Execution replay: auditing our own fills
A backtest assumes fills. A live record proves them. We replay every fill against recorded market data because slippage is a dataset, not an excuse.
3 min read →Systematic derivatives research
Alpha, measured against its worst path.
Axiom Point Research builds market-neutral systematic strategies in listed derivatives. Every edge we run is captured tick by tick, tested against full market history, and sized by how it behaves on its worst days — not its best.
25derivative surfaces under continuous tick-level capture
10⁹+market events processed per trading session
nsnative exchange timestamps preserved end-to-end
24/7self-healing pipeline with automated integrity audits
The standard
A track record is a claim. Evidence is what survives cross‑examination.
Most performance stories are built to be admired. Ours is built to be interrogated — drawdown by drawdown, regime by regime, fill by fill. We hold our own work to the diligence standard a skeptical allocator would apply, before any allocator has to.
The evidence standard
Six disciplines. One record.
Before a strategy is considered ready for outside scrutiny, it has to clear the same six-part examination — applied continuously, not once.
01
Forward record
Strategies earn capital in forward test against predefined risk benchmarks — drawdowns, exposure, and execution quality tracked as part of the record, not footnotes to it.
Observed results over narrative02
Downside ledger
Sortino, Calmar, loss-tail behavior, recovery time, and capital efficiency are reported together. A gain that cannot survive its own adverse path does not count.
Gains must survive adverse paths03
Regime attribution
Results are decomposed across volatility, trend, correlation, and liquidity states — so strength and weakness are inspectable rather than averaged away.
Robustness must be inspectable04
Path stress
Risk is treated as a distribution of outcomes: drawdown sequencing, scenario analysis, and stress ranges rather than a single equity curve.
One curve is not evidence05
Capacity controls
Liquidity, spread, turnover, and per-position exposure are constraints applied before scaling — capacity assumptions stay grounded in market mechanics.
Size changes the trade06
Auditable proof
Every performance claim ties back to audit-ready records — the numbers stay checkable while proprietary signal, sizing, and execution logic stay protected.
Evidence without trade replicationThe machine behind the record
Research infrastructure, built like a trading system.
Conclusions are only as good as the data and the machinery underneath them. We build and operate the entire stack in-house — capture, storage, research compute, and execution audit — so that no claim rests on data we cannot replay.
Capture
Tick-native market capture
Full-depth listed options and equities feeds recorded live, session after session, with native exchange timestamps preserved end-to-end. If it printed, we have it.
Data engine
GPU-scale research archive
Multi-terabyte compressed columnar archives engineered for GPU decompression — full-history scans that would take a conventional stack days come back in minutes.
Operations
Self-healing infrastructure
Layered sentinels audit capture, storage, and processing minute by minute and repair failures automatically. Humans get paged only when automation loses.
Integrity
Nightly archive audits
Every data surface is audited nightly against trailing baselines; gaps are detected and repaired before research ever touches them. Bad data is the quietest way to lose money — we treat it as an incident, not an inconvenience.
Execution
Fill-level execution replay
Live fills are audited against replayed market data, tick by tick. Slippage is not an excuse in our shop — it is a dataset, and it feeds directly back into strategy evaluation and sizing.
Who we are
Engineers first. Allocator‑grade discipline.
Axiom Point is a small team that builds everything it depends on — the market data capture, the GPU research stack, the risk engine, and the execution audit. Backgrounds span quant trading infrastructure, market microstructure, machine learning pipelines, and production AI data systems for financial services.
Validate, don’t guess
Every assumption is checked against data before it is trusted. Measured beats plausible, every time.
Honest accounting
Slippage, fees, capacity, and failure modes are part of the result — not an asterisk under it.
Own the full stack
From exchange feed to fill report, we operate what we rely on. No black boxes between us and the market.
Research notes
How we think, in writing.
Qualified allocators
The record is built to be examined. Examine it.
If you allocate to systematic strategies and care about downside-adjusted performance more than headline returns, we should talk.