# Browser continuation — DEX live-candidate search and paper-live plan — 2026-05-07 ## Scope used I deliberately reduced the search scope to bounded, inspectable jobs: - refine CHECK/USDC 2% Q1 exact-event candidate; - test CHECK/USDC 0.25% newer April/May pool against the same candidate set; - screen SYND/USDC high-fee pool as a possible alternative; - re-test BIO/USDC router on April, Feb-Mar, May, and fresh paper-live data; - smoke-run paper-live commands for BIO and CHECK. No signed transaction work. No wallet work. No secrets. ## Main decision There are two S2 paper-live tracks now: 1. **BIO/USDC macro-router, $600 virtual capital** — best full-size candidate. It passed April and the fresh paper-live replay window. It correctly idled in Feb-Mar and May under the tested macro gate. 2. **CHECK/USDC 2% static/periodic candidate, $25 first capital** — best small-cap candidate. It improved versus the old 90/1/336h candidate, but the existing router paper-live implementation does not yet reproduce the static replay cleanly, so run it as static replay watcher until a dedicated static virtual LP runner is implemented. ## CHECK/USDC 2% refined Q1 result Best bounded candidate: ```text pool: base_CHECK_USDC_AERODROME_2PCT strategy: periodic_85_0.5_336h capital: 25 first, 50 max for paper test range: lower 85%, upper 0.5% rebalance: periodic every 336h Q1 return: +36.66% Q1 MDD: -19.16% p95 share: 1.39% p99 share: 4.29% max share: 7.69% ``` Top strict Q1 rows: | strategy | capital_usd | return_pct | mdd_pct | p95_liquidity_share_pct_when_in_range | p99_liquidity_share_pct_when_in_range | max_liquidity_share_pct_when_in_range | score | |:---------------------|--------------:|-------------:|----------:|----------------------------------------:|----------------------------------------:|----------------------------------------:|--------:| | periodic_85_0.5_336h | 25.0000 | 36.6615 | -19.1551 | 1.3900 | 4.2897 | 7.6876 | 36.5415 | | periodic_85_1_336h | 25.0000 | 36.4018 | -19.2943 | 1.3822 | 4.2663 | 7.6472 | 36.2818 | | periodic_85_1.5_336h | 25.0000 | 36.1655 | -19.4315 | 1.3634 | 4.2435 | 7.6077 | 36.0455 | | periodic_85_2_336h | 25.0000 | 35.9507 | -19.5718 | 1.3561 | 4.2214 | 7.5695 | 35.8307 | | periodic_85_0.5_336h | 50.0000 | 35.7122 | -19.1764 | 2.7416 | 8.2257 | 14.2763 | 35.5922 | | periodic_85_3_336h | 25.0000 | 35.5195 | -19.8463 | 1.3416 | 4.1778 | 7.4941 | 35.3995 | | periodic_85_1_336h | 50.0000 | 35.4618 | -19.3156 | 2.7263 | 8.1826 | 14.2064 | 35.3418 | | periodic_85_1.5_336h | 50.0000 | 35.2345 | -19.4527 | 2.6898 | 8.1407 | 14.1384 | 35.1145 | | periodic_85_2_336h | 50.0000 | 35.0282 | -19.5930 | 2.6755 | 8.1000 | 14.0724 | 34.9082 | | periodic_85_3_336h | 50.0000 | 34.6137 | -19.8675 | 2.6475 | 8.0198 | 13.9419 | 34.4937 | Monthly selected CHECK breakdown, $25: | strategy | period | ret_pct | mdd_pct | p95_share_pct | p99_share_pct | score | |:---------------------|:---------|----------:|----------:|----------------:|----------------:|---------:| | periodic_90_1_336h | feb | -0.6543 | -1.9607 | 0.0303 | 0.0699 | -0.6743 | | periodic_85_0.5_336h | feb | -0.6676 | -2.1057 | 0.0339 | 0.0810 | -0.6876 | | periodic_95_1_168h | feb | -0.6757 | -1.6086 | 0.0282 | 0.0638 | -0.7357 | | periodic_80_5_336h | feb | -1.1633 | -3.1365 | 0.0386 | 0.0828 | -1.1833 | | periodic_70_5_336h | feb | -1.4112 | -3.7927 | 0.0466 | 0.0999 | -1.4312 | | periodic_60_5_336h | feb | -1.7202 | -4.6050 | 0.0564 | 0.1209 | -1.7402 | | periodic_85_0.5_336h | mar | 1.4854 | -16.4921 | 1.6490 | 7.7254 | 1.4454 | | periodic_95_1_168h | mar | 1.4254 | -12.7384 | 1.3071 | 6.2032 | 1.3454 | | periodic_90_1_336h | mar | 1.1872 | -15.0156 | 1.4745 | 6.9533 | 1.1472 | | periodic_80_5_336h | mar | -0.1624 | -20.2725 | 1.7365 | 8.1087 | -0.7475 | | periodic_70_5_336h | mar | -0.2457 | -24.4268 | 2.0889 | 9.6274 | -9.1393 | | periodic_60_5_336h | mar | -0.3710 | -29.5240 | 2.5197 | 11.4315 | -23.7537 | | periodic_70_5_336h | apr | 32.9405 | -20.2059 | 1.1261 | 5.8158 | 32.4887 | | periodic_95_1_168h | apr | 30.9143 | -19.2567 | 0.5995 | 3.1591 | 30.8343 | | periodic_60_5_336h | apr | 39.4810 | -24.5482 | 1.3696 | 6.9862 | 30.3447 | | periodic_80_5_336h | apr | 27.5185 | -16.7015 | 0.9317 | 4.8517 | 27.4785 | | periodic_85_0.5_336h | apr | 24.5387 | -13.8591 | 0.8763 | 4.5634 | 24.4987 | | periodic_90_1_336h | apr | 22.0276 | -12.5622 | 0.7834 | 4.0961 | 21.9876 | Hard read: `85/0.5/336h` is the new CHECK candidate by Q1 score, but it has April-driven edge and Feb is slightly negative. That is acceptable for a small $25 paper-live, not for scaling. ## CHECK/USDC 0.25% newer pool check The same candidate shape also works on the newer 0.25% CHECK pool sample, but capacity stats include extreme max-share spikes. p95/p99 are fine, max share is a warning. | strategy | capital_usd | return_pct | mdd_pct | p95_liquidity_share_pct_when_in_range | p99_liquidity_share_pct_when_in_range | max_liquidity_share_pct_when_in_range | score | |:---------------------|--------------:|-------------:|----------:|----------------------------------------:|----------------------------------------:|----------------------------------------:|---------:| | periodic_85_0.5_336h | 25.0000 | 20.0104 | -11.8072 | 0.0719 | 0.3751 | 100.0000 | -17.5096 | | periodic_90_1_336h | 25.0000 | 18.4072 | -10.7234 | 0.0649 | 0.3351 | 100.0000 | -19.1128 | | periodic_85_0.5_336h | 50.0000 | 16.6883 | -12.0534 | 0.1433 | 0.7473 | 100.0000 | -20.8317 | | periodic_90_1_336h | 50.0000 | 15.1823 | -10.7271 | 0.1283 | 0.6680 | 100.0000 | -22.3377 | | periodic_95_1_168h | 25.0000 | 4.9664 | -9.3817 | 0.0242 | 0.3109 | 100.0000 | -32.5936 | | periodic_95_1_168h | 50.0000 | 2.1033 | -9.3821 | 0.0481 | 0.6199 | 100.0000 | -35.4567 | Decision: useful confirmation, but not enough proof. Continue collecting fresh CHECK events. ## SYND/USDC screen SYND is rejected for now. Even with 0.95% fee, the valid-capacity rows had deep drawdown and weak return. | strategy | capital_usd | return_pct | mdd_pct | p95_liquidity_share_pct_when_in_range | p99_liquidity_share_pct_when_in_range | score | |:-------------------|--------------:|-------------:|----------:|----------------------------------------:|----------------------------------------:|---------:| | periodic_90_1_168h | 25.0000 | 1.1607 | -30.6827 | 0.5917 | 0.7788 | -20.2846 | | periodic_90_1_168h | 50.0000 | 1.1206 | -30.6841 | 1.1764 | 1.5455 | -20.3275 | | periodic_90_1_168h | 100.0000 | 1.0440 | -30.6868 | 2.3254 | 3.0438 | -20.4095 | | periodic_90_2_168h | 25.0000 | 0.9916 | -31.0228 | 0.5039 | 0.7742 | -21.1341 | | periodic_90_2_168h | 50.0000 | 0.9518 | -31.0242 | 1.0028 | 1.5365 | -21.1766 | | periodic_90_2_168h | 100.0000 | 0.8760 | -31.0270 | 1.9856 | 3.0264 | -21.2579 | | periodic_90_1_336h | 25.0000 | 0.8464 | -31.3418 | 0.5038 | 0.7764 | -21.8771 | | periodic_90_1_336h | 50.0000 | 0.8063 | -31.3431 | 1.0024 | 1.5408 | -21.9198 | ## BIO/USDC router retest BIO router result table: | case | ret_pct | mdd_pct | strict_pass | entries | health_exits | pnl_mdd | |:-------------------------|----------:|----------:|:--------------|----------:|---------------:|----------:| | baseline_apr | 42.5463 | -9.9234 | True | 3 | 2 | 4.2875 | | baseline_febmar | 0.0000 | 0.0000 | False | 0 | 0 | 0.0000 | | baseline_fresh | 9.6549 | -4.3967 | True | 1 | 0 | 2.1959 | | baseline_may | 0.0000 | 0.0000 | False | 0 | 0 | 0.0000 | | macro2_dd5_apr | 34.9479 | -12.2178 | True | 3 | 2 | 2.8604 | | macro2_dd5_febmar | 0.0000 | 0.0000 | False | 0 | 0 | 0.0000 | | macro2_dd5_fresh | 9.6549 | -4.3967 | True | 1 | 0 | 2.1959 | | macro2_dd5_may | 0.0000 | 0.0000 | False | 0 | 0 | 0.0000 | | macro2_micro2_dd5_apr | 34.9479 | -12.2178 | True | 3 | 2 | 2.8604 | | macro2_micro2_dd5_febmar | 0.0000 | 0.0000 | False | 0 | 0 | 0.0000 | | macro2_micro2_dd5_may | 0.0000 | 0.0000 | False | 0 | 0 | 0.0000 | Decision: keep BIO macro router as the main $600 paper-live candidate. The tested macro-positive gate idles in Feb-Mar and May, and trades the fresh paper-live window. ## Paper-live smoke results | case | ret_pct | mdd_pct | strict_pass | entries | exits | health_exits | pnl_mdd | |:--------------------------|----------:|----------:|:--------------|----------:|--------:|---------------:|----------:| | bio_fresh_macro_router | 9.6549 | -4.3967 | True | 1 | 0 | 0 | 2.1959 | | check_q1_85_0p5 | 21.2752 | -25.5348 | False | 7 | 6 | 0 | 0.8332 | | check_q1_85_0p5_dh12_dd15 | 27.0696 | -24.6437 | False | 7 | 6 | 0 | 1.0984 | | check_q1_85_0p5_dh6_dd15 | 26.2054 | -24.6166 | False | 7 | 6 | 0 | 1.0645 | | check_q1_85_0p5_dh6_dd20 | 26.2054 | -24.6166 | False | 7 | 6 | 0 | 1.0645 | Important: CHECK via the current `portfolio_router_v4` paper-live path failed strict MDD. Static replay for CHECK is cleaner than router paper-live. Therefore CHECK should use the static replay watcher until a dedicated `paper_live_static_periodic_lp_v1.py` is added. ## Promotion status ```text BIO/USDC: S2 paper-live candidate, $600 virtual only. CHECK/USDC 2%: S2 paper-live candidate, $25 virtual only, static replay watcher. CHECK/USDC 0.25%: S1 confirmation candidate, needs fresh collection. SYND/USDC: rejected for now. WETH/USDC 0.30: baseline only, not business live edge. ``` ## Next action Run two supervised paper-live watchers: 1. `scripts/watch_paper_live_bio_macro_router_v1.sh` 2. `scripts/watch_paper_live_check_static_85_0p5_336h_v1.sh` Let them run for 3-7 days. Do not sign live tx. Do not scale CHECK above $25 before fresh paper-live evidence.