Research: does it have an edge, or not?
Popular crypto trading ideas, tested with rules written down before any data is loaded, a control for luck, an out-of-sample check and trading costs. Results are published whatever they say.
All results side by side, grouped by strategy type: crypto trading strategies, tested. Or guess them first: Edge or Not? quiz.
Does extreme funding predict reversals?
Shorting high funding on 522 coins won 55% of trades, but a few squeezes took it all back. Buying negative funding mostly lost.
Does altcoin season predict altcoin gains?
After 8 altcoin season readings since 2019, altcoins beat bitcoin the next month only 3 times. Over 82 months the top 50 lagged BTC by 1.8% a month.
Does the bitcoin rainbow chart work?
Drawn with hindsight it beats buy-and-hold. Drawn only with data known at the time, the same buy-blue, sell-red rule turned $1 into $22.
Is Uptober real?
October was bitcoin's best month by median and green in 10 of 13 years - but with 13 Octobers that is not distinguishable from chance. No month passes.
Do crypto pumps always dump?
1,215 pumps on Binance futures. The pullback came first no more often than chance; shorting won 69% of trades, but 2.2% lost more than the whole position.
Do liquidation heatmaps predict price?
Does price get pulled toward the biggest liquidation cluster? First 30 days: no edge. The forward test on new data is collecting events.
How every test works
- Rules first. Entry, exit, time frame and the pass criterion are written down and saved before any data is loaded.
- A control for luck. The signal is compared with random entries, the opposite side or shuffled labels. It has to beat what chance produces.
- Out of sample. A result that holds only on the part of history used to look at the idea does not count.
- Costs included. Fees and slippage are subtracted where a trade is involved.
- Published as is. "No edge" is a result. The full method and labels are on the about page.