I am not sure if this is already discussed but I found out that 20 EMA crossing 100 EMA has marked Bitcoin Bottom in the past.
What do you think, is this correct even this cycle or we are about to see this fail for the first time?
I think that small sample error is too big to even consider this as a valid indicator. So far number of successful repetitions equals 1 because in 2014-15 you have event N=0 based on which you could create indicator and in 2018-19, during N=1, it successfuly predicted bottom. Now we have N=2 and its not yet confirmed. So its not like we are talking about possible "fail for the first time" event, we are taking about succeeded for the second time.
Its like you play coin toss. During first game heads won. Based on that you predict that heads will always win. You throw again and heads won again. Based on that you create a thread, here on bitcointalk, asking as if heads will fail for the first time during your next throw. Of course its possible and going back to TA and EMA20 and EMA100, based only on outcome we cant even estimate probability (because unlike coin toss its not 50:50) and N=1 or N=2 is way too small sample to say anything.
People say -"ok, small sample error, but we had only 3/4 bull runs in crypto, there is no more data". Ok but we had hundreds of those in stocks/comedies/gold etc. and based on this dozens of TA books were created. There is no need to create new indicator only for bitcoin only because it succeeded once or twice.
"bad data is worse than no data at all"