Research angle
The question behind the note
Relative strength is attractive only until it becomes too extended for the review plan. The lab looks for a balance between leadership and manageable distance from reference areas.
What the evidence showed
What repeated review changed
A strong candidate should be compared with its sector and with a broad benchmark.
Overextension often appears when the candidate is far above its 20-day average and already widely visible.
The lab does not punish strength. It asks whether strength still offers a sensible review point.
Evidence checklist
Evidence to record before the outcome
- Compare recent performance with the sector ETF.
- Check distance from the 20-day average.
- Review RSI as a caution gauge, not a final answer.
- Ask whether a pullback would improve the setup or damage the structure.
Historical ticker results
IP, IVZ, DXCM, ZBH
| Ticker | Signal dates | 5-day raw | 5-day excess vs QQQ | 20-day raw |
|---|---|---|---|---|
| IP | 34 | 1.72% | 2.50% | -0.33% |
| IVZ | 39 | 0.58% | 0.95% | 4.58% |
| DXCM | 29 | -1.10% | -0.80% | 1.84% |
| ZBH | 38 | -0.98% | -1.74% | -3.90% |
V4.67.1 historical sample. Next-session-open entry assumption. A zero count means no matching observation, not a negative rating.
Failure modes
How the observation can be misused
- Assuming strength must continue because it has already been impressive.
- Ignoring risk distance when the chart looks clean.
- Confusing relative leadership with a durable business thesis.
- Promoting a descriptive pattern into a forecast without an out-of-sample test.
- Reporting the favorable horizon while omitting weaker horizons or the benchmark.
Open the full V4.67.1 model audit | Read the data methodology
Reader replication guide
How to evaluate Relative Strength Versus Overextension
Treat this case as a documented research example rather than a current recommendation. First identify the original question and observation date. Then separate company-reported facts, market data, model-derived results, and editorial interpretation. Each layer can update on a different schedule.
Reproduce the evidence
Open the cited primary source, confirm the covered period, preserve the raw input, and recalculate the reported comparison. If a historical model result is discussed, keep its entry assumption, horizon, benchmark, and overlapping-sample limitation attached.
Test an alternative explanation
Ask whether sector performance, rates, currency, acquisition effects, reporting-definition changes, or a small sample could explain the observation. Record the strongest contradictory evidence before accepting the interpretation.
A useful conclusion states what the example supports, what it does not establish, and which new filing or dated result would require an update. Preserving the earlier version prevents hindsight from silently changing the research question.