Polymarket Adapter Benchmarks

Numbers measured 2026-05-19 on AMD Ryzen Threadripper 9980X under rustc 1.95.0, bench-lto profile (release opts + lto = "fat" + codegen-units = 1, debug = full). ASLR is disabled via setarch -R for the run. The CPU governor was powersave (not performance) for this capture; absolute numbers will improve under performance, but same-machine deltas remain meaningful.

Refresh on substantive perf change or before release; bump the date. Absolute numbers vary by machine; only same-machine deltas are meaningful.

How to reproduce

sudo cpupower frequency-set -g performance
setarch -R cargo bench -p nautilus-polymarket --profile bench-lto \
    --bench data --bench exec --bench micros --bench signing
sudo cpupower frequency-set -g powersave  # restore default

For policy and the general noise-reduction recipe see BENCHMARKING.md at the repo root.

Inbound pipeline (data.rs)

Raw WS frame bytes (market channel) or REST row (user channel) -> Nautilus domain type. Covers decode + parse + cache lookup + Nautilus type construction. No I/O, no async runtime, no channel.

Rows ordered from the most fundamental market-data stream (book deltas) down through the snapshot variant, the derived top-of-book quote streams, trades, and finally the user-channel reports. order_event and order_fill use the REST GET /orders and GET /trades parse paths because the WS user-channel -> report conversion is private to the dispatch loop; both paths share the string-decimal + status logic.

Execution pipeline (exec.rs)

Strategy intent -> per-request JSON body + L2 HMAC-SHA256 signature. Covers maker/taker amount math, EIP-712 order signing (submits only), JSON body serialization, and the HMAC body signature auth_headers attaches via Credential::sign. The fixed-cost work auth_headers does around the signature (timestamp string format + the five POLY_* header entries) is omitted; it's constant overhead unrelated to the regressions these benches are meant to catch. Polymarket has no in-place modify on the CLOB (cancel-replace is two independent ops), so there is no modify row.

Crypto path (signing.rs)

Decomposes the exec-pipeline signature cost into its components and covers the L2 HMAC path used by every authenticated REST call.

Component breakdown (micros.rs)

Diagnostic benches that decompose the pipeline numbers above. Use these to localise where time goes when a pipeline bench regresses.

Notes

• Inbound is JSON-decode dominated. decode_only/book (1.56 µs) accounts for ~82% of the book_snapshot pipeline (1.90 µs); decode_only/trade (384 ns) is ~71% of the trades pipeline (537 ns). Decimal / Price / Quantity / UUID / TradeId construction are all sub- µs and not meaningful in the absolute pipeline number.
• String -> Price / Quantity is Decimal-direct. parse_price and parse_quantity in websocket::parse route through Decimal::from_str then Price::from_decimal_dp (matches hyperliquid). All Decimal-typed REST fields (PolymarketOpenOrder, PolymarketTradeReport, PolymarketMakerOrder) and the WS user-channel string fields skip the intermediate f64 parse entirely. This removed the Decimal -> String -> f64 -> Price round-trips that dominated the earlier order_event row, dropping it from ~742 ns to ~603 ns (~19%), and dropped parse_only/book_snapshot from ~422 ns to ~350 ns (~17%). The change also eliminates a class of float-rounding risk on the string-to-fixed-point boundary.
• Exec submits are EIP-712-bound. sign_order is 44 µs and dominates every exec_pipeline/submit_* row; LTO collapses the per-shape differences so limit, market, and neg-risk converge near 47-49 µs. The remaining ~3-5 µs is maker/taker amount math, builder state, JSON body serialise, and the L2 HMAC step (~332 ns, lost in the noise next to ECDSA). Optimisations that don't change the EIP-712 + keccak + secp256k1 path won't move these numbers.
• cancel is HMAC-bound (~399 ns). REST cancels don't need an EIP-712 signature, so the whole client-side cost is the JSON body serialise plus the L2 HMAC-SHA256 signature auth_headers attaches via Credential::sign. The HMAC alone (hmac_l2_sign, 332 ns) is ~80% of this row; the network round-trip dominates wall time in production.
• sign_clob_auth carries hidden signer construction. The function builds a fresh PrivateKeySigner from the hex key on every call (~31 µs of overhead, exactly the signer_construction cost) before signing. This path is cold (only used by the CLOB /auth/api-key and /auth/derive-api-key flows at credential bootstrap), so the overhead is not a production hotspot. If sign_clob_auth ever ends up on a hot path, accept a pre-constructed signer instead.
• trade_id_determine (102 ns) is the FNV-1a hash over (asset_id, side, price, size, timestamp) used to make trade IDs deterministic across reconnects. Lower numbers are not desirable here; the cost reflects the security property.
• Maker fills don't show up in order_fill. The REST trade-report fixture exercises the taker fast path; the maker fan-out (build_maker_fill_report per maker order in maker_orders) is one small allocation per leg and is not separately benched. Add a maker_fan_out row if you need to track it.