A cross-cutting view of the unified roster: the 73 engines written by people who publicly
claim a professional trading-industry role — engineers and quants at market-making, prop-trading,
and hedge-fund firms, exchange and vendor staff. It also includes the official vendor/org engines
and the production DEX matchers. Every engine here also appears in
CONSENSUS_CONFORMING_ENGINES.md or
NON_CONFORMING_ENGINES.md (and in the
CORRECTNESS_FINDINGS.md roster); this file only gathers them in one
place.
Of the 73: 20 conform as shipped (two of them measured accessor-only), 25 conform only after
a documented fix, and 28 diverge, cannot finish their worst scenario, or crash. The fastest
worst-case among the 73 is 8.19 M/s; the FlashOne reference measures 33.20 M/s on the same
workload. The per-row findings below carry the specifics; the filed issue for each is linked in
CORRECTNESS_FINDINGS.md. Each row is a snapshot at that engine's
pinned commit — several have since been fixed upstream — and most are personal side projects that
reflect their authors' own goals.
No number or finding here ranks or judges an author's engineering quality — the harness merely reports what each pinned commit does. Affiliations are as the authors state them publicly.
‡ = authored by a professional trading-industry engineer — a personal side project with no commercial intent, not their employer's work, except where the Author column explicitly labels an official vendor/org repo. Affiliations are as the authors state them publicly (GitHub profile / bio). · ★ = repository has 50+ GitHub stars. Rows are sorted by worst-case throughput — the lowest of an engine's five scenario results, the benchmark's definitional result.
On authorship and affiliations. Each ‡ engine is an individual developer's personal side project with no commercial intent — not their employer's product or work. No engine here is a company's official work unless its author has explicitly published it as such; the few official vendor / protocol / org repositories are the ones labeled as such in the Author column. The professional affiliations shown are the authors' own public self-descriptions (GitHub profile, bio, or a linked résumé), reproduced as stated and not independently verified by us as facts of employment — offered as context on the community that builds matching engines, not as a statement about, or on behalf of, any named company.
Unlike the personal side projects that make up most of this file, these 12 are official repositories — a vendor's shipped product, an exchange's own engine, or a firm's org repo. Of the 12: 4 conform as shipped, 3 conform after a documented fix (among them the 10,236★ StockSharp, whose same-price orders could match out of arrival order at the pinned commit), and 5 diverge or cannot finish (among them the Lykke exchange's engine, whose static deep-book run exceeded the 2 M-message budget at the pinned commit).
| Engine | Lang | Author (as publicly claimed) | Conformance | Worst-case M/s | Finding (one line) |
|---|---|---|---|---|---|
| parity (502★) ‡ | Java | paritytrading — open-source org repo (self-described 'open source trading technologies') | as shipped | 2.21 | RB-tree: TreeSet + fastutil id-map |
| coralme (56★) ‡ | Java | CoralBlocks — official vendor repo | as shipped | 1.97 (flash-crash) | clean |
| stocksharp (10236★) ‡ | C# | StockSharp — official vendor repo | with fix | 1.64 (swing-25) | same-price orders could match out of arrival order (Dictionary enumeration replaces FIFO); conforms with the documented one-line fix; #681; fixed upstream — RESOLVED_FINDINGS.md (pin stays pre-fix) |
| vdt ‡ | JavaScript | BitFloor's co-founders — the exchange's own engine, mirrored (BitFloor closed 2013; repo owner is an uninvolved re-hoster) | with fix | 0.71 | assert never imported → ReferenceError on side guard |
| viabtc (2784★) ‡ | C | the founder/CEO of ViaBTC, founder of CoinEx (official org repo) | as shipped | 0.29 | skiplist + dict (clean) |
| trademacher ‡ | Java/JNI | TradeMatcher — the org's own published matching-engine project | as shipped | 0.15 (swing-25) | clean |
| zorrofix ‡ | C++ | DSEC Capital's own repo (org bio: 'Algo Trading Tech Shop') | with fix | 0.01 (static) | sweep loop never checks whether the aggressor has closed → execute(0)→NAN→abort #12 |
| abides ‡ | Python | JPMorgan research org repo | non-conforming | infeasible (static) | byte-identical on small cells, but static times out at 2 M |
| lykke ‡ | Kotlin | the Lykke exchange — official repo | non-conforming | infeasible (static) | fast on normal; its static deep-book run exceeded the 2 M-message budget at the pinned commit (conforms on the scenarios it completes) |
| opexdev ‡ | Kotlin | OPEX — official org repo (open-source crypto-exchange platform) | non-conforming | infeasible (2 M) | modify after a partial fill rests less than requested (stale filledQuantity) + an unconditional O(book-depth) tax per message → infeasible at 2 M; [Bug Report] #688 |
| peatio ‡ | Ruby | principal author an ex-core developer at Yunbi (closed crypto exchange); the open-source Peatio exchange itself | non-conforming | infeasible (2 M) | byte-identical on completed cells; the Ruby rbtree engine exceeds the watchdog |
| buttercoin ‡ | CoffeeScript | Buttercoin — a Bitcoin exchange that closed in 2015; open-sourced engine | non-conforming | crash | partial residual re-inserted under inverted price key → book locks; buttercoin/buttercoin-engine, MIT; #9 |
The order books that run (or ran) real decentralized exchanges — a different deployment target for the same matching problem. Measured de-chained as native code, 5 conform (Serum, Manifest, Phoenix cleanly; dYdX and Vega with a read-only accessor patch) at 0.08–2.6 M/s worst-case.
What
accessor-onlymeans. It denotes the only modification made to those two engines, not a limit on what was tested. dYdX and Vega keep their book private to their own package, so the harness's book-state audit has no way to readbest_bid/best_ask/depth_at. The patch is a minimal, additive, side-effect-free read-only accessor that forwards to the engine's own already-correct state. No matching, ordering, or crossing line is touched, and the full matching path is driven through each engine's own public API.
The 4 measured through their chain's VM cannot complete the 2 M-message tape under VM interpretation — an execution-environment cost plus documented per-call design caps (gas-safety parameters, not correctness defects): under revm interpretation Clober runs at ~0.02 M/s and its 2^15 orders-per-tick cap (a representational limit) diverges the deep book; DeepBook's own MAX_FILLS=100 per-call cap (a documented gas-safety cap, not a defect) truncates a deep sweep; Econia and Osmosis exceed the 600 s watchdog under the Move VM and wasmer (~444 ms/message).
| Engine | Lang | Author (as publicly claimed) | Conformance | Worst-case M/s | Finding (one line) |
|---|---|---|---|---|---|
| serum ‡ | Rust | Project Serum — the original Solana on-chain order book | as shipped | 2.625 (static) | de-chained Solana CLOB — clean |
| manifest ‡ | Rust | Manifest (production Solana CLOB; pinned via the Bonasa-Tech repo) | as shipped | 2.145 (static) | de-chained — clean |
| phoenix ‡ | Rust | Ellipsis Labs (founders ex-Jane Street/Citadel) — official repo | as shipped | 1.5 (static) | de-chained production Solana CLOB — clean |
| dydx ‡ | Go | dYdX — official protocol org repo (v4-chain memclob); a dYdX founder | as shipped | 0.11 (swing-25) | de-chained v4 memclob (accessor-only) |
| vega ‡ | Go | Vega Protocol — official org repo; a Vega Protocol founder (designed a London Stock Exchange matcher) | as shipped | 0.08 (static) | accessor-only |
| clober ‡ | Solidity/EVM | Clober — official org repo (v2-core, production EVM CLOB) | non-conforming | 0.02 (static diverges) | de-chained via revm; the static deep-book scenario diverges at Clober v2's own 32,768-order-per-tick (2^15) OrderId cap — conforms 400/400 on the four feasible scenarios; a representational limit (QuantCup-class), not a correctness defect; ~0.02 M/s under revm interpretation |
| econia ‡ | Aptos Move | Econia Labs — official org repo (production Aptos CLOB) | non-conforming | infeasible (2 M) | de-chained via the Aptos Move VM (FakeExecutor); a deep same-price sweep exceeds the 600 s watchdog (~444 ms/msg, gate 32/33) → infeasible |
| osmosis ‡ | CosmWasm/Rust | Osmosis — official org repo (sumtree-orderbook, production Cosmos CLOB) | non-conforming | infeasible (2 M) | infeasible at 2 M (the wasmer-driven audit exceeds the 600 s watchdog); separately, a latent stale-best-tick pointer (next_bid_tick/next_ask_tick never retracted on a cancel or exact-drain sweep) fails the gate's state dimension without the fix; stale-best-tick bug filed #211 |
| deepbook ‡ | Sui Move | Mysten Labs — official org repo (DeepBook v3, Sui's CLOB) | non-conforming | infeasible (2 M) | de-chained via the Sui Move VM (simulacrum); a deep recursive sweep is truncated by DeepBook's own MAX_FILLS=100 per-call cap — a documented gas-safety design cap, not a defect (gate 32/33) + infeasible at 2 M |
FlashOne — the harness publisher's production engine — is shown at the top as the reference point; it is not counted in the 73.
| Engine | Lang | Author (as publicly claimed) | Conformance | Worst-case M/s | Finding (one line) |
|---|---|---|---|---|---|
| FlashOne | C++ | Flash One Technologies LLC | as shipped | 33.20 (normal) | reference target |
| e820 / weekend-orderbook ‡ | C | an IMC Trading engineer | with fix | 8.19 | singly-linked orphan + aggressor-price fix #1 |
| CppTrader (1041★) ‡ | C++ | Head of C++ Development at Finstek (FX/CFD trading-platform vendor) | as shipped | 7.26 (normal) | a ModifyOrder defect off the canonical path is fixed upstream — RESOLVED_FINDINGS.md #42 |
| ndfex ‡ | C++ | an ex-Citadel Securities engineer (17y in HFT) | as shipped | 6.825 (swing-25) | std::map RB-tree book (clean) |
| yashkukrecha ‡ | C++ | incoming at Jump Trading | as shipped | 6.26 (normal) | two priority_queues + timestamp FIFO tiebreak (clean) |
| daniele ‡ | C++ | an Optiver engineer | with fix | 5.36 (normal) | fill-report reads a maker freed in the same fill (matching is correct); #1 |
| ghosh (677★) ‡ | C++ | a low-latency trading-systems developer and Packt author | with fix | 5.05 (swing-25) | MIT; flat 256-slot price index shared by both sides had no collision handling → cross-side bucket merge crash; + a 1,048,576 order-id cap; both fixed #9 |
| hroptatyr/clob ‡ | C | a Data Manager at GA Financial Solutions GmbH (GA Group / GA Asset Management, systematic funds) | as shipped | 4.73 (normal) | b+tree CLOB, _Decimal64 (no patch) |
| ranjan2829 (131★) ‡ | C++ | an ex-Baven (GigaBrain) trading-agents engineer; ex-MaticAlgos quant intern | with fix | 4.07 | 4 memory-safety defects fix #3 |
| faulaire ‡ | C++ | a software engineer at One Eleven Capital (Paris quant hedge fund) | as shipped | 3.62 | Boost.MultiIndex: hashed id + ordered price (clean) |
| Tzadiko (307★) ‡ | C++ | YouTube's 'Coding Jesus' (getcracked.io) — an ex-quant developer at an undisclosed proprietary trading firm | with fix | 3.39 (flash-crash) | IOC self-deadlock; two-site lock-wrapper fix #11 + #12 |
| serum ‡ | Rust | Project Serum — the original Solana on-chain order book | as shipped | 2.625 (static) | de-chained Solana CLOB — clean |
| parity (502★) ‡ | Java | paritytrading — open-source org repo (self-described 'open source trading technologies') | as shipped | 2.21 | RB-tree: TreeSet + fastutil id-map |
| shivaganapathy ‡ | C++ | an IMC engineer | as shipped | 2.15 (normal) | two priority_queues + timestamp FIFO tiebreak (clean) |
| manifest ‡ | Rust | Manifest (production Solana CLOB; pinned via the Bonasa-Tech repo) | as shipped | 2.145 (static) | de-chained — clean |
| coralme (56★) ‡ | Java | CoralBlocks — official vendor repo | as shipped | 1.97 (flash-crash) | clean |
| robdev ‡ | Rust | a CME Group engineer | with fix | 1.76 (static) | clear the emptied price level on cancel, return the real cancel result, and kill the IOC residual — all latent as-shipped (match path immune; the stale best_price is caught by the gate's state audit) #1 |
| stocksharp (10236★) ‡ | C# | StockSharp — official vendor repo | with fix | 1.64 (swing-25) | same-price orders could match out of arrival order (Dictionary enumeration replaces FIFO); conforms with the documented one-line fix; #681; fixed upstream — RESOLVED_FINDINGS.md (pin stays pre-fix) |
| matchina ‡ | Rust | at GSR (crypto market maker) | with fix | 1.60 (static) | taker-exhaustion guard — no phantom zero-quantity trades; fixed upstream — RESOLVED_FINDINGS.md #3 |
| phoenix ‡ | Rust | Ellipsis Labs (founders ex-Jane Street/Citadel) — official repo | as shipped | 1.5 (static) | de-chained production Solana CLOB — clean |
| tembolo ‡ | C | a quantitative developer at Tradeweb | with fix | 1.475 (swing-25) | two capacity ceilings (8192-order pool silent-drop + 512 price-level abort) #1 |
| piquette ‡ | Go | a financial services engineer at Watershed Technologies (equity-options ATS) | non-conforming | 1.28 (flash-crash) | bid/ask maps never populated → every cancel rejected; best-ask uses > not < #2 |
| koral ‡ | C++ | a Coinbase software-engineering intern | as shipped | 1.255 (normal) | FIX exchange (clean; thread-affinity plumbing only) |
| ironcrypto ‡ | Rust | self-described 'TradFi/DeFi Quant' | with fix | 1.04 (6.2 normal) | no-license; adapter restores engine's removed cancel impl (faithfulness caveat) |
| javalob ‡ | Java | a JPMorgan engineer | as shipped | 1.03 (swing-40) | teaching LOB (clean); worst case with bidirectional engine_on_batch (per-message 0.86) |
| trusted ‡ | Rust | a KRX market-maker at IBK Securities | with fix | 0.925 (static) | latent bid-side market-order double-subtract underflow #9 |
| kennethzhang ‡ | C++ | a Squarepoint quant researcher | with fix | 0.86 (static) | price the limit-vs-limit cross at the resting maker, not the taker (the adapter normalizes it today) #1 |
| swirly ‡ | Java | a trading-systems developer; co-founder of Reactive Markets | as shipped | 0.79 (swing-40) | clean — native revise changes only lots, so modify = cancel+reinsert per contract |
| vdt ‡ | JavaScript | BitFloor's co-founders — the exchange's own engine, mirrored (BitFloor closed 2013; repo owner is an uninvolved re-hoster) | with fix | 0.71 | assert never imported → ReferenceError on side guard |
| damian ‡ | Kotlin | a developer with 20y at a bank | as shipped | 0.38 (static) | clean |
| pyob ‡ | Python | an FX e-trading quant at mBank | with fix | 0.30 (swing-25) | deque IndexError on a full fill + stale best_price after cancel #1 |
| joaquinbejar ‡ | Rust | a quant developer at Capital Delta | with fix | 0.29 (static) | Book::replace crossing modify never re-matches → rests a crossed book #59; fixed upstream — RESOLVED_FINDINGS.md (pin stays pre-fix) |
| viabtc (2784★) ‡ | C | the founder/CEO of ViaBTC, founder of CoinEx (official org repo) | as shipped | 0.29 | skiplist + dict (clean) |
| oceanbook ‡ | Go | self-described HFT developer (bio 'HFT / C++ / Go'; @spectra-fund) | with fix | 0.26 (flash-crash) | Depth writes quantity into the qty field, not the price field #44 |
| silue ‡ | Python | an ex-Kairon Labs quantitative-research intern (crypto market maker) | with fix | 0.26 | get_pnl computes Decimal × None → crash on the first cross #1; fixed upstream — RESOLVED_FINDINGS.md (pin stays pre-fix) |
| rakuzen25 ‡ | C++ | an Optiver intern | with fix | 0.18 (flash-crash) | within-level FIFO fix (swap-with-last broke arrival order); uint16 ceiling residual; issues disabled upstream |
| sculd ‡ | Python | a quant/developer who has worked at Two Sigma | with fix | 0.18 (swing-25) | guard unknown-id cancel/status against KeyError and skip cancelled heads in _get_best_price — latent as-shipped (matching path unaffected) #1 |
| mkhoshkam ‡ | Go | a software engineer at Technance (trading-infrastructure vendor) | with fix | 0.18 | heap Less ignores seq → FIFO fix (adapter) #10 |
| trademacher ‡ | Java/JNI | TradeMatcher — the org's own published matching-engine project | as shipped | 0.15 (swing-25) | clean |
| OrderBook-rs (477★) ‡ | Rust | a quant developer at Capital Delta | with fix | 0.13 (static) | partial-fill maker keeps FIFO priority (push_front, not re-queue to tail); fixed upstream — RESOLVED_FINDINGS.md #88 |
| dydx ‡ | Go | dYdX — official protocol org repo (v4-chain memclob); a dYdX founder | as shipped | 0.11 (swing-25) | de-chained v4 memclob (accessor-only) |
| qa-rs ‡ | Rust | a private-fund manager (Shanghai Binghao) | with fix | 0.09 (static) | OrderQueue lazy-deletion — same-id reinsert leaves a stale heap entry #1, get_depth over-counts a plain cancel until swept #2; + 5 latent Orderbook match-loop bugs off the limit-only workload: 1000-id recycle drops orders #3, market remainder rested not killed #4, amend skips the crossing check #5, NaN price passes validation #6, per-order sweep recursion overflows the stack #7 |
| vega ‡ | Go | Vega Protocol — official org repo; a Vega Protocol founder (designed a London Stock Exchange matcher) | as shipped | 0.08 (static) | accessor-only |
| rhodey ‡ | JavaScript | an engineer at Velo Data (crypto-derivatives data & trading platform) | as shipped | ~0.02–0.12 (swings) | clean |
| amer ‡ | Java | a former Nasdaq engineer | non-conforming | 0.02 (static) | same-price orders execute LIFO not FIFO — >=/<= insertion splices ahead of equal-priced peers (33/40 sweep cells diverge); unconditional O(depth) contra-scan → deep 2 M runs can exceed the watchdog; #1 |
| clober ‡ | Solidity/EVM | Clober — official org repo (v2-core, production EVM CLOB) | non-conforming | 0.02 (static diverges) | de-chained via revm; the static deep-book scenario diverges at Clober v2's own 32,768-order-per-tick (2^15) OrderId cap — conforms 400/400 on the four feasible scenarios; a representational limit (QuantCup-class), not a correctness defect; ~0.02 M/s under revm interpretation |
| zorrofix ‡ | C++ | DSEC Capital's own repo (org bio: 'Algo Trading Tech Shop') | with fix | 0.01 (static) | sweep loop never checks whether the aggressor has closed → execute(0)→NAN→abort #12 |
| khrapovs ‡ | Python | a senior ML engineer at ING (bank) | with fix | very slow (pure Python) | MIT; orders_by_expiration not pruned on fill #25; fixed upstream — RESOLVED_FINDINGS.md (pin stays pre-fix) |
| darkpool ‡ | C++ | founder of Bitwyre (crypto exchange) and Systematic Trading Group (HFT market maker) | with fix | very slow (static) | aggressor pricing in the execution-price report (fixed); matcher core vendored from QuickFIX examples/ordermatch #1 |
| abides ‡ | Python | JPMorgan research org repo | non-conforming | infeasible (static) | byte-identical on small cells, but static times out at 2 M |
| bitex ‡ | Python | a founder of BlinkTrade (the open-source platform behind the Foxbit exchange) | non-conforming | infeasible (static) | correct on normal/swings, but the static deep-book scenario times out at 2 M |
| figgie ‡ | OCaml | an ex-Jane Street engineer — figgie is Jane Street's Figgie card game (a FIFO matcher), not a commercial engine | non-conforming | infeasible (static) | byte-identical where it completes, but static times out at 2 M |
| isaaccheng ‡ | Python | ex-T. Rowe Price fixed-income quant developer | non-conforming | infeasible (static) | byte-identical where it completes, but static times out at 2 M |
| ismailfer ‡ | Java | self-described systematic trading developer / quant trader | non-conforming | infeasible (2 M) | processTrade() clears the wrong side's active flag on a full fill — truncates multi-order sweeps and orphans resting quantity (a filled id can also never be reused); diverges + infeasible at 2 M; [Bug Report] #1 |
| liqian ‡ | C++ | ex-Virtu Financial quant trader | non-conforming | infeasible (2 M) | O(20 M-tick) domain scan per order → infeasible at 2 M; occupancy bitsets never skip empty levels; #1; fixed upstream — RESOLVED_FINDINGS.md (the infeasible verdict describes the pinned commit) |
| lykke ‡ | Kotlin | the Lykke exchange — official repo | non-conforming | infeasible (static) | fast on normal; its static deep-book run exceeded the 2 M-message budget at the pinned commit (conforms on the scenarios it completes) |
| nilesh05apr ‡ | C++ | a Tower Research Capital SWE intern | non-conforming | infeasible (2 M) | matches only one counterparty per order; O(n²) re-sort → infeasible at 2 M; #1 |
| opexdev ‡ | Kotlin | OPEX — official org repo (open-source crypto-exchange platform) | non-conforming | infeasible (2 M) | modify after a partial fill rests less than requested (stale filledQuantity) + an unconditional O(book-depth) tax per message → infeasible at 2 M; [Bug Report] #688 |
| pylob ‡ | Python | a JPMorgan engineer | non-conforming | infeasible (static) | SQLite-backed; static times out at 2 M; 2 bugs #8; the two bugs are fixed upstream — RESOLVED_FINDINGS.md (the infeasible verdict is unchanged) |
| econia ‡ | Aptos Move | Econia Labs — official org repo (production Aptos CLOB) | non-conforming | infeasible (2 M) | de-chained via the Aptos Move VM (FakeExecutor); a deep same-price sweep exceeds the 600 s watchdog (~444 ms/msg, gate 32/33) → infeasible |
| osmosis ‡ | CosmWasm/Rust | Osmosis — official org repo (sumtree-orderbook, production Cosmos CLOB) | non-conforming | infeasible (2 M) | infeasible at 2 M (the wasmer-driven audit exceeds the 600 s watchdog); separately, a latent stale-best-tick pointer (next_bid_tick/next_ask_tick never retracted on a cancel or exact-drain sweep) fails the gate's state dimension without the fix; stale-best-tick bug filed #211 |
| lll ‡ | C++ | a Citadel Securities engineer (HFT systems; engine predates employment) | non-conforming | infeasible (static) | infeasible at 2 M — stalls at the 600 s watchdog on deep static |
| mattdavey ‡ | Java | two then-Lab49 engineers (trading-technology consultancy) | non-conforming | infeasible (2 M) | O(n²) insert rescan — infeasible at 2 M #26 |
| peatio ‡ | Ruby | principal author an ex-core developer at Yunbi (closed crypto exchange); the open-source Peatio exchange itself | non-conforming | infeasible (2 M) | byte-identical on completed cells; the Ruby rbtree engine exceeds the watchdog |
| prystupa (58★) ‡ | Scala | an ex-Lab49 engineer (15+ years building trading systems) | non-conforming | infeasible (static) | infeasible at 2 M — a deep stressful-static book stalls past the watchdog; the latent finding is fixed upstream — RESOLVED_FINDINGS.md (the infeasible verdict is unchanged) |
| thelilypad ‡ | Python | a quantitative options trader/researcher (NOPE indicator; Salience Capital) | non-conforming | infeasible (2 M) | same-price orders can fill out of arrival order (no heapq tiebreak); diverges + infeasible at 2 M |
| zzsun777 ‡ | C++ | the founder of trading-software vendor Coreware Ltd (llfix); at Thomson Reuters at time of authorship (2016); the repo is an unmarked mirror by an uninvolved re-hoster | non-conforming | infeasible (2 M) | byte-identical on every completed cell; O(n) by-owner find exceeds the watchdog |
| raunakchopra ‡ | C++ | a Flow Traders engineer | non-conforming | crash (sweep) | recursive re-submit overflows the stack; not price-time; O(trades²) I/O; #1 |
| buttercoin ‡ | CoffeeScript | Buttercoin — a Bitcoin exchange that closed in 2015; open-sourced engine | non-conforming | crash | partial residual re-inserted under inverted price key → book locks; buttercoin/buttercoin-engine, MIT; #9 |
| soham ‡ | C++ | an iRage quant-analyst intern | non-conforming | — | binary Yes/No prediction-market matcher — prices hard-validated to [1, 99]; rejects the benchmark's full-range tapes in full by design (not a FIFO defect); internally sound (author-run differential fuzzing; in-domain checks match exactly), no bug report was filed |
| lethalazo ‡ | C++ | at Marshall Wace (hedge fund) | non-conforming | — | add-only book, missing the cancel/modify operations the benchmark drives; 2 bugs #1 |
| pyobsim ‡ | Python | a trading-systems engineer at tplus (decentralized prime exchange) | non-conforming | — | Side.remove deletes only the head; __match mutates the level while iterating #2; fixed upstream — RESOLVED_FINDINGS.md (the verdict describes the pinned commit) |
| konqr ‡ | Python | a Tower Research Capital quant (ex-JPMorgan Equities Electronic Trading) | non-conforming | — | Hawkes-process simulator — not a price-time CLOB in the benchmark's form #19 |
| deepbook ‡ | Sui Move | Mysten Labs — official org repo (DeepBook v3, Sui's CLOB) | non-conforming | infeasible (2 M) | de-chained via the Sui Move VM (simulacrum); a deep recursive sweep is truncated by DeepBook's own MAX_FILLS=100 per-call cap — a documented gas-safety design cap, not a defect (gate 32/33) + infeasible at 2 M |
As everywhere in this repository: these observations are a reproducible, time-stamped snapshot of
a specific pinned commit, offered back to the authors — several are already fixed upstream — and
not a judgment of anyone's engineering quality; personal side projects reflect their authors' goals,
which may differ from a venue's. The one-line finding and the filed-issue link per engine are in
CORRECTNESS_FINDINGS.md.