# analyst3 on scout items 22-23 (16 Oct)
## 23 same-name decoys (s23.py, births.db symbols x misses.csv)
- 141 missed runners; 73 have a symbol+birth time in births.db (symbol coverage only 61k of 495k pools).
- 52/73 (71%) share their symbol with another token; 43 have an earlier same-symbol token, 33 a later one.
- Base rate: 70.8% of ALL 21k tokens share a symbol with another token. Runners 71% -> lift 1.0. Same-name is the norm (symbols like AI, PAPER, QI), so it neither flags runners nor decoys.
- Verdict: no signal as a filter. A real decoy test needs name+symbol+image similarity plus "did our bot buy the earlier copy"; bot buy logs not joined, EV not shown. Not matured.
## 22 one-contract bundler / curve filled <=3 blocks
- Needs per-pool block-level opening buys by one contract; we have launcher/kit and funder data (bundle_fund.db) but no "filled in <=3 blocks" feature. Already covered by rug-gate work (exec/research); handed over as a feature idea, no test run today.