# AGEN — The 5 Sessions Before 9/28 (9/21 – 9/25)

Same format as the 9/28 report: premarket-to-close chart (VWAP, the bot's buys/sells), price vs the open and VWAP through the day, volume by hour, the bot's scanner ranks, its entry checks, and its trades. Source: Alpaca SIP 1-minute bars; the live bot's scanner lists, decision logs and trade logs.

## The week at a glance

| Day | Open → close | Range | Day's high (time) | On the bot's list? | Bot trades |
|---|---|---|---|---|---|
| 9/21 | **+6.8%** | 13.2% | 8.60 (15:09) | yes, 9:28 rank 17 | 2 trades: +$50.60, −$9.88 |
| 9/22 | **+5.3%** | 8.1% | 8.82 (13:57) | **no** | none (bot also crashed at 10:49) |
| 9/23 | −3.4% | 3.7% | 8.70 (9:30) | no | none |
| 9/24 | +2.2% | 4.8% | 8.59 (15:00) | no (11 rescans) | none |
| 9/25 | −0.4% | 3.3% | 8.68 (9:41) | no (13 rescans) | none |
| 9/28 | **+8.3%** | 14.6% | 9.71 (14:57) | from 9:40 (rank 2 → 1) | 1 trade: +$51.72 |

## What the week shows

1. **Its two big days looked the same.** On 9/21 and 9/28 the bot bought early and well (9/21 at $7.91, 9:47; 9/28 at $8.84, 11:35), then **sold early** (9/21 at 13:21, $8.11, by the "deterioration" exit; 9/28 at 12:23, $9.12, by the lock-in-profit exit). Both days the stock made its **high in the afternoon** (15:09 and 14:57). Held to the close: 9/21 ≈ +$106 instead of +$50.60; 9/28 ≈ +$111 instead of +$51.72.
2. **9/22 was missed entirely** — a +5.3% open-to-close day (high 13:57) with a −0.8% gap, so it never made the list. Small-gap, steady-climb stocks don't score well on the current scanner (see the scanner ideas: opening relative volume).
3. **9/23 – 9/25: quiet base, correctly ignored.** Ranges of 3.3–4.8% — the three-day base AGEN broke out of on 9/28.
4. **9/21's list data was wrong:** the 9:28 scan showed a +9.8% gap for AGEN, but the real gap was −0.6% (the two-sessions-old data bug fixed on 9/25). It made the list for the wrong reason that day.

---