Case Study #001 is live — paper +64%, live 0-of-5, fully audited.

Read Case Study #001
Strategy OSApply for an audit pilot

Strategy audit platform for active traders

Your backtest said +64%. Your live account said zero. We built the platform that tells you why.

Strategy OS reconciles paper vs live for systematic strategies. Upload the backtest, the paper run, and the broker fills. Get back a structured report: where the edge actually lives, where it leaked, and what claim is honest to make. We audited our own strategy first — that report is on this site.

  • Paper-vs-live reconciliation with trade-level diff and cost attribution
  • Alpha decay scoring across windows — catch silent regime change early
  • Explicit claim ladder: which numbers you can publish, which you can't

Audit engagement. No signal delivery. No managed money.

Case Study #001

Insider cluster buying strategy — paper +64%, live 0-of-5

Context

Strategy showed t-stat 2.24 across 169 out-of-sample trades 2023-2025 with +2.73% avg per trade. Paper environment reported +64% over a synthetic window built from cached events.

Risk note

Live deployment closed five trades, lost on every one. The audit shows the +64% was structurally non-reproducible (backfill of past events) — not the strategy that ran live.

AI review

Alpha was real 2018-2024, decayed through 2025, flat-negative in 2026. We publish the half-year breakdown, the ablation matrix, and the claim ladder so you can verify the audit yourself.

Paper-vs-live reconciliationAlpha decay scoringExplicit claim ladderCost-realistic attributionReproducible methodology

The problem

A clean backtest doesn't mean a working strategy. Most traders find out the expensive way.

Strategy OS forces the comparison nobody enjoys: what the backtest claimed, what paper actually produced, and what your live broker actually filled. The gap is where the lesson lives.

What the backtest says

Curve, drawdown, Sharpe. Almost always good. Almost always overfit, lookahead-leaky, or running on a universe that won't repeat.

What paper says

Usually different. Sometimes wildly. Cached events, replay assumptions, and forward-only stream behavior diverge from the original simulation.

What live says

Different again. Slippage, fills, broker risk caps, missed signals. The honest number lives here — and it deserves its own audit.

How it works

Three runs. One reconciliation. One claim ladder.

An audit pilot is a structured engagement, not a SaaS trial. You hand us the data; we hand you a report.

01

Strategy intake

You provide backtest output, paper-run logs, and broker execution records. We normalize trade-level data into a single reconciliation schema.

02

Window decomposition

We split the historical sample into in-sample, out-of-sample, and live windows. T-stat, expectancy, and decay scored per window.

03

Gap attribution

Every divergence between paper and live is attributed: slippage, missed signal, sizing delta, regime shift, or implementation bug.

04

Claim ladder

Final report lists which claims are publishable (Level 1), defensible internally (Level 2), speculative (Level 3), or unsupported (Level 4).

What the audit includes

Six deliverables. No fluff. No marketing summary.

Reconciliation report

Trade-by-trade diff between paper and live, with cost and timing attribution.

Window ablation

Three-window decomposition with t-stat, expectancy, and significance per window.

Decay scoring

Time-series view of edge magnitude — early signal of regime change or crowding.

Claim ladder

Explicit Level 1-4 verdict on which numbers are honest to publish externally.

Implementation gap log

Every divergence categorized: leak, sizing, slippage, regime, or genuine alpha decay.

Closeout brief

One-page summary: continue, modify, or shelve the strategy. Plus the reasoning behind it.

Worked example

Our own strategy. Audited in public.

EC_V1 is the insider-cluster strategy we built first. It looked good on paper. It failed in live. The full audit is published as Case Study #001 — same format every external engagement receives.

Ticker

EC_V1

Event

Insider cluster buying — Form 4 signals on US equities

Context

Three-window ablation: 2018-2020 in-sample (t=2.78), 2020-2023 in-sample (t=2.44), 2023-2025 out-of-sample (t=2.24, 169 trades, +2.73% avg). All three windows alpha-positive and statistically significant.

Why it matters

The historical signal was real. The naive implementation was not. Alpha decayed through 2025 (biotech blowups dominated losses) and went flat-negative in 2026 Q1-Q2 (23 trades, -1.10% avg, t=-0.13).

Risk notes

Live ran five trades over five weeks and closed all five at a loss. Comparing that to the paper +64% is not a like-for-like comparison: paper backfilled 90 days of prior events at startup; live only saw forward events. The audit makes the gap explicit.

AI review

Claim ladder verdict: Level 1 — the historical edge 2018-2024 is publishable as a finding. Level 2 — the strategy is currently in decay and should not be re-deployed without sector filter, regime gate, and a much larger account. Level 3+ rejected: the +64% paper figure and any forward-return claim built on it.

Full audit is on the case-studies page. Same format applied to every external strategy submitted for review.

Why this is different

Not a signal service. Not a backtester. An accountability layer.

Not a signal product

We do not deliver trade ideas. We audit strategies you already run or plan to deploy.

Not a backtester

You bring the backtest. We compare it to paper and live. The interesting math is in the gaps.

Not anonymous

Case Study #001 is our own strategy. Every external audit follows the same public-grade format.

Not optional honesty

The claim ladder forces every number to declare its evidence level. No room for marketing arithmetic.

Who this is for

Built for

  • Systematic traders with at least one strategy in paper or live
  • Anyone whose backtest and live results disagree by more than they're comfortable with
  • Quant teams who need an external sanity check before publishing or scaling capital
  • Discretionary traders moving toward systematic — and want the diagnostic before deploying

Not built for

  • Anyone looking for trade ideas or signal subscriptions
  • Investors hoping for a managed account or capital allocation
  • Strategies with no live or paper data — there's nothing yet to reconcile
  • Traders who want validation rather than diagnosis

Audit pilot

One strategy. Two weeks. One honest report.

Audit pilot

€900 / audit pilot

1 strategy. Backtest, paper, and live data reconciled into one report. Claim ladder included. No retainer. No upsell to a signal product.

Limited audit slots per month. Case Study #001 is the worked example; your engagement output looks the same.

Included

  • Trade-level paper-vs-live reconciliation
  • Three-window ablation with t-stat and expectancy
  • Alpha decay scoring across the historical sample
  • Implementation gap log with attribution
  • Claim ladder verdict (Level 1-4) per finding
  • One-page closeout brief with continue/modify/shelve recommendation
  • 30-minute readout call with the analyst

Not included

  • Trade signals or live recommendations
  • Strategy development or parameter tuning
  • Managed accounts or capital allocation
  • Ongoing monitoring (separate engagement)

FAQ

Questions that actually matter

Is this a signal service?+

No. We do not deliver trade ideas. The deliverable is an audit report on a strategy you already have.

What do I need to submit?+

Backtest output (trade-level CSV preferred), paper-run logs if available, and broker execution records (IBKR FlexQuery, fills CSV, or equivalent). The more granular, the sharper the reconciliation.

What if my strategy is profitable in backtest but I haven't traded it live?+

That is exactly when an audit is most useful. We can still run a paper-vs-backtest reconciliation and surface the implementation risk before you commit live capital.

Why should I trust an audit firm whose own strategy lost money?+

Because we published the loss. Case Study #001 is our own strategy, audited the same way we'd audit yours. A firm that has never put a real loss on the table has not actually tested its own methodology.

Will you tell me whether to deploy capital?+

No. The audit provides evidence and a claim ladder. The deployment decision is yours, and it should be — we are not your fiduciary.

Is the audit confidential?+

Yes by default. We do not publish client audits without explicit written consent. The methodology and report template are public; the data and findings are yours.

If your backtest and your live account disagree, you have an audit problem before you have a strategy problem.

Submit a strategy for an audit pilot. Two weeks. One honest report. Same format as Case Study #001.

Audit engagement. No signal delivery. No managed money. No return promises.