Tools for the quant
Find spreads that survive the statistics.
Research tool, not advice.
From universe to evidence, in three screens.
Set the thresholds, get a ranked shortlist, then open any pair and see every test behind its score. No black box at any step.
| Pair | Sector | Score | Corr | 90d MA | |
|---|---|---|---|---|---|
| V / MA | Payments | A87 | 0.94 | +7.6% | Drilldown → |
| KO / PEP | Beverages | A84 | 0.91 | +4.8% | Drilldown → |
| UNH / CI | Insurers | A82 | 0.89 | +4.1% | Drilldown → |
| XOM / CVX | Integrated | B76 | 0.87 | +2.9% | Drilldown → |
| GS / MS | Banks | B68 | 0.85 | +1.4% | Drilldown → |
Illustrative mockup — not live data, historical results, or a recommendation.
Four steps, all documented.
No black box. Every stage below links to the exact method, test, and threshold used.
Build the spread
Daily adjusted closes across 521 symbols — the S&P 500 plus 21 sector and index ETFs — are turned into a price ratio for every one of 135,000+ candidate pairs, one consistent construction applied the same way every time.
The method →Test for cointegration
An escalating test hierarchy — Johansen, Gregory–Hansen, Phillips–Ouliaris — separates a genuine long-run relationship from two things that merely drifted together.
The method →Measure the reversion
ADF stationarity testing plus an Ornstein–Uhlenbeck half-life estimate: not just whether a spread might revert, but how fast it historically has.
The method →Rank and backtest
Survivors are scored, gated for regime breaks and imminent earnings, then backtested net of commission, spread, and market impact.
The method →Questions, answered plainly.
Isn't this just correlation?
No — correlation measures co-movement, not a stable long-run relationship. RINS runs a full cointegration test suite (Johansen, Gregory–Hansen, Phillips–Ouliaris) specifically because correlated pairs can still drift apart forever. Correlation is one input among several, not the answer.
Do your backtests include trading costs?
Yes. Every backtest nets out commission at $0.01 per share per leg, bid–ask at 2 basis points per leg, and market impact on a square-root model — sized at $50,000 per leg by default, and all of it configurable. A reversion has to clear those costs before it counts as a win. See the cost model.
How is the score calculated?
Each pair gets a 0–100 score built from five weighted components — mean reversion, regime stability, cointegration, ARIMA structure, and correlation stability. The weights and the tier cutoffs are documented rather than hidden; see the composite tradability score.
What happens when a pair stops working?
That's the case the tool is built for. PELT change-point detection partitions the history into regimes, the Gregory–Hansen test looks for cointegration that only holds either side of a structural break, and rolling stability windows track whether the relationship is still holding. A pair whose relationship has decayed is shown as decayed, not quietly re-ranked.
Why do some pairs get excluded?
Because a gate rejected them — an earnings report inside the blackout window (5 trading days by default), a regime break, or a failed cointegration test. The gates are designed to reject candidates, not to flatter them, and the reason is always shown alongside the pair.
Can I analyse my own pairs, or only what the screen returns?
Both. The overnight screen ranks the universe for you, and you can also look up any specific pair directly or keep your own watchlist and track it over time.
Do I need a statistics background?
No. Every method RINS uses is written up in plain English in the methodology, with the statistical terms explained as they come up. You should understand what a diagnostic means before you act on it — which is exactly why it is public.
Is this financial advice?
No. RINS is a statistical research and information tool. Every output — scores, signals, analysis, sizing calculations — is a research diagnostic, not a recommendation to buy, sell, or hold anything. See our Disclaimers & Risk Disclosures for the full statement.
What data do you use?
Daily adjusted closes across the S&P 500 plus 21 sector and index ETFs — 521 symbols, or 135,460 candidate pairs — sourced from third-party market-data vendors. Professional and above also cover 43 FX instruments (majors, crosses and EM). See the methodology for exactly how the price ratio is constructed.
Do you place trades for me?
No. RINS never executes, routes, or places trades. It surfaces research diagnostics; any decision to act on them is yours alone.
Built by RINS.COM LLC.
The name is the method. RINS builds statistical research tools for pairs and spread traders — integrating published numerical techniques into one consistent pipeline. We publish the methods we use: every test, threshold, and cost assumption is documented in public, because a research tool you can't audit isn't research. RINS is a software company — we don't custody funds, place trades, or manage money.
Stop eyeballing charts.
Start measuring relationships.
Research tool, not advice.