Alpha Engine — Daily Report (2026-03-20)

RunId: manual Scratchpads (2):

• .scratchpad/2026-03-20T17-51-18-554Z_cc355342b9c5.jsonl
• .scratchpad/2026-03-20T23-45-51-502Z_4cea1589c494.jsonl

1) P&L + risk

• Start cash (paper): $937.04
• End cash (paper, computed): $931.98
• Realized PnL: $-5.06 (-0.54%)
• Open exposure: $0.00 / $120.00 cap
• Open positions (today ledger): 0
• State now: cash $951.98, positions 0, exposureCounter $0.00

2) Trades list (proof)

• 2ddc63560069:0xe432:1773971832543 · (unknown) · 0xe432e4b79830fa534914b9e7767f904346b7bba3
  • entry 2026-03-20T01:57:12.544Z @ $5.99e-7
  • exit 2026-03-20T17:51:20.131Z @ $3.20e-7 · PnL $-4.66 · stop_loss_15%
• 2ddc63560069:0xea64:1773971832543 · (unknown) · 0xea644f7d935ced4915674f28fc2fe7380a663ba3
  • entry 2026-03-20T01:57:12.543Z @ $8.60e-7
  • exit 2026-03-20T17:51:20.049Z @ $8.25e-7 · PnL $-0.40 · time_30m

3) What worked vs didn’t (measured)

• Trades closed: 2 · Win rate: 0.0%
• Avg win: $0.00 · Avg loss: $-2.53
• Biggest winner: —
• Biggest loser: 2ddc63560069:0xe432:1773971832543 ($-4.66)

4) Learning log

• Change today: fixed Pulse chain id parsing + token address selection; added paper-trading skip diagnostics; added price sanity + major-token denylist + min liquidity filter; fixed tsx runner flag for Node >=20.6.
• Evidence: see paper_diag in scratchpads; it reports why entries were skipped (no price, caps, etc.).
• Next hypothesis: if skipped_major dominates, we need better filtering to target non-major tokens; if skipped_liquidity dominates, tune minLiquidityUsd; if skipped_no_price dominates, add pricing fallbacks.

Diagnostics (latest paper_diag)

RunId: 4cea1589c494

{
  "pulse_txs": 25,
  "skipped_no_addr": 0,
  "skipped_not_base": 0,
  "skipped_already_held": 0,
  "skipped_daily_loss_cap": 0,
  "skipped_exposure_cap": 0,
  "skipped_cash_cap": 0,
  "skipped_no_price": 0,
  "skipped_price_sanity": 0,
  "skipped_liquidity": 0,
  "skipped_activity": 0,
  "skipped_confidence": 8,
  "skipped_confirmation2": 17,
  "skipped_macro_riskoff": 0,
  "skipped_major": 0,
  "entries": 0
}

5) Next 24h plan

1. Add pricing fallback for common Base tokens + better Dexscreener pair selection (expected: fewer skipped_no_price). Guardrail: keep $/trade + exposure caps unchanged.
2. Enforce maxPositions consistently (expected: prevent over-entry). Guardrail: maxPositions from config.
3. If still zero entries, log first 5 candidate token addresses + chains for manual inspection.