Research · open data
Open data: our test results and statistics
The tables behind our tests, as CSV files you can download, check and reuse. CC BY 4.0: use them, including commercially, and credit Edge or Not with a link. These are our own derived tables (results, counts, distributions), not raw exchange data.
Downloads
| File | What it is | Columns | Rows | Size |
|---|---|---|---|---|
| tests_summary.csv | All ten tests in one table. Pre-registered tests and their verdicts. | test number, date, claim, data, sample, control, headline result, verdict, page | 10 | 3 KB |
| funding_rate_distribution.csv | Funding rate distribution. 1.8 million Binance funding payments, January 2024 - October 2026. | rate bucket per 8 hours, payments and share for all 522 coins and for bitcoin | 8 | 1 KB |
| btc_daily_level_sweeps.csv | Bitcoin: previous day's high and low taken out. Binance BTCUSDT daily candles, 2018 - October 2026. | days, share of days, closed back inside, closed beyond, median depth | 2 | 1 KB |
| bmsb_touch_episodes.csv | Bull Market Support Band touches. 17 touches, bitcoin 2018 - 2026. | date, close, band top, returns after 30/90/180 days, worst dip, band broke | 17 | 1 KB |
| new_listings_results.csv | New Binance futures listings: results by entry and hold. 348 listings, 2024 - 2026. | side, entry delay, hold, period, listings, mean, median, share in profit, t, p, control | 30 | 2 KB |
| funding_open_interest_results.csv | Funding plus rising open interest: results. 522 Binance coins, 2024 - 2026. | side, period, trades, coins, mean, median, share in profit, t, p vs pool, p vs funding alone | 6 | 1 KB |
UTF-8, comma-separated, a header row, shares and returns as fractions (0.05 = 5%), dates in UTC. Charts on the test pages are drawn from the same numbers.
The rule of every test, in one line
Each rule below was written before the data was loaded; the full version with controls, pass mark and known weaknesses is on the test page. How a verdict is decided: methodology.
| Test | Rule | Control | Pass mark |
|---|---|---|---|
| 1. Liquidation clusters | Every 4 hours take the biggest modeled cluster 0.5-3% above and below price; if one side is at least 1.5x heavier it is the target; which level does price touch first within 24 hours? | Mirror level on the other side | 55%+, p < 0.01, 200 events, forward period |
| 2. Pumps | Pump = hourly close at least 20% above the lowest close of the previous 12 hours on 3x volume; entry at the close; within 72 hours does price first reach half the move back or the same distance further? | Same-distance level on the other side; shuffled sides | 55%+, p < 0.01, both periods |
| 3. Uptober | Average and median bitcoin return by calendar month 2013-2026; is October's record more than shuffling returns between months produces? | 100,000 shuffles of returns between months | p corrected for 12 months |
| 4. Rainbow chart | Buy in bands 1-3 and sell in bands 7-9, with bands refitted each month on the data known at the time | Buy and hold; random strategies with the same trades | see the test page |
| 5. Altcoin season | At each month-end compare the next-month return of the top 50 altcoins (equal weight, 0.2% costs) with bitcoin, after readings of 75 or more | Every month vs bitcoin | see the test page |
| 6. Extreme funding | Short at funding of +0.10% per 8 hours or more, long at -0.10% or less; entry at the close of the payment hour, exit after 24 hours; 0.15% costs plus funding | Same position at every payment | t > 2, p < 0.05 on both periods, 30+ trades |
| 7. Funding + open interest | Funding of +0.05% or more (short) or -0.05% or less (long) with open interest up 10% in 24 hours; same entry, exit and costs as test 6 | Funding alone | t > 2, p < 0.05 on both periods, 30+ trades |
| 8. New listings | Enter 24 hours after the first candle of a new Binance perpetual, exit 7 days later, short or long; 0.15% costs plus funding | Older coins on the same dates | t > 2, p < 0.05 on both periods, 30+ listings |
| 9. Bull Market Support Band | After two weekly closes above the band, buy the close of the first day whose low reaches the band's top; hold 90 days; 0.2% costs | Random days in the same bull context | t > 2, p < 0.05 on both periods, 8+ touches |
| 10. Best month for altcoins | Average return of each calendar month for six altcoins; is the best of twelve more than 100,000 shuffles of returns between months produce? Monthly candles, Binance spot, 2017-2026 | Shuffled calendar months within each coin | p < 0.0083 (5% / 6 coins), positive in both halves, 6+ observations |
All results side by side: crypto trading strategies, tested.
Embeddable charts
Five charts you can put on your own site or blog. Free under CC BY 4.0: keep the source line under the chart. Each is a plain page that works in an iframe; add ?theme=light for a white background.
| Chart | Code |
|---|---|
| Ten popular crypto signals, tested | <iframe src="https://edgeornot.com/embed/scoreboard" width="100%" height="560" style="border:0;max-width:760px" loading="lazy" title="Ten popular crypto signals, tested"></iframe> |
| How funding rates are distributed | <iframe src="https://edgeornot.com/embed/funding-rate-distribution" width="100%" height="400" style="border:0;max-width:760px" loading="lazy" title="How funding rates are distributed"></iframe> |
| Bull Market Support Band: every touch, 90 days later | <iframe src="https://edgeornot.com/embed/bmsb-touches" width="100%" height="430" style="border:0;max-width:760px" loading="lazy" title="Bull Market Support Band: every touch, 90 days later"></iframe> |
| Altcoins by calendar month: share of positive months | <iframe src="https://edgeornot.com/embed/altcoin-june" width="100%" height="340" style="border:0;max-width:760px" loading="lazy" title="Altcoins by calendar month: share of positive months"></iframe> |
| Open interest by exchange | <iframe src="https://edgeornot.com/embed/oi-by-exchange" width="100%" height="260" style="border:0;max-width:760px" loading="lazy" title="Open interest by exchange"></iframe> |
How to cite
Edge or Not, "<table or test name>", https://edgeornot.com/data, CC BY 4.0, accessed <date>. A link to the test page is enough for web use.
Limits
- Binance data for most tests, mostly 2024-2026; coins delisted before 21 October 2026 are missing (each test page says which way this leans).
- Tables are snapshots at the date of publication: a test is not re-run when a month passes unless we say so.
- Past results describe past data. Not investment advice.
Changelog
- 24 October 2026: first release: nine tests summary, funding rate distribution, bitcoin level sweeps, Bull Market Support Band touches, new listings results, funding plus open interest results.
FAQ
Can I reuse these tables?
Yes. They are published under CC BY 4.0: use them, including commercially, and credit Edge or Not with a link to edgeornot.com/data.
Do you publish raw exchange data?
No. These are our own derived tables - test results, counts and distributions. Raw market data belongs to the exchanges and is available from their public APIs and archives; the sources are named on each test page.
How were the tables made?
By the scripts and rules described on the methodology page; each test page states its data, sample, control and pass mark. The rules were written before the data was loaded.
How often is this updated?
When a new test or descriptive table is published. The date of the last change is in the changelog on this page.