APPLIED QUANTS — SAVED-RESULTS CASE STUDY

This is an aggregate description of an older owner-supplied Buy Bot research export. It is separate from the current Applied Quants calculation engine and is not the dataset used by the public practice backtester. The original protected CSV is not distributed here.

Observed file contents:
- 550 unique tickers
- 29,323 records; 319 columns
- Session values span 2024-03-23 through 2026-04-21
- NO_TRADE: 10,937 records
- FILTERED: 18,208 records
- OK: 98 records
- Provider errors: 80 records
- 98 records have positive entry_count

These are inventory counts, not an independent validation of prices, strategy performance, costs or coverage. A record is not necessarily a completed trade. Some records contain multiple entry/exit fields. Session values include non-trading dates. The export contains later-outcome research fields that must not be used as if known at entry time.

EXERCISE
1. Add the four status counts. Do they reconcile to the record count?
2. Explain why a provider error differs from a valid no-trade result.
3. Name the evidence needed before calling a row a completed trade.
4. Separate decision-time fields from hindsight outcomes.
5. Write one conclusion this inventory supports and one it does not.

The online preview uses artificial practice prices to teach the existing engine's workflow. Its results do not establish historical market profitability.