CHART COLLEGE · THE CLASS

The oversold bounce

What the market says

"Down 20% in a month and finally bouncing hard: the selling is exhausted, buy the bounce."

The pattern, as the memory screens it: s.ret_20d <= -20 AND s.intraday_pct >= 5 AND s.rvol >= 2, liquid names ($50M traded on the anchor day, above $5), 2019 to the latest close. A name that is normally thin can qualify on a spike day, so the instances lean toward the names that spike. Questions this lesson answers:

Did the bounce hold the next session and the next week?
A month on, were more above the bounce day than an ordinary day?

What the charts did

More of them gave some back the next session than on an ordinary day: 57% of 1,983 closed below the anchor, against 49% of ordinary days. Median next-session move -1.4%.

A week on, 43% were above the anchor, fewer than the 55% of ordinary days; median -1.8%, and the middle 80% ran from -23.2% to +19.2% (ordinary: -6.1% to +6.8%).

A month on, 44% were above the anchor, fewer than the 53% of ordinary days; median -3.1%, and the middle 80% ran from -36.0% to +32.4% (ordinary: -12.6% to +13.5%).

Three months on, 49% were above the anchor, fewer than the 54% of ordinary days; median -0.7%, and the middle 80% ran from -47.6% to +68.6% (ordinary: -19.9% to +24.8%).

A year on, 47% were above the anchor, fewer than the 56% of ordinary days; median -10.3%, and the middle 80% ran from -81.4% to +141.3% (ordinary: -38.2% to +58.2%).

The week after this pattern is 3.3 times as wide as an ordinary week. That spread, not the median, is what the pattern tells you.

1,984 instances across 1,128 stocks, 2019-01-15 to 2026-09-17. The ordinary-day base rate is every liquid stock-day on one Wednesday a month over the same years.

Raw move from the anchor close	Instances	Share above	Median	Middle 80%	Ordinary day: share above	Ordinary day: middle 80%
Next session	1,983	43%	-1.4%	-12.2% to +11.3%	51%	-3.0% to +2.9%
1 week (5 sessions)	1,974	43%	-1.8%	-23.2% to +19.2%	55%	-6.1% to +6.8%
1 month (21 sessions)	1,956	44%	-3.1%	-36.0% to +32.4%	53%	-12.6% to +13.5%
3 months (63)	1,889	49%	-0.7%	-47.6% to +68.6%	54%	-19.9% to +24.8%
6 months (126)	1,823	45%	-6.3%	-64.9% to +89.5%	55%	-27.6% to +35.7%
1 year (252)	1,641	47%	-10.3%	-81.4% to +141.3%	56%	-38.2% to +58.2%

How it looks

Two dozen instances chosen at random, each cut at its own anchor: dark is the run-in, orange is what followed. Pick how far ahead to look; the tiles regroup by what happened.

Try it on any chart

Pick any ticker and date on the wall and the memory shows you the charts that looked like it. The class is the pattern in bulk; the wall is one chart at a time.

JSON · All lessons · built 2026-09-21