Basket rule: dashboard v4.6 wording + new coin money study + 120 ideas (round J). 3 Oct 2026, Bangkok.

DASHBOARD (live, rule unchanged)
- Top banner "WHOLE MARKET WARNING: money is chasing brand-new coins" when the Monday share of top 1000 volume in coins
  listed under 30 days is in the top 20% of the prior 52 Mondays. Telegram on switch on and off.
- Extended word is now "TAKE % PROFIT": Take % Profit. (25% if not add, 75% if you doubled). Young baskets: TAKE 50% PROFIT.
- Two exits shown on every risk on card: trailing stop 20% (only once 30%+ above the buy) and a close under the 15 day low.

NEW COIN MONEY, 3 ITERATIONS (it1, it2, it_nc3; CMC point in time weekly snapshots Oct 2019 to Sep 2026)
1) Robustness. Live definition: next 4 weeks worse in all 4 eras; all years mean ON minus OFF: top 10 alts -9% (p 0.01),
   top 100 alts -11% (p 0.00), memes -18% (p 0.02), new coins -14% (p 0.01), BTC -4% (p 0.09, not reliable).
   p = circular shift test (keeps ON weeks clustered). Under 60 days and top 500/2000 give the same answer.
   Under 7 days does NOT work (p 0.5); count of new coins does not work; market cap share is weaker.
2) What it is. Share is high because new coin volume jumps (+310% in 4 weeks), not because old coin volume dries up (-9%).
   The coins are mostly big new launches and airdrops (UNI, SUSHI, CRV 2020; APE, APT 2022; ARB, SEI, MANTA 2023-24;
   TRUMP 2025), meme tagged share of their volume 0 to 2%. Effect grows with time: top 100 alts ON vs OFF median
   1w -0/+1, 2w -4/+1, 4w -8/+3, 8w -15/+3, 12w -17/+1. Strongest after a pump or flat month (-12 vs +9, -15 vs +1),
   absent after a fall (-2 vs -1). Weeks that stay ON (2nd week+) worse than the switch on week (-11 vs -4).
3) Trading use (bench, both eras). Live meme block beats no block (2023-26 +1901% vs +1503%, 2019-22 +3370% vs +3248%).
   Blocking ALL buys is worse (+1811%, +2643%); block all but majors worse; take profit or tighter exits when ON mixed.
   Live (Monday 28 Sep 2026): 0.25%, percentile 0. Research and live agree to 0.01%.

120 IDEAS (it_j: 30 breakout day K, 30 bottom Q, 30 extended E, 30 profit P), base v4.5 = 2023-26 +1901% dd -24,
2019-22 +3370% dd -24. Only 7 beat the base in both eras (K05, K12, K25, K26, P01, P03, P06):
- Sizing (K05 90d high 1.5x, K25 hot market 1.5x, K26 calm new coins 1.5x): a plain 1.5x size on every buy does better
  (+3182% / +7206%, dd -30) so the condition adds nothing; it is just more money in (idle cash is 72% of the time).
- Profit split grid (it_j2, it_j3): selling less is better in bulls and worse in chop, monotonically. 20%/70% gives
  +2059% (2025-26 half cagr 45 vs 47) and +4011%. No market condition (hot market, calm new coins, BTC above 200d)
  removes the chop cost. Kept 25%/75%.
- K12 (2+ other baskets broke out in 5 days): +2012% / +3447%, too small to matter.
Nothing shipped. Every other idea lost in at least one era. Mixed ones: Q30 bottoms never trim (old +9339%, new +1306%),
E14 trim on 2nd HOT (old +5334%, new +1021%), E24 z 2 anywhere (better late halves, worse totals).