protocols/B-autosell-exit.json · v1

B · AutoSell: trigger-to-inclusion, exit probability, cost and execution quality

For exit intents triggered by a pre-defined condition, the Rout arm lands within the window more often, at no higher cost per landed exit and no worse implementation shortfall, than the control path.

draft

10 values must be agreed before this can run

$.parameters.bootstrap_resamples · $.parameters.landing_window_slots · $.parameters.max_abs_clock_offset_ns · $.parameters.target_n_per_arm · $.parameters.budget_lamports_max · $.parameters.tip_lamports_by_arm · $.parameters.cluster_slots · $.acceptance.min_landing_difference · $.acceptance.max_cost_ratio · $.acceptance.max_shortfall_difference_bps

Check with python3 -m rbench.prereg check protocols/B-autosell-exit.json.

Design

Why it matters: Auto-sell and counter-trade exits that still land, at a bounded price, when the market moves fast.
Unit: One exit intent on a position held by the experiment's wallets. One arm per intent; never two arms on the same token at once.
Primary metric: landed_within_window: Share of all exit intents with a landed_ok copy at slot - anchor.slot <= landing_window_slots; anchor = trigger_tx slot.
Denominator: Every exit intent, including refused and expired.
Randomization: Seeded CSPRNG per intent; one position per token per wallet set; arms never concurrently on one token (sha256_mod_v1)
Stopping rule: Fixed n; single analysis.

Arms

Arm	What it does
rout	Rout sell rule, bounded slippage start_bps..max_bps, own tip.
control	Control path agreed in the protocol (reference RPC, or the partner's path when the partner runs it).

Secondary metrics

Metric	Definition
`implementation_shortfall_bps`	(curve price at anchor slot - realized price) / curve price at anchor slot, from tx meta; computed for landed exits.
`spend_per_landed`	As E0.
`delta_slot_quantiles`	As E0.

Exclusions, decided before the run

None. Every intent stays in the denominator.

Acceptance

Criterion	Agreed value
`min_landing_difference`	Not agreed
`max_cost_ratio`	Not agreed
`max_shortfall_difference_bps`	Not agreed
`max_double_landed_intents`	0

What each tier hides

Signatures, slots and ids are never redacted: they are how anyone re-checks a result. A run can't be published wider than the tier it was registered at.

Partner-confidential removes: host_id, nonce_account, stratum
Public also removes: mint, size_raw, quote, endpoint_region, refusal_code, send_error

Waits on

E1E2E4MAINNETFILL-QUALITY-DECODERBUDGET-B

The file

protocols/B-autosell-exit.json

{
  "protocol_id": "B-autosell-exit",
  "version": 1,
  "experiment": "B",
  "status": "draft",
  "title": "B · AutoSell: trigger-to-inclusion, exit probability, cost and execution quality",
  "hypothesis": "For exit intents triggered by a pre-defined condition, the Rout arm lands within the window more often, at no higher cost per landed exit and no worse implementation shortfall, than the control path.",
  "customer_value": "Auto-sell and counter-trade exits that still land, at a bounded price, when the market moves fast.",
  "unit": "One exit intent on a position held by the experiment's wallets. One arm per intent; never two arms on the same token at once.",
  "arms": [
    {
      "arm_id": "rout",
      "description": "Rout sell rule, bounded slippage start_bps..max_bps, own tip."
    },
    {
      "arm_id": "control",
      "description": "Control path agreed in the protocol (reference RPC, or the partner's path when the partner runs it)."
    }
  ],
  "primary_metric": {
    "id": "landed_within_window",
    "definition": "Share of all exit intents with a landed_ok copy at slot - anchor.slot <= landing_window_slots; anchor = trigger_tx slot."
  },
  "secondary_metrics": [
    {
      "id": "implementation_shortfall_bps",
      "definition": "(curve price at anchor slot - realized price) / curve price at anchor slot, from tx meta; computed for landed exits."
    },
    {
      "id": "spend_per_landed",
      "definition": "As E0."
    },
    {
      "id": "delta_slot_quantiles",
      "definition": "As E0."
    }
  ],
  "parameters": {
    "confidence": 0.95,
    "quantiles": [
      0.5,
      0.9,
      0.95,
      0.99
    ],
    "bootstrap_resamples": null,
    "landing_window_slots": null,
    "commitment": "finalized",
    "max_abs_clock_offset_ns": null,
    "target_n_per_arm": null,
    "budget_lamports_max": null,
    "tip_lamports_by_arm": null,
    "comparisons": [
      [
        "rout",
        "control"
      ]
    ],
    "strata": [
      "launchpad",
      "congestion_tercile",
      "leader_window_position",
      "region"
    ],
    "cluster_slots": null
  },
  "randomization": {
    "unit": "intent",
    "method_id": "sha256_mod_v1",
    "method": "Seeded CSPRNG per intent; one position per token per wallet set; arms never concurrently on one token",
    "strata": [
      "launchpad",
      "congestion_tercile"
    ]
  },
  "denominator": "Every exit intent, including refused and expired.",
  "exclusions": [],
  "stopping_rule": "Fixed n; single analysis.",
  "acceptance": {
    "min_landing_difference": null,
    "max_cost_ratio": null,
    "max_shortfall_difference_bps": null,
    "max_double_landed_intents": 0
  },
  "redact_by_tier": {
    "partner_confidential": [
      "host_id",
      "nonce_account",
      "stratum"
    ],
    "public": [
      "mint",
      "size_raw",
      "quote",
      "endpoint_region",
      "refusal_code",
      "send_error"
    ]
  },
  "requires": [
    "E1",
    "E2",
    "E4",
    "MAINNET",
    "FILL-QUALITY-DECODER",
    "BUDGET-B"
  ]
}

Search