# Brief: "Your only job is to catch S2" (stage analysis), Ben 10 Oct 2026

Source: a Price Maturation chart (Weinstein stage analysis, Amgen 1990-97) shared by Ansem with the line "your only job is to catch S2". Ben: this could apply to new coins a few minutes old and to older coins.

## The idea in plain terms
Every chart cycles through 4 stages:
- **S1, base:** price goes sideways, the moving average flattens, volume is quiet.
- **S2, advance:** price breaks above the base on a volume spike, stays above a rising moving average, and makes higher highs and higher lows. **This is the only stage to hold.**
- **S3, top:** price stops making progress, the average flattens, and volume spikes without progress (distribution).
- **S4, decline:** price falls below a falling average. After S4 a new S1 base can form, and then a second S2 follows (the right side of the chart).

The original uses a 30- and 40-period moving average on weekly bars. For us the same rules apply at different bar sizes.

## Rules to test (start with these numbers, then let the data move them)
- **S1 base:** the last N bars have high and low within B% of each other, and the slope of the MA(30) is close to 0.
- **S2 entry:**
  - a bar closes at least K% above the base high
  - that bar's volume is at least V times the average base volume
  - the MA(30) slope turns positive
  - price closes above the MA(30)

  Start values: N = 20, B = 25% on new coins and 40% on older coins, K = 5%, V = 3.
- **Exits, test all four:**
  - (a) a close below the MA(30)
  - (b) the MA(30) slope flattens to 0 or below (S3 starts)
  - (c) a close below the MA(40) (slower)
  - (d) the base low as a hard stop, plus (a)
- **Do not buy S3 or S4.** Skip a coin if the price is below a falling MA, or if volume spikes while the price makes no progress.

## Three timescales = three tests
| Test | Coins | Bars | MA(30) covers | Owner |
|---|---|---|---|---|
| **S2-micro** | New coins, first 60 min | 10 s bars | 5 min | Analyst 1 |
| **S2-meso** | Coins 1 h to 3 days old | 1 min bars | 30 min | Analyst 1 |
| **S2-macro** | Older coins, 3+ days, still trading | 15 min and 1 h bars | 7.5 h / 30 h | Analyst 2 |
| **S2-second-wave** | Any coin that went through S2-S3-S4 and builds a new S1 base | the bar size it lived on | same | Analyst 2, combine with gMAK second-wave (#334) |

For S2-micro there is no real S1 on a coin that is minutes old. The test is whether the first pause (a mini base after the first push) followed by a breakout on volume beats our 5-Min Cap entry. Compare it on the same coins.

## How to judge
- Replay on the last 30 days of pools: trades, hit rate, mean and median result, results without the top 3, worst losing streak, and depth to exit at our size ($10 and $50).
- **Baselines:** buy at the same time without the S2 filter (dumb baseline), and for new coins our 5-Min Cap rule.
- Breakouts on one print do not count: the jump trap (one-print prices) and the exit-depth rule apply.
- No look-ahead: the base, MA and volume use only bars closed before the entry.

## Then
Whatever beats the baseline after fees goes to the Test Manager as a paper test with exit arms (a) to (d), gets a tab on the tests page, and goes to the Auditor. Data: our existing tick and swap tapes for new coins. For older coins, Data Engineer builds 1 min / 15 min / 1 h OHLCV bars for every pool alive more than 3 days (free sources only).