You now have six lessons of measurements. This one turns them into something you can carry.
The scoreboard
| Rule | TRAIN | TEST | Verdict |
|---|---|---|---|
| Close above 52-week high | 1.91× | 1.29× | works |
| Breakout + above 200-day average | 1.55× | 1.25× | works |
| Breakout on 2×+ volume | 1.63× | 1.26× | works |
| RSI above 80 | 1.60× | 1.35× | works |
| Riding above the upper Bollinger band | 1.36× | 1.21× | works |
| Fallen more than 50% from the 52-week high | 1.74× | 1.26× | works |
| Bullish divergence | 1.18× | 1.66× | unstable, candidate only |
| RSI below 30 | 1.36× | 1.09× | decayed |
| Bearish divergence | 1.47× | 1.06× | decayed |
| Above/below the 200-day average, alone | 1.10× | 0.98× | nothing |
| Above/below the 50-day average, alone | 1.08× | 1.00× | nothing |
| Bollinger squeeze | 0.88× | 0.90× | worse than random |
| Buying at the 20-day low ("support") | 0.86× | 0.87× | worse than random |
The pattern inside the pattern
Read the "works" block and the "worse than random" block together. Every rule that worked involves a stock that is already moving hard. Every rule that failed involves a stock that is quiet or falling.
That is one finding stated six ways, and it is the central finding of this course:
On PSX, over 10.6 years, the setups that feel safe are the ones that measure worst. Support, the squeeze, the healthy mid-range pullback. These are the comfortable entries, and they are where the base rate goes to die.
Building your own checklist
A workable one, drawn only from rows that survived both periods:
- Is it moving? Above the upper Bollinger band, RSI high, or breaking a recent high. If none of these, the odds of a 20% move are below average, however good the story is.
- Did volume confirm it? 2×+ normal on the breakout day. This is the difference between 1.63× and 1.13×.
- Is it above its 200-day average? As a qualifier only, after 1 and 2 have fired.
- Where is the exit, and is the order placed? Decided before entry, resting at the target.
- What size? Never more than 10% of the stock's daily traded value.
What a checklist cannot do
Nothing above tells you the move will happen. The best single row in this course is 1.91×, on an 18-24% base, meaning roughly two thirds of even the best setups did not reach +20%. A checklist raises your odds from poor to fair. It does not make you right.
And every number here carries the same four limits, stated in full in Where ZEME is weak: observations overlap in time, delisted companies are absent from our price history, entries are modelled at the close, and a rule that worked in two periods can still stop working in a third.
Do this now, practical activity
- Write your checklist on paper, actually on paper. Five lines.
- Take the three stocks you are most interested in right now and score each one honestly.
- For any stock that fails step 1, write down what made you interested in it anyway. That sentence is your real strategy, and it is usually a story rather than a measurement.
- Then compare your list against today's signals and the Track Record, where ZEME's own picks are published with every miss included.
The graduation test for this course: you should now be able to look at any PSX chart and say which measured row it belongs to, and roughly what that row's odds were, before you have any opinion about the company.
Measured across all seven lessons: 67,098 point-in-time observations, 2016-01 to 2026-06, walk-forward TRAIN 2016-2022 / TEST 2023-2026. Three of the most widely taught rules on this market, support buying, the Bollinger squeeze, and moving-average position, measured at or below the base rate in the held-out years.