# LIVE PREP STATUS 2026-05-05

## Active Direction

Signed live is still rejected. The active live-prep path is:

```text
1. BIO/USDC portfolio_router_v4 paper-live virtual LP
2. no signed tx
3. same tuned v4 gates from the external light bundle
4. 14-day watch period before any $25-$50 signed pilot
```

The older “find 150-200% quarter champion” track is not live-ready. It is kept
only as research history.

## Imported External Agent Work

Source archive:

```text
../dex_lp_current_status_LIGHT_2026_05_05.tar.gz
```

Copied into `../DEX_full_cycle`:

```text
dex_platform/backtest/portfolio_router_v4.py
dex_platform/scripts/paper_live_virtual_lp_v1.py
dex_platform/scripts/run_portfolio_router_v4_tuned_best.sh
dex_platform/scripts/run_portfolio_router_v4_key_npzs.sh
```

Local fix applied after import:

```text
paper_live_virtual_lp_v1.py now forwards:
  --no-reentry-after-exit
  --health-max-total-dd-pct
```

Without this fix, the paper-live runner would not exactly match tuned v4 gates.

## BIO Router Validation

Offline tuned validation in `../DEX_full_cycle`:

```text
out: ../DEX_full_cycle/DEX_REPORTS/portfolio_router_v4_tuned_best_local_2026_05_05
BIO Apr corrected: +42.546% return, -9.923% MDD, strict pass
BIO Feb-Mar: 0%, idle, correct bear-regime block
BIO May: 0%, idle, truncated/latest-window block
```

Paper-live virtual smoke:

```text
BIO Apr corrected:
  out: ../DEX_full_cycle/DEX_REPORTS/paper_live_virtual_lp_bio_v4_tuned_local_2026_05_05
  return_total_pct: +42.546%
  mdd_total_pct: -9.923%
  strict_pass: true

BIO Feb-Mar:
  out: ../DEX_full_cycle/DEX_REPORTS/paper_live_virtual_lp_bio_febmar_v4_tuned_local_2026_05_05
  idle: true

BIO combined Apr17-May3:
  out: ../DEX_full_cycle/DEX_REPORTS/paper_live_virtual_lp_bio_combined_v4_tuned_local_2026_05_05
  idle: true

BIO May:
  out: ../DEX_full_cycle/DEX_REPORTS/paper_live_virtual_lp_bio_may_v4_tuned_local_2026_05_05
  idle: true
```

Active paper-live watchdog:

```text
tmux: bio_paper_live_watch
script: ../DEX_full_cycle/dex_platform/scripts/watch_paper_live_bio_v4_tuned.sh
input NPZ: ../DEX_full_cycle/DEX_DATA/fast_npz/bio_usdc_may2026_full.npz
out: ../DEX_full_cycle/DEX_REPORTS/paper_live_virtual_lp_bio_v4_tuned_watch
mode: virtual LP only, signed_transactions=false
```

Current watchdog state is idle on BIO May. This is expected.

## WETH/Oleg Conservative Baseline

Local WETH/Oleg reports:

```text
DEX_REPORTS/oleg_live_preparation/combined_weth_live_candidates.csv
DEX_REPORTS/oleg_tranche_deploy_fraction_weth_usdc_existing_quarter
DEX_REPORTS/revert_range_static_oor_weth_usdc_existing_quarter
```

Best static WETH/USDC 0.30% quarter baseline:

```text
strategy: wide_30_0.5
quarter return on deployed: ~8.28%
MDD: ~-15.51%
capacity share: negligible
```

Best reserve-aware Oleg tranche focused result:

```text
strategy: oleg_tranche_3p_0.25f_20_10_double_once_1x_24h_rolling_low_12h_0off
initial deploy fraction: 25%
lower/upper: 20% / 10%
quarter return on total capital: ~6.45%
MDD: ~-8.47%
fees earned: ~$8.62 per $100 total
inventory PnL ex-fees: ~-$2.17 per $100 total
time in range: ~99.38%
```

Interpretation: static WETH earns more in this historical quarter, but Oleg
tranche has a better drawdown profile and better matches the reserve discipline
from the chat.

## Data Downloader Status

The WETH/USDC monthly mainnet downloader hit RPC rate limits and stopped before
building monthly NPZ files. Do not restart it aggressively with the same RPC.
Use slower chunking/sleep or another endpoint.

Fallback added after the 429 failure:

```text
dex_platform/data_collectors/full_cycle_import/fetch_geckoterminal_pool_ohlcv_v1.py
dex_platform/data_collectors/full_cycle_import/build_ohlcv_proxy_npz_v1.py
```

The WETH monthly scripts now fall back to GeckoTerminal OHLCV and build separate
`*_ohlcv_proxy.npz` files if exact RPC event collection fails. These proxy NPZs
are explicitly marked as `source_type=geckoterminal_ohlcv_proxy`.

Important limitation: OHLCV proxy data is acceptable for rough regime filters,
pool screening, and router dry-runs. It is not exact swap-event fee-share or
capacity evidence, so it must not be used as the final live proof.

Created fallback WETH/USDC 0.30% files:

```text
DEX_DATA/ohlcv_proxy/mainnet_WETH_USDC_UNIV3_030_feb2026_quote_hour.csv
DEX_DATA/ohlcv_proxy/mainnet_WETH_USDC_UNIV3_030_mar2026_quote_hour.csv
DEX_DATA/ohlcv_proxy/mainnet_WETH_USDC_UNIV3_030_apr2026_quote_hour.csv
DEX_DATA/fast_npz/mainnet_WETH_USDC_UNIV3_030_feb2026_ohlcv_proxy.npz
DEX_DATA/fast_npz/mainnet_WETH_USDC_UNIV3_030_mar2026_ohlcv_proxy.npz
DEX_DATA/fast_npz/mainnet_WETH_USDC_UNIV3_030_apr2026_ohlcv_proxy.npz
```

Sanity check for orientation: quote OHLCV gives WETH prices around
`$1,816-$2,457`; base OHLCV gave ~$1 and is not suitable for this replay
orientation.

The current priority remains BIO paper-live virtual logging plus exact event
collection retries at a slower rate.

## Active tmux Sessions

```text
bio_paper_live_watch
oleg_tranche_tune
```

The focused tranche result already gives the main WETH/Oleg insight.