2026-08-24
Three names have a standout record for the last week of August, and the biotech is the one with the strange shape. HALO's worst trip through Aug 25–31 since 2004 cost 4.48%, which is a remarkably polite bad year for a company that can move 17% in a week, and not one of those years had an earnings report sitting inside the window. That makes it a calendar record rather than a print reaction wearing a calendar costume.
VLO has the strongest typical showing for Aug 24 in our universe since 2000: up 16 of 18 years (89%), with a median move of +1.70% (average +1.33%). Best year +5.0% (2015); worst -3.7% (2000).
AVGO has the strongest typical showing over the next 5 trading sessions from Aug 24 in our universe since 2009: up 13 of 17 years (76%), with a median move of +3.11% (average +2.55%). Best year +10.4% (2015); worst -6.4% (2022).
The Seasonal Watchlist
The names that have owned Aug 25 – Aug 31, year after year:
| Ticker | Hit rate | Median | Sample |
|---|---|---|---|
| AVGO | 76% | +3.11% | 17 yrs |
| HALO | 77% | +2.58% | 22 yrs |
| SIMO | 76% | +1.78% | 21 yrs |
The Seasonal Board ranks every qualifying name for the month, each with its own window.
AU didn't drift into overbought, it sprinted: up 27.2% in five sessions to the most stretched reading on the board. What's funny is how little that buys you. From readings like that, the next week has been a coin flip for it, and the beaten-up utility a few rows down carries the better forward record of the two.
Most overbought
| Ticker | Level | RSI Percentile | Close | Last 5 Days |
|---|---|---|---|---|
| AU | Extreme overbought | 100th | $121.22 | +27.2% |
| A | Extreme overbought | 100th | $159.00 | +7.0% |
| SEIC | Extreme overbought | 100th | $109.53 | +3.0% |
Most oversold
| Ticker | Level | RSI Percentile | Close | Last 5 Days |
|---|---|---|---|---|
| NI | Extreme oversold | 0th | $40.62 | -4.4% |
| TJX | Extreme oversold | 0th | $140.53 | -7.6% |
| PEG | Extreme oversold | 0th | $72.61 | -4.5% |
What's happened after this before
| Ticker | Next 3 Days | Next 7 Days | Next 30 Days | Prior Times |
|---|---|---|---|---|
| NI | 58% · +0.5% | 60% · +0.7% | 61% · +1.7% | 473 |
| it did tend to bounce — higher 60% of the time on average across the three windows | ||||
| TJX | 56% · +0.3% | 61% · +0.7% | 62% · +1.8% | 453 |
| it did tend to bounce — higher 60% of the time on average across the three windows | ||||
| SEIC | 54% · -0.1% | 56% · +0.1% | 58% · +0.7% | 478 |
| it kept going — higher 56% of the time, so "overbought" has not meant due to fall for this name | ||||
| PEG | 53% · +0.0% | 55% · +0.1% | 59% · +0.8% | 502 |
| it did tend to bounce — higher 56% of the time on average across the three windows | ||||
| A | 53% · +0.1% | 53% · +0.2% | 59% · +1.3% | 494 |
| it kept going — higher 55% of the time, so "overbought" has not meant due to fall for this name | ||||
| AU | 51% · +0.3% | 50% · +0.6% | 51% · +1.1% | 492 |
| a coin flip — this level has told you nothing about this name | ||||
Each cell: how often the stock was higher, and its average move, every prior time it hit this level. Under the hood the windows are 3, 5 and 20 trading sessions. The line under each name is what that record says.
Worth sitting with: SEIC, A are on this list for being at an extreme, and their own history says the extreme has not been the thing to trade against.
The Levels Map gives you the whole board: every name at an extreme, with where it stalls.
The line worth staring at is HL. Getting knocked back from a swing high reads like a warning, and HL's own history has been the opposite: five days after those rejections it sat up 2.39% at the median, the best forward record anywhere on this card and the wrong direction for a warning sign. Worth noting MRVL lost the 250 round the same session and reports Thursday night, so that level gets re-tested by the print rather than by the tape.
XEL broke below its volume shelf at $79.04 (closed -3.5% through it).
Across 51,813 prior times a stock did this, it was higher five sessions later 54% of the time — the move through the level has not been the direction to follow.
Also at a level
BABA broke below a round number at $130.00 (closed -8.2% through it).
MRVL broke below a round number at $250.00 (closed -5.2% through it).
EQIX got rejected at a swing high at $1099.72.
HL got rejected at a swing high at $21.29.
APA got rejected at its 52-week high at $45.05.
FRO got rejected at its 52-week high at $45.17.
The Levels Map carries every name being tested right now, not just this one.
Everyone is pointed at Wednesday night. The part of NVDA's record that nobody brings up is what has happened on the misses: the median move on those days is exactly 0.0%, so the disaster scenario has historically been a shrug. PDD, which printed this morning, is the opposite animal entirely and gives back double digits when it comes up short.
The biggest names reporting in the next 7 days — and what each has typically done on report day:
| Wed Aug 26 | NVDA | $5454B | after close | median swing ±3.0% over 107 reports; typically +0.9% on a beat, 0.0% on a miss |
| Thu Aug 27 | RY | $301B | before open | median swing ±1.6% over 107 reports; typically +0.5% on a beat, -1.7% on a miss |
| Wed Aug 26 | CRWD | $221B | after close | median swing ±6.0% over 29 reports |
| Thu Aug 27 | TD | $205B | before open | median swing ±1.3% over 107 reports; typically +0.3% on a beat, -1.0% on a miss |
| Thu Aug 27 | MRVL | $199B | after close | median swing ±5.5% over 104 reports; typically +3.8% on a beat, -1.6% on a miss |
| Wed Aug 26 | CRM | $162B | after close | median swing ±4.9% over 88 reports; typically +1.0% on a beat, -3.7% on a miss |
| Tue Aug 25 | BMO | $130B | before open | median swing ±2.0% over 107 reports; typically +0.5% on a beat, -2.3% on a miss |
| Mon Aug 24 | PDD | $121B | before open | median swing ±13.6% over 32 reports; typically +3.5% on a beat, -10.5% on a miss |
| Thu Aug 27 | CM | $113B | before open | median swing ±2.1% over 107 reports; typically +1.0% on a beat, -2.6% on a miss |
| Tue Aug 25 | BNS | $112B | before open | median swing ±1.4% over 107 reports; typically +0.3% on a beat, -1.4% on a miss |
Get issues like this by email: