> ## Documentation Index > Fetch the complete documentation index at: https://docs.usetissue.xyz/llms.txt > Use this file to discover all available pages before exploring further. # FAQ & Limitations > Common questions, paper-trading disclaimers, and what depends on TxLINE data availability. ## Does Tissue execute real trades? Within TxLINE's own on-chain program, no — `txoracle` has no intent-book or order-matching instruction, so live mode publishes risk-approved quotes there and never invents a fill, a counterparty, or PnL against it. On a separate venue, yes: Slip is a real settlement venue with its own order/execution instructions. A decision that clears a second, stricter, off-by-default capital-risk gate (deliberately higher edge threshold and its own exposure caps — publishing a recommendation and signing a transaction that spends funds are not the same risk decision) is turned into a real signed, confirmed transaction on Slip. See [Verifiability](/verifiability) for how to check a real execution's transaction signature yourself. ## Is the "learning" real, or marketing? The Strategy Arena and regime ablation matrix are real, re-runnable, computed comparisons over real captured corpora — that's the actual evidence. Tissue does not autonomously rewrite its own policy overnight; see [Autonomous Operation](/autonomous-operation) for exactly why, and what the honest alternative (human-reviewed tuning suggestions) looks like instead. ## How rigorously is this actually tested? 322 tests across the daemon, dashboard, and analyst, all real — no mocked units under test. That includes suites purpose-built to find real bugs, not just confirm happy paths: adversarial/malformed input against the TxLINE ingest pipeline (corrupted timestamps, tampered proof responses, NaN-producing prices), adversarial prompts and tool-call spoofing against the analyst's LLM+MCP loop, a real end-to-end browser suite against every dashboard page and desk state, and two independent process-level chaos drills (crash/restart, and SSE disconnect/reconnect) run against real TxLINE and Solana infrastructure. This testing effort itself found and fixed two real bugs — see the [Changelog](/feedback-roadmap#changelog) for exactly what and why. See [Deployment → Testing & verification](/deployment#testing-verification) for how to run every suite yourself. ## Why does the calibration/holdout tool say "underpowered"? Because it currently has only two real captured TxLINE fixtures to work with — its own threshold requires at least three fixtures per side before it will trust a comparison. That's the tool working correctly, not failing. More real captures make the flag go away on their own. ## What's StablePrice, and why does it matter for informed-flow? TxLINE's StablePrice is a single de-margined consensus odds stream, not per-bookmaker lines — confirmed against both captured real data and TxLINE's current live docs. This is why the informed-flow signal classifies a move's *velocity* against a market's own trailing distribution, instead of assuming cross-bookmaker propagation data that doesn't exist in this feed. ## Limitations * **No order execution against TxLINE's own on-chain program** — `txoracle` has no intent-book. Real execution exists on Slip instead (off by default, its own stricter risk gate) — see [Verifiability](/verifiability). * **Underpowered calibration** — only two real fixtures captured so far; more real match captures are needed before any tuning conclusion is trustworthy * **One-fixture-focus by default** — portfolio risk caps support concurrent fixtures, but the desk's design center is one fixture at a time * **Devnet/mainnet IDL mismatch** — a real process-level restart drill against mainnet data surfaced that the bundled IDL is devnet-specific; documented honestly rather than silently worked around * **Public devnet RPC rate limiting** — observed during real restart-drill testing; affects proof-fetch reliability under load on the public devnet RPC, not a code bug. Local Solana anchoring tests now run against a local Surfpool validator instead, sidestepping this specific limitation for anchoring-flow testing. * **Historical proof availability during chaos drills** — TxLINE's proof endpoints don't reliably serve proofs for old/replayed messages, which blocks a live SIGKILL-during-anchor-submission drill against a replayed corpus specifically. Worked around with an independent stream-drop/reconnect drill (`drill:streamdrop`) that proves real SSE reconnect resilience without depending on proof-verification success. ## Paper-trading disclaimer All PnL, fills, and settlement figures shown anywhere in Tissue's output are labeled `simulated` and apply only to replay/research mode. Nothing in this project constitutes financial advice, and no real capital is placed or at risk through Tissue's current execution model.