# Virality Expert: the three checks on the hour-4 rule (V1.1, V1.2, V1.3, plus v2.2)

## READ THIS FIRST
- **Verdict: the hour-4 rule FAILS. Do not put it in a paper money book.** Priced swap by swap with real pool depth and pulls, it only works in the first half. The second half and the latest two weeks lose money in every version I tried.
- **V1.1 real money pricing (134 coins, $500 per trade):**
  - Same 115 coins as my round 2: average +84% (hourly) drops to **+31%** (swap by swap). The middle trade stays +25%.
  - First half: +35% average, +34% middle trade (78 trades).
  - **Second half: +2% average, −26% middle trade (56 trades).** Without the best 3 trades: −12%. Only the 37 second-half coins from round 2: +24% average, −11% middle trade.
  - **21 Sep to 4 Oct (the v3 rule's period): −15% average, −39% middle trade (31 trades).**
  - Why hourly was too kind: these coins drop 50% inside the hour and come back. Hourly closes never see the dip, so the stop never fired. Example: 0xf15aab29c3d22cf2b469d4a9b0c0a72f31ba0711 was +1815% hourly, but swap by swap it fell to 0.48x within 22 minutes of entry. That is a real stop at −51%.
- **Lane test: not a honeypot lane.**
  - After an outside $300 to $800 buy, a pull (half or more of the last hour's liquidity removed) came within 30 s in 60 of 149,060 cases (0.040%).
  - After any swap: 0.027%. At a random moment: 0.011%.
  - So it happens 1.5x more than after any swap, but only about 1 time in 2,500. 7 of 136 coins had it at all.
- **V1.2 overlap with v3: almost none.** Only 1 of the 58 v3 trades is in my hour-4 list: 0xf15aab29c3d22cf2b469d4a9b0c0a72f31ba0711, same pool, 0x4cfb58fa2a846b3b9ed8c1dc0124d1092ccf0c65. The v3 rule bought it at about $1M, 0.7 h old, and stopped out at 0.54x. My rule also stops out at real prices (−51%).
  - What the hour-4 rule adds on coins v3 did not buy, in the same period: **30 trades, −13% average, −34% middle trade, 33% winners.** So it adds nothing.
- **V1.3 bots before the gate: not a filter.** On the 58 v3 trades, copy bot buys before the gate are almost absent: median 1 of about 500 buyers (0.2%), for winners and losers alike. The split flips sign depending on how you count (count: more bots did better; share: worse in the second half only). Nothing for the Coder.
- **v2.2 crew repeat wallets removed:** they are under 1% of new buyers after $500k (0.1 to 0.6% by tier). New buyers per hour barely move, and the hour-4 rule is unchanged: 119 trades, +83% hourly, both halves the same as before. So crews do not explain my slow-burn result.

## V1.1 table (swap by swap, $500 fill against the pool, pull = exit at what the pool pays)
| Version | All 134 | Round 2 set (115) | First half | Second half | 21 Sep to 4 Oct |
|---|---|---|---|---|---|
| Old: hourly closes, no pulls | n/a | +84% / +25% | +93% / +41% | +65% / −11% | n/a |
| Swap checks, stop −50% (base) | +21% / +5% | +31% / +25% | +35% / +34% | +2% / −26% | −15% / −39% |
| Swap checks, stop −70% | +23% / +23% | +34% / +39% | +41% / +48% | −3% / −34% | −19% / −39% |
| Hourly checks, pulls live | +39% / −7% | +54% / +13% | +72% / +23% | −7% / −34% | −27% / −52% |
| Hourly checks, only 90% pulls exit | +41% / −2% | +56% / +14% | +74% / +28% | −5% / −34% | −23% / −52% |

- Each cell is average / middle trade.
- "Second half" here covers all 56 late coins, including 19 that had no hourly data in round 2.
- Exits (base): trail 46, stop 44, time 28, pull 16. Pull exits average −29% (second half −46%, middle −95%).
- Entry cost: the median price impact of a $500 buy is 1.5% (p90 2.5%). The median pool has $84k on the quote side. Fees are the real pool fee, with 1% a side assumed on hooked pools that show a 0 fee.
- By outcome: coins that later reached $5 to $20M made +47%, and coins that stalled under $1M lost −38%. The rule cannot tell them apart well enough at hour 4.

## Method
- Pool: for each coin, I sampled 10 transactions between hours 3.5 and 4 after $500k and took the pool with the most swaps for that token. v4 pool keys came from the Initialize event.
- Data: every Swap and liquidity event (ModifyLiquidity, Mint/Burn, Sync/Burn) of that pool, from 1 h before entry to 7 days after, from the public RPC. In total 2,666,363 swaps.
- Fills: a constant product on the pool's live liquidity and price at that swap (exact for full range, close for wide ranges).
- Prices: ETH from Coinbase hourly. USDG = $1. Other quotes (stock tokens, memes) are priced each hour from the coin's hourly cap.
- Pull: a liquidity removal that leaves 50% or less of the last hour's peak. We sell right after it, into what is left.

## Caveats
- The 569 coin set is still survivor-skewed (the runner tracker). Real live results would be lower, not higher, so the fail verdict stands.
- One pool per coin. Some coins also traded in side pools; we only sell in the main one.

## What survives from round 2
- The slow-burn map (winners keep pulling new buyers for days, bots before $500k is a bad sign) is description, not a trade. It still holds with crew wallets removed.
- I am moving on to V3 (spread patterns) and V4 ($5M vs $20M), and the side task (X before the v3 gate). Every new idea gets swap-by-swap pricing from the start. The swap-by-swap pricer is built and can go to the Coder on request.

## First principles score
- **8 of 10.** Every trade is priced from the pool's own swaps and liquidity, with our $500 moving the price and pulls paying what the pool really pays.
- Missing: the exact tick map for narrow concentrated ranges, side pools, and the fair all-venue coin list.

Version: v1, 6 Oct 2026, 14:30 Bangkok. Latest version: http://159.223.88.168/hx-c9ois/sw/sw-virality-v11.md