Before you trust your system.
Try to break it.
You have the system. You have the results. But do they survive scrutiny?
Elite Pass provides independent analysis and stress-testing for racing systems, models and methodologies. We examine where performance is coming from, where it may be vulnerable, and what could be tightened before you commit further time or capital.
Confidential analysisNDA availableIndependent assessment
A profitable backtest is the beginning of the investigation. Not the end.
Racing systems can look exceptional historically and still fail on new data. Small samples, market movement, selection bias, overfitting, changing race conditions and a handful of exceptional results can all create performance that looks stronger than it is.
Elite Pass approaches it differently. We do not begin by trying to prove your system works. We begin by trying to find out why it might not.
If it survives that process, the evidence becomes considerably more interesting.What we can examine
Ten questions a system has to answer.
ROI and return structure
Headline ROI alongside strike rate, average return, drawdown, losing sequences, profit distribution and dependence on a handful of exceptional results.
Market benchmarking
Performance compared with Betfair Starting Price, the benchmark every Elite Pass finding is measured against, and with bookmaker starting price and pre-off exchange prices where the records cover the period. The question is not “did it make money?” but “was the system identifying information the market had not already priced?”
Out-of-sample testing
Discovery separated from validation wherever the supplied data permits, so the rule is judged on races it was never built on.
Overfitting and rule fragility
Do small changes to thresholds or filters collapse performance? A system that only holds at one exact setting may be describing its own dataset rather than the sport.
Segment analysis
Course, distance, class, race type, field size, going, surface, price range, market position, trainer and jockey conditions, and time period.
Drawdown and variance
How deep the losing runs go, how long they last, and whether the return survives the variance a real staking history would have met.
Time stability
Whether the result holds across seasons and years, or rests on a period that has already passed.
Sensitivity testing
What happens to the result when one assumption at a time is moved, and which assumptions it cannot do without.
Data quality
Missing information, leakage from after the race, duplicates, inconsistent timestamps and survivorship in the sample.
Implementation reality
Market availability, timing, liquidity and commission: whether the price the backtest assumed was a price anyone could have taken.
Scope depends on the system and the data supplied. Not every audit includes every test.
We have spent years trying to disprove our own ideas.
Elite Pass was built by investigating racing data at scale: 633,474 historical runner-race records across 71,626 races, 531,115 runners benchmarked against the market’s own price, and 1,180 races with our own timestamped pre-off market record. How that came about is told on the founder’s page; how every number is defined and checked is in the Evidence Room.
Bring us a system you believe works. Our job is not to agree with you. Our job is to interrogate the evidence.
The system audit.
Every engagement is scoped individually. Depending on the system and the data you can supply, an audit may include:
- Initial system and data review
- Performance reconstruction
- ROI and strike-rate analysis
- Profit distribution
- Market benchmarking
- Drawdown analysis
- Out-of-sample validation
- Sensitivity testing
- Segment analysis
- Identification of potentially unstable variables
- Data-quality review
- Assessment of potential overfitting
- Implementation considerations
- Written findings
- Priority recommendations
- Suggested areas for further investigation
Where the evidence supports it, we identify where the system could be simplified, tightened or further tested. Where it does not, we say so.
Four kinds of statement, kept apart
Every finding in a written audit is labelled as one of these, so a hunch is never mistaken for a result.
- Observation
- What the data shows, before any interpretation.
- Evidence
- What survives a test that could have gone the other way.
- Hypothesis
- What might explain it, and has not yet been tested.
- Recommendation
- What we would do next, and what we would leave alone.
We will not manufacture an improvement simply to justify the consultancy. Sometimes the most valuable conclusion is: do not add another rule.Ways of working
Four ways to engage.
Each engagement is scoped to the dataset, its complexity and the analysis required. Terms are agreed before any work begins.
- Initial consultationA private conversation about how the system was built, what data exists and what you want independently tested. It decides whether a full audit would be useful and what it should include.Scoped and quoted individually.
- System auditIndependent reconstruction and stress-testing of one system against the supplied data: return structure, market benchmarking, out-of-sample validation, sensitivity, segments and data quality, with written findings and priority recommendations.Scoped and quoted individually.
- Advanced quantitative auditFor larger datasets, multi-rule models or methodologies that need bespoke reconstruction: everything in the system audit, extended to the depth the data and the question require.Scoped and quoted individually.
- Ongoing research consultancyContinuing independent review as a system is developed, re-tested and run forward, so each change is judged against evidence rather than against the last good month.Scoped and quoted individually.
Not ready to submit the system? Start with a private conversation.
Book an initial consultation to talk through:
- How the system was developed
- What data exists
- Historical performance
- Current concerns
- Where you believe the edge originates
- What you want independently tested
From there we determine together whether a full audit would be useful, and what it should include.
No obligation to proceed to a full audit.
ConfidentialityYour system remains your system.
A racing methodology may represent years of work. Information supplied for an audit is treated as confidential and used only for the agreed analysis.
- NDA
- Available on request
- Use of your material
- The agreed analysis only
- Publication
- Never without your explicit permission
- Data transfer
- Arranged securely, not by contact form
We are happy to sign an appropriate NDA before any proprietary rules, algorithms or datasets are supplied. We will not publish a client’s methodology, selections, datasets or findings without explicit permission. A secure transfer route is agreed with you before anything proprietary is sent; the enquiry form is not the place for it.
People with a system, and a question about it.
- Independent racing-system developers
- Quantitative racing researchers
- Professional and semi-professional bettors
- People who publish selections and want them independently validated
- Racing-data businesses
- Developers building racing models
- Syndicates
- Anyone considering purchasing or commercialising a racing system
It is equally valid to arrive with “I think I have found something” or with “I have been running this for two years and something has changed.”
What we will not doIndependent means independent.
- We will not guarantee that a system is profitable.
- We will not manipulate an analysis to produce the answer a client wants.
- We will not describe historical performance as proof of future returns.
- We will not certify a system as certain to win.
- We will not claim an edge exists when the evidence does not support one.
Our job is to tell you what the analysis actually shows.
Five stages. One question throughout.
- TalkAn optional, confidential initial consultation.
- ScopeTell us what you have, what data exists, and what you want tested.
- ProtectAn NDA where required, and a secure data-transfer route agreed.
- Stress testElite Pass independently interrogates the system and its results.
- FindingsThe analysis, the weaknesses, the observations and the recommended next investigations.
An optional follow-up consultation is available once you have read the findings.
Think your system has an edge? Let’s find out what happens when somebody tries to break it.
All enquiries are treated confidentially.
Received. Your enquiry is treated as confidential and we will reply by email.
We could not accept that enquiry. Please check the name and email address and try again, or write to the address in the site footer.
Enquiries by email
Write to info@elitepassracing.com with a short description of the system and what you would like tested. Please do not send proprietary rules, code or datasets by email; a secure transfer route is agreed once we have spoken.
Research and analytical consultancy: Elite Pass provides research and analytical consultancy. Historical results do not guarantee future performance. Nothing in an audit is financial advice, a betting recommendation or a guarantee of profitability, and decisions made using the analysis remain the client’s own.
Not building a system? Elite Pass Gold is the race-by-race product, and the research behind both is public.
Read next
- Do horse racing systems actually work? What happens to a racing system when it is tested against the price that was actually available, rather than against chance.
- Why horse racing systems look brilliant in backtests The statistical reasons a historical result so often stops working, with the research behind them.
- How should a horse racing system actually be tested? Twelve steps for testing a racing idea properly, and why the answer is allowed to be no.
- Evidence and methodology How every number on this site is defined, sourced and checked.