How the learning lab works
VeltaCapital is a historical-data reasoning exercise. It is not a trading system, brokerage, investment recommendation, or claim that chart patterns reliably predict future returns.
The included legacy dataset does not preserve exact dates. The interface discloses that limitation. The updated dataset builder preserves provider, generation time, start date, prediction split date, end date, adjustment status, and selection procedure. The legacy dataset must be regenerated before production launch.
Chart selection
The current builder requests a fixed list of liquid U.S. stocks and broad-market ETFs from Alpha Vantage, takes the most recent 60 available daily sessions for each symbol, and shuffles them using a fixed seed. It does not select windows based on whether a later move looks easy to predict. This is still a limited, non-representative sample.
What the score measures
The path score compares five estimated closing prices with five historical closing prices. Each point receives declining credit as its distance from the historical close increases relative to the selected tolerance. The displayed direction result is separate and is not blended into the path score.
What the score does not measure
It does not prove investing skill, future performance, causal understanding, or the reliability of technical analysis. A reasonable interpretation can have a wrong outcome, and a weak interpretation can be lucky. Confidence and evidence selections are recorded for reflection, not treated as objective truth.
Explanations
Post-round notes describe visible features such as price change, prior range, relative volume, and a moving average. They deliberately avoid saying that a pattern caused a market move. News, filings, and macroeconomic events require separately sourced reporting.
Hypothetical $100 option comparison
The result panel includes a simplified intrinsic-value comparison for a $100 call or put allocation. It uses the last visible close as the strike and assumes a premium equal to 5% of that strike. The calculation permits a fractional option equivalent so different charts can be compared from the same $100 starting amount. Listed U.S. equity options usually represent 100 shares and generally cannot be purchased fractionally, so the interface also reports the assumed cost of one whole contract. This comparison is not a historical option quote, executable return, or forecast.
Simulation portfolio and certificate
For authenticated course users, the server recomputes the $100 comparison from its own copy of the selected historical round and stores the budget, strike, assumed premium, closing price, ending value, and net result. The portfolio totals these independent $100 exercises; it is not a compounding brokerage account. A printable milestone becomes available after at least 25 logged rounds and at least $1,000 of cumulative net hypothetical profit. The milestone is deliberately labeled as a historical simulation certificate and does not certify real investment skill or professional competence.
Known limitations
- Short-horizon prices contain substantial noise.
- The ticker list is not representative of all securities, countries, or regimes.
- Unadjusted data can contain apparent gaps around corporate actions.
- Historical exercises create hindsight and selection risks even when windows are selected mechanically.
- The option comparison omits time value, implied volatility changes, bid-ask spreads, liquidity, commissions, taxes, early exercise, and assignment.
- Transaction costs, taxes, liquidity, spreads, and investor circumstances are outside the chart exercise.
Quality controls
Financial formulas have automated regression tests. Changes to scoring require a versioned method and test updates. Before paid launch, course material and calculations should receive documented review by an appropriately qualified independent professional.