8/19/20 Probable xCC and ConsecuSig

Note the third signal in xEM is E, if that holds the signal is xCC. Whatever, Thursday mornings make me nervous anyway. Kind of surprised $SPX hasn’t set a high, that really shouldn’t be a big deal, I’m pretty sure SPY and ES has. Talk about non-events, but that’s the dog days.

An interesting point was brought up in the comments to my latest article about performance in February 2020. During the week of February 24th, there were 5 x00s in a row. It would have behooved the algorithm to not go long on any of those. On the other hand, it made the money back in March and more. Not everyone has the strength to sit through February and not have that affect what you do in March, I sure don’t.

The first x00 was a good trade, that was buying the Monday weakness, very reliable but not here. That could have been exited when Monday’s low was taken out, but if you play monkey see monkey do there is no human to save the day, not to mention the days afterwards.

The red trend lines toward the left, show the day’s in question. The algorithm was long on 3/2 because that was also x00. It played March like a professor.

Anyway, in the FF5 game structure (today/yesterday) only buy signals can repeat. So I wrote a little routine that numbers consecutive signals. I considered this issue during the early design phase and figured that repeating signals were OK even though the x00 issue is the obvious elephant in the room.

This required a deep dive to 3/29/2001 with DIA/IWM/QQQ/ and SPY.

Turns out this has only happened 5 times, and it never happened more than that. Moreover, these situations are positive. That is, buying x00 after 4 x00s in a row makes money. 3/2 is an example. That is the same for all the x00 consecutive signals. The data issue seems to be that the information for the previous day’s losses is not available, and maybe it shouldn’t be. It is the same decision whether you were long or flat the day before.

Leave a Reply