Strategy Examples

This page collects public Stingray examples where a plain-English market idea became a typed rule and a backtest artifact. The goal is proof over pitch: show the prompt, the data sources, the replay window, the trigger count, and the forward-return shape before anything goes live.

These are published product walkthroughs, not anonymized customer case studies. When customer examples are published, they should be permissioned, anonymized, and specific enough to audit.

What each example should show

• The prompt or thesis the user started with
• The typed condition Stingray produced
• The data sources used during the replay
• The backtest window and trigger count
• Forward returns or hit-rate summaries at relevant horizons
• The live path: notify, monitor, and review

Funding-rate squeeze example

The prompt:

Create a draft alert that fires when BTC funding rate on Hyperliquid is below 0 bps/hr and Binance BTCUSDT moves up at least 0.5% over 60 minutes. Keep it as a draft and do not send it live.

What Stingray tested:

• Data sources: Hyperliquid funding and Binance BTCUSDT price
• Window: 2026-04-20 to 2026-05-18
• Raw matches: 150 five-minute windows
• Fired alerts after 1h cooldown: 37
• 24h profile: average +0.41%, 62% positive, median +0.55%
• Compound comparison from non-overlapping 24h holds: +8.3% versus BTC buy-and-hold at +4.9%

Read the full funding-rate walkthrough.

BTC breakout alert example

The prompt:

Alert when RSI is above 65, price touches or breaks the upper Bollinger Band, and volume is 2x the 24h average, all true at the same time on the 1h chart.

What Stingray tested:

• Data sources: BTCUSDT price, RSI, Bollinger Band, and volume
• Window: April 2025 to April 2026
• Fired alerts after 1h cooldown: 109
• Frequency: about one fire every 3.3 days
• Forward-return profile: +0.03% at 1h, -0.26% at 4h, +0.17% at 24h
• Use case: reject loose breakout logic before a live alert starts waking anyone up

Read the full BTC breakout walkthrough.

How to read these results

The funding-rate example has the stronger 24-hour profile and beats its stated buy-and-hold baseline, but it contains only 37 cooldown-filtered fires in a four-week window. Treat it as evidence worth testing out of sample, not as live performance or proof that the edge persists.

The breakout example covers a longer period and 109 fires, but its average forward returns are close to zero and turn negative at four hours. That is a useful negative result: a familiar setup can be made precise, tested, and rejected before it becomes a noisy live alert.

More strategy inputs to test

These examples use markets with unusually complete public data. The pattern is simple: state a thesis, resolve the data, replay it, then decide whether the idea deserves live monitoring and more resources.

• Funding rates plus macro events
• Polymarket odds as a trading signal
• FOMC transcripts as strategy inputs
• Whale flow and open-interest divergence

Case studies by market question

The following articles are research frameworks rather than published performance records. Each shows what a credible test would need and where the evidence can fail:

• DOGS and meme-token momentum: separates price, volume, liquidity, and event-time evidence from social attention. The main limitation is short, regime-dependent history.
• STBL yield and spread strategies: separates principal, yield rights, liquidity, peg risk, and redemption conditions. Missing protocol history must be reported, not replaced with a convenient proxy.
• Hyperliquid algorithmic trading: combines venue-aware funding, open interest, price rules, cooldowns, and monitoring boundaries. A backtest is only relevant when its sources and rule match the live venue workflow.

What these examples are not

They are not trading advice, live performance promises, or proof that a rule will work in the future. They are proof that the rule was made explicit, tested against a stated historical window, and turned into an artifact someone can inspect before capital is involved.

For the full workflow, start with How Stingray Works, learn how to read a backtest card, or browse the Strategy Guides.

For a versioned side-by-side table with citation guidance, use the Backtesting Benchmark Report. To reproduce the first example step by step, use the No-Code Backtesting Tutorial.