⚠️ Personal research and trading journal — not investment advice. The author does not provide licensed advisory services.
I spent several months analyzing transcripts from IBD (Investor's Business Daily) video commentary, trying to understand which phrases actually preceded profitable moves. The full corpus: 86 videos, thousands of stock mentions, tagged by phrase context.
Most phrase categories produced noise. Two produced a signal that survived my testing protocol.
The Signal
When IBD described a stock as being in a fresh base — specifically not extended, not yet broken out — AND described the technical pattern as a breakout in progress (cup-with-handle, VCP, flat base completing), the subsequent 20-day return was measurably different from the baseline.
The combination I tagged as PATTERN_BREAKOUT_FRESH: mentions where the pattern was explicitly named AND the stock was described as breaking out or at a pivot, AND no language indicating the stock was extended or "late stage."
Result: n=106 mentions over 2019-2026. Median return +0.36% at 20 days. +3.17 percentage point lift versus a baseline of mentions without these conjunction tags.
That +3.17pp is real: bootstrap CI excludes zero.
Adding RS≥80
When I filtered down to the PATTERN_BREAKOUT_FRESH mentions where the stock also had an RS Rating of 80 or above at the time of mention, the sample shrank but the signal strengthened:
- n=16 mentions
- 44% hit rate for ≥2R gain within 20 days (versus 22% baseline)
- 31% stop rate (stop = declined 5-7% from entry)
- Median return: +1.88%
Two examples that drove much of the right tail: PL (+214% maximum run-up from the mention date) and GRPN (+114%). These weren't the only contributors — the hit rate held after excluding the two largest outliers — but they illustrate the type of move this filter is designed to catch.
The combination RS≥80 + fresh pattern breakout description creates a small, concentrated set of high-signal candidates.
Why Buy-Zone Language Is a Warning Sign
Counter-intuitive finding from the same analysis: when IBD commentary included language indicating a stock was "within buying range," "in the buy zone," or "still buyable," the 20-day return was worse — significantly worse.
The buy-zone language group showed a -7.77pp lift compared to baseline. This is a large negative effect.
The interpretation: "still buyable" commentary tends to appear when a stock has broken out but hasn't moved decisively. The breakout may be stalling. IBD is describing a window that's about to close. By the time that language appears in a video, the optimal entry has already passed. Investors buying on "still in buy zone" mentions are often buying into a move that's running out of momentum.
This isn't guaranteed — individual cases vary — but the directional finding is clear across the sample: fresh breakout language is early. Buy-zone language is late.
What Changes in My Process
This finding is wired into my US scanner (scan_di_live_us.py): every candidate now shows an ibd_mentions_30d field showing how many times it appeared in recent IBD commentary. When that count is 3+, the mention type matters: was it "fresh" or "extended"?
The scanner doesn't automate the decision — it surfaces context. A stock showing up 4 times in IBD videos as "fresh breakout" with RS≥80 moves to the top of the review queue. A stock showing up 4 times as "still buyable" stays at the bottom, or gets filtered out entirely.
The practical filter: if I'm watching a stock that IBD mentioned and the language was buy-zone or extended, I wait for a new base to form before treating it as an active candidate.
Limitations
n=16 is a small sample. The 44% 2R hit rate comes from a specific 7-year period (2019-2026) that includes two significant bull markets. The filter hasn't been tested on a dedicated out-of-sample period — it was discovered and described in the same dataset.
I've tagged it as "production-ready post-freeze" in my notes, meaning I won't change the definition, but I'll track its performance in real-time before weighting it heavily. The 3+pp lift from the broader n=106 sample gives me more confidence in the directional finding than the n=16 hit rate alone.
The RS≥80 threshold is the same one that passed bootstrap walk-forward validation in a separate test (see: "The Filter That Finally Passed"). The combination with fresh-breakout language adds a second orthogonal filter that appears to concentrate the signal further.
Track. Study. Wait. Strike.
Personal research and trading journal — not investment advice. The author does not provide licensed advisory services. — MOEasymmetry
Draft 2026-06-12. Source: 86 IBD video transcripts, 2019-2026. PATTERN_BREAKOUT_FRESH tag = pattern named + breakout in progress + no extended/late-stage language. n=16 is the RS≥80 filtered subset. PL and GRPN cited as illustrative outliers; hit rate analysis excludes them and the directional finding holds.