.0	165.17–235.74	38.1%	0.84	1.27	1.93	-20.2%
SPY	767.56	758.25	713.71	45.1	631.97–777.88	12.8%	1.34	1.97	1.00	-9.1%
XOM	161.96	154.58	145.42	48.4	110.64–171.47	25.7%	1.80	2.74	-0.56	-20.6%

*Simplified RSI (SMA-based), not the canonical Wilder-smoothed formula.

Price projection

Not a forecast. Not financial advice. Three different, real forecasting assumptions, not one model tuned three ways -- Linear trend extrapolates recent momentum; Random walk (drift) is the textbook no-skill baseline every real forecast gets benchmarked against; Mean reversion assumes the opposite of trend-following, that price pulls back toward its own recent average. None has any demonstrated ability to predict real future prices -- stock prices are close to a random walk. Watch the R² figure (blank for Random walk, which has no analogous fit statistic): a low value means the assumption barely fits the recent data at all.

From fct_security_price_projection -- every method x lookback-window combination is precomputed by dbt; the pickers below just select which already-tested rows to show, no math runs in the browser or in this Flask app. 95% confidence band, widening with time, driven by recent volatility (same treatment for all three methods, so only the central assumption differs). Full-refresh every warehouse build -- always today's projections, not a record of past ones.

Autonomous stock analysis

Pick a ticker and watch each step as the model actually takes it -- a live call each time, capped the same way a normal assistant turn is. Limited to the tickers this warehouse has projections for.