# #559 Radar cross-event history (ppfeeds, 10 Oct 2026)
**File:** research-box `/opt/sw/radar/radar_hist.jsonl` (18,359 ev=cross lines, 13 Sep 00:40 BKK to 9 Oct 07:00 UTC, sorted by t). Scripts: /opt/crew/work/ppfeeds/radar_hist.py, radar_hist2.py, radar_hist3.py. Same field names as radar.jsonl (pool, venue, token, th, cap, cap_first, age_s, depth_usd, thin, n_senders, fee, hooked, quote, backfill=true) plus `src`. Radar's own 72 h rule applies: crosses only within 72 h of the pool's first swap. Dedupe key (pool, th). From 7 Oct on use radar.jsonl for the live part (overlap is harmless).

| src | covers | depth/thin | n_senders |
|-|-|-|-|
| r6 (r6weth/r6nat/r6usd tapes) | pools with first swap 21 Sep - 9 Oct 07:00 UTC, all of v3/v4 WETH, native, USDG, SUSD, hooked or not | yes (virtual depth from swap liquidity, same formula as radar v4) | yes (distinct swap senders so far, cap 64) |
| v3hist | v3 + pons_grad born 13-21 Sep | yes (pool_eth_after x ETH) | null |
| xall | v4 born 13-21 Sep, **only the ~1.2k pools in the E1 candidate set** (the E1 swap tape was never built for all 13-21 Sep v4 pools) | null (thin = null, treated as non-thin below) | null |

Known gaps: v2 pools not included (no tape); v4 13-21 Sep is partial (candidate set); no launcher/symbol/cap_5m/tx; caps are raw prints with the E1 glitch rule (cap > $20B or a 20x print the next print undoes is dropped), so no radar lag fields.

## First counts (pool-level, non-thin, climber = cap_first < 0.5 x level)
| venue | cross 1M | climbers 1M | climbers 5M | 300k non-thin age<1h cap_first<150k | of those reached 1M |
|-|-|-|-|-|-|
| v3 | 393 | 191 | 59 | 468 | 179 |
| v4:native | 537 | 74 | 13 | 148 | 33 |
| v4:native:hook | 287 | 100 | 7 | 166 | 89 |
| v4:usdg | 2571 | 6 | 3 | 2 | 1 |
| v4:usdg:hook | 199 | 8 | 4 | 20 | 4 |
| v4:weth (nohook) | 268 | 45 | 1 | 371 | 44 |
| v4:weth:hook | 226 | 150 | 34 | 201 | 140 |
| v4:susd | 9 | 0 | 0 | | |

So USDG nohook confirms your finding (6 climbers of 2571 crossers). Quick read of your pass rule: native nohook (74 climbers, P about 0.22) and WETH nohook (45, 0.12) clear the climber count but not P(1M) >= 0.5; the hooked WETH (150, 0.70) and hooked native (100, 0.54) clear both (hooked pools were the focus of other tests). Example: `python3 -c "import json;[print(l) for l in open('/opt/sw/radar/radar_hist.jsonl') if '\"v4:weth:hook\"' in l][:3]"`.