Before any strategy, know the size of the thing
Most PSX opinions are formed from a few dozen remembered trades. That is not enough data to see anything. Here is the ground this whole Academy stands on, counted rather than estimated.
| What | How much | Period |
|---|---|---|
| Daily price bars | 1,029,511 | 2016-01-01 to 2026-07-31 |
| Symbols covered | 485 | same period |
| Intraday rows | 7,084,054 | 2026-07-10 to 2026-07-31 |
| , of which carry the trade side (buyer- or seller-initiated) | 2,364,649 | same |
Ten and a half years is roughly 2,600 trading sessions. A trader who takes two positions a month for ten years sees 240 trades. The panel above is four thousand times that. This is the only reason a claim in this course can carry an n.
What that lets you ask
With one symbol you can only ask "what happened to me?". With 485 symbols and 2,600 sessions you can ask "what happens to stocks like this, and how often?", which is the only question with an answer.
Evidence · History · Confidence · Counterargument
Evidence, SELECT COUNT(*), MIN(date), MAX(date), COUNT(DISTINCT sym) FROM prices returns 1029511 | 2016-01-01 | 2026-07-31 | 485. [MEASURED 2026-07-31]
History. The intraday tape is young: it starts 2026-07-10. Anything taught from tick data in this Academy has three weeks behind it, not ten years, and is labelled as such.
Confidence, high on the counts themselves; they are a row count, not an estimate.
Counterargument. The panel has survivorship bias, confirmed and unfixable. Of 484 symbols, 478 traded in the last three sessions and not one stopped trading more than 250 sessions ago across 10.5 years. Companies that delisted are simply missing. So every positive result in this course is an upper bound, and every negative result is, if anything, understated. We will repeat this warning; it is the single most important limitation of everything we publish.