HOW I TESTED IT
I swept 40 symbols and found three that looked profitable. I called them validated and put them live.
Then I tested the exact same settings on 751 days of price history the search had never touched. All three reversed. Not marginally - completely.
Three survivors out of 40 is roughly what pure chance produces at that sample size.
THE NUMBERS
In-sample result: +1.74
Out-of-sample result: -208.72
Days of untouched data: 751
Symbols that survived: 0 of 3
Searching hard enough through random data will always produce a winner. The out-of-sample test is the only thing that tells you whether you found an edge or just found noise wearing a costume.
Part of a public experiment: 5,063 USDT to 1,000,000 USDT, every result published - including the ones that make me look bad.
Run on $BTC $ETH $BNB
#Binance #Futures #Backtesting #TradingBot #CryptoData
I swept 40 symbols and found three that looked profitable. I called them validated and put them live.
Then I tested the exact same settings on 751 days of price history the search had never touched. All three reversed. Not marginally - completely.
Three survivors out of 40 is roughly what pure chance produces at that sample size.
THE NUMBERS
In-sample result: +1.74
Out-of-sample result: -208.72
Days of untouched data: 751
Symbols that survived: 0 of 3
Searching hard enough through random data will always produce a winner. The out-of-sample test is the only thing that tells you whether you found an edge or just found noise wearing a costume.
Part of a public experiment: 5,063 USDT to 1,000,000 USDT, every result published - including the ones that make me look bad.
Run on $BTC $ETH $BNB
#Binance #Futures #Backtesting #TradingBot #CryptoData
