[Myth Busted] The Sunday Dip

I crunched the numbers correlating close price and dips with days of the week. The idea of the "Sunday dip" or the "weekend dip" is prevalent in this sub, and people here are basing their financial decisions on it. But is it real? Let's find out.The data is taken from Binance (the largest trading platfom) and from the BTCUSDT pair (the most traded BTC pair). I fetched the candles from the 1d chart, and ran two types of analysis:Did the day close down from the previous day.Was there a significant dip (>5%) during the day, whether or not the day closed down.I ran this over two timeframes: 2 years to get a broad range of data, and 3 months to get the most recent data.The count was tallied in each case to produce a single figure - i.e. the number of Sundays where the price went down over the specified period, or the number of Mondays where there was a significant dip over the specified period.Here are the results:2 year timeframeDown daysSunday: 50Monday: 49Tuesday: 50Wednesday: 47Thursday: 51Friday: 47Saturday: 39Dip daysSunday: 19Monday: 16Tuesday: 18Wednesday: 17Thursday: 16Friday: 16Saturday: 103 month timeframeDown daysSunday: 9Monday: 6Tuesday: 4Wednesday: 8Thursday: 7Friday: 5Saturday: 4Dip daysSunday: 4Monday: 4Tuesday: 2Wednesday: 3Thursday: 4Friday: 4Saturday: 2ConclusionThe most statistically-significant set of data is the 2 year timeframe, this shows:no correlation with Sunday being a down dayslight correlation with Sunday being a dip daygood correlation with Saturday being an up daygood correlation with Saturday having fewer dip daysOn the 3 month timescale (caveat here is that the sample size is probably too small to make a conclusion, but I've added it in anyway):slight correlation with Sunday being a down dayno correlation with Sunday being a dip daygood correlation with Tuesday and Saturday being up daysgood correlation with Tuesday and Saturday having fewer dip daysThe long timeframe confirms that this myth is busted, with the data being well within the margin of error. For illustration of the margin of error, imagine just one of the Sunday dip days being on a Wednesday instead - that would have left Sunday, Tuesday, and Wednesday all on 18.The only interesting thing I would take from this as a potentially tradable correlation is Saturday having fewer dip days as it had 10 vs the mean of 16. But this figure is probably still on the edge of the margin of error.

Submitted May 28, 2021 at 03:25PM