The Score

One number an allocator can trust, precisely because the manager being rated cannot touch it.

Any Sharpe ratio can be dressed up. Test fifty variants, publish the winner, and start the chart at the foot of a drawdown. The QuantLite Score ends that game. It is a single, defensible grade, built from the deflated Sharpe ratio, bootstrap robustness, tail risk, consistency, and track-record sufficiency, and it runs integrity checks that catch return smoothing, outlier dependence, and cherry-picked start dates.

Every rating travels as a verifiable artefact. Anyone who holds the original returns can reproduce it, to the last bit.

Independent by construction

A rating is worthless if the party being rated can buy a better one. That single flaw hollowed out the credit-rating agencies, and the QuantLite Score is built to be immune to it. A grade is firewall-clean only when its inputs are attested to a source the manager does not control, such as an exchange’s own fills or a fund administrator’s books, and never from figures the manager has supplied. The grade is a pure function of independently sourced data, so no amount of money can buy a better one.

Six grades, from exceptional to failing. The bands are fixed, public, and identical for every manager.

Integrity overrides the arithmetic. A single critical flag caps the grade at 40, a warning caps it at 70, and every flag travels with the rating, so a capped grade always states its reason.

Self-serve, without self-reporting

You never type in your own returns, which is the entire point, yet getting rated is still self-serve. You connect the source that already holds your record, and we read it there.

Connect a venue or broker

Authorise a read-only key to your exchange or broker. We pull your fills from the source, so the record is firewall-clean and you never touch the figures.

Rated by your venue

If your exchange or marketplace has integrated QuantLite, it rates the strategies it lists from its own fills, automatically.

Through an administrator or allocator

Your fund administrator, prime broker, or an allocator who holds your record can act as the attester.

You can run a self-scored figure for your own diagnostics, but it is marked not firewall-clean, and no serious allocator should treat it as more than a draft.

Open, like the methodology

You pay for reach and monitoring, never for the grade. Managers onboard themselves; venues and allocators are priced to the size of what they rate.

Self-serve onboarding by connected source is launching. Today you can integrate directly through the API. Paying more buys reach, never a better number.

Score and verify over HTTP

One open methodology, behind three endpoints.

What the figures mean

Deflated-Sharpe skill

A Sharpe ratio flatters anyone who tried many strategies and showed you the best. This measures the probability the edge is real once you account for how many attempts were made, and for fat tails. High means the returns are unlikely to be luck.

Bootstrap robustness

We rebuild the record thousands of times by resampling its own returns, and re-score each. High means the grade barely moves, so it does not hang on a handful of lucky months.

Tail risk

How bad the worst stretches are: the expected loss in the worst five per cent of periods, and the largest peak-to-trough fall. High means losses stay contained when it matters.

Consistency

Whether the gains are spread across time or concentrated in a single run. High means steady, and repeatable.

Track-record sufficiency

Whether the record is long enough to justify its Sharpe, measured against the minimum track-record length. High means there is enough data to believe the number.

Firewall-clean

The returns came from an independent source, such as an exchange, a fund administrator, a prime broker, or an allocator, and not from the manager. They cannot have been edited to flatter.

Integrity flags

Red flags the engine raises: return smoothing, dependence on a few outliers, a short-volatility signature, a cherry-picked start date, or too short a record. A critical flag caps the grade at 40, a warning at 70.

Input digest and content hash

SHA-256 fingerprints of the returns and of the rating. They let anyone reproduce the exact grade, and detect any tampering.

Questions

Can I rate my own track record?

Not for an official grade. The firewall requires the returns to come from an independent source, so a self-scored figure is always marked not firewall-clean. Score through your venue, administrator, prime broker, or allocator.

What stops a manager from gaming the score?

Three things at once. The firewall means the manager never supplies the data, the deflated Sharpe ratio penalises records that were cherry-picked from many attempts, and the integrity checks cap any grade built on smoothing or a few lucky outliers.

Is the methodology open?

Yes. QLS is open and versioned, and every rating is reproducible from the original returns. There is no black box to take on trust.

How is a rating verified?

Each rating carries a SHA-256 input digest and content hash. Anyone holding the original returns can call the verify endpoint and reproduce the grade, bit for bit.

How often is a rating refreshed?

Continuously, wherever the venue or allocator streams new data. Monitoring raises an alert when a record begins to deteriorate, before the headline grade moves.

What does a capped grade mean?

That integrity, not arithmetic, set the ceiling. A critical flag holds the grade at 40 and a warning at 70, and the flag travels with the rating so the reason is always on the record.