Experimental indicator, anyone? I give you the... reversal index

[Wary]

2014-06-26

trend: down

intraday median daily price: $582
intraday daily MACD hist: -11.0

RI1 = 57
RI2 = 50

comment: interesting. one or two days of positive price action, and RI1 crosses 50, i.e. signals reversal. But as I said one post above, I'm worried about false positives now that I made it more sensitive, which is why I'd rather look at RI2, which is at 50 right now, i.e. waiting for a bit more confirmation before calling it a reversal.

The indicator is broken! It shows wrong date! 

[oda.krell]

2014-06-28

trend: down

intraday median daily price: $598
intraday daily MACD hist: -7.0

RI1 = 60 78 [corrected wrong value for R1]
RI2 = 51

comment: RI1 now firmly in 'reversal' territory. RI2 still more cautious, but slightly leaning towards reversal as well. If I would trade on those signals (which I don't.) (yet.) now would probably be the time to take their advice and buy. MACD substantially up since yesterday, but still negative, so I'm happy that RI and MACD "disagree", which will make it more interesting later to compare which signal turned out to be more profitable.

[oda.krell]

2014-06-29

trend: down

intraday median daily price: $596
intraday daily MACD hist: -5.5

RI1 (2) = 90
RI1 (3) = 65

RI2 (1) = 58
RI2 (2) = 53

comment: hope it doesn't get too confusing, but I need to introduce one more distinction, this time in the form of a parameter for both variants that I'd like to keep track of, to see which one performs better.

RI1 (2) = previous RI1 = the "original" RI
RI1 (3) = slightly more "cautious" version of RI1

RI2 (1) = previous RI2 = the "cautious" variant of the original RI
RI2 (2) = even more cautious version of the (cautious) RI2

In any case, all 4 of them are now well above 50, for the second day, which means the RIx signals are clearly saying 'up' by now.

[oda.krell]

2014-06-30

trend: down

intraday median daily price: $617
intraday daily MACD hist: 0.2

RI1 (2) = 99
RI1 (3) = 85

RI2 (1) = 60
RI2 (2) = 53

[oda.krell]

2014-07-01

trend: down

intraday median daily price: $648
intraday daily MACD hist: 8.4

RI1 (2) = 99
RI1 (3) = 99

RI2 (1) = 99
RI2 (2) = 99

comment: um, yes... Not much point anymore reporting the RI values during this swing, unless it turns around again, they're all at max anyway.

MACD signal would have, according to the common interpretation, signaled a buy today, because yesterday MACD hist went positive again.

So... buying signal MACD at ~$640 (today's opening price) vs. RI signal at $581 (on the 27th, the "risky" choice) or $598 (a day later, the "safer" choice).

[Raystonn]

So when does your trend change to up?

[oda.krell]

So when does your trend change to up?

Not my choice. The only "manual" choice is the starting point, which is June 1st @$683.

So I guess when and if we go above that value, and stay there for a bit.


EDIT: though I admit, it might make more sense to set a new starting point once the RI signaled reversal. As you can tell, I haven't fully thought this through yet

[Kramerc]

I'm liking this a lot. Interesting experiment but fortunately I was already all-in by the time the indicator reversed. Maybe next time I'll use it, provided you keep updating the thread. So thanks for a good read.

[oda.krell]

I'm liking this a lot. Interesting experiment but fortunately I was already all-in by the time the indicator reversed. Maybe next time I'll use it, provided you keep updating the thread. So thanks for a good read.

Thanks as well.

It's probably too early to really trust the signal with any significant amount of money, at least before testing it some more. As I wrote a few posts above, I tuned it to be a lot more "sensitive" to an impending reversal, so the question is now whether it'll give too many false positives.

Apart from that, I don't think it'll ever become as accurate and profitable as a good trader who goes bottom fishing. My goal is not to beat discretionary trading, but to get an automated (or almost automated) signal that is *slightly* better than, say, daily MACD. If I can get there, I'd be pretty happy. I still wouldn't trade based on the RI alone, but I would be able to include it in my inventory of indicators that I use to make decisions.

[zimmah]

2014-07-01

trend: down

intraday median daily price: $648
intraday daily MACD hist: 8.4

RI1 (2) = 99
RI1 (3) = 99

RI2 (1) = 99
RI2 (2) = 99

comment: um, yes... Not much point anymore reporting the RI values during this swing, unless it turns around again, they're all at max anyway.

MACD signal would have, according to the common interpretation, signaled a buy today, because yesterday MACD hist went positive again.

So... buying signal MACD at ~$640 (today's opening price) vs. RI signal at $581 (on the 27th, the "risky" choice) or $598 (a day later, the "safer" choice).

so, you are predicting it to go down from here?

or do i misunderstand how this works?

[oda.krell]

2014-07-01

trend: down

intraday median daily price: $648
intraday daily MACD hist: 8.4

RI1 (2) = 99
RI1 (3) = 99

RI2 (1) = 99
RI2 (2) = 99

comment: um, yes... Not much point anymore reporting the RI values during this swing, unless it turns around again, they're all at max anyway.

MACD signal would have, according to the common interpretation, signaled a buy today, because yesterday MACD hist went positive again.

So... buying signal MACD at ~$640 (today's opening price) vs. RI signal at $581 (on the 27th, the "risky" choice) or $598 (a day later, the "safer" choice).

so, you are predicting it to go down from here?

or do i misunderstand how this works?

No, very much not. The prediction is more or less over for now, but a few days ago, my RI predicted we'd go up. As of today, MACD agrees.

Could you let me know where you got that idea from, so I can work on improving my way of presenting the results?

[oda.krell]

Quick single day update..

2014-07-11

larger trend (since June 1) = down

RI1(2) = 85
RI2(1) = 46


subtrend (since June 26) = up

RI1(2) = 62
RI2(1) = 46

comment: By my interpretation (where RI1 is the leading indicator and RI2 should be used for confirmation), this means mid term, we're most likely breaking out of the downtrend since June 1st (but would want to see the RI2 at 50 or above to be sure). Short term there's potential to go down some more, i.e. break down from the June 26 upwards trend, especially if the shorter term RI2 crosses 50. (EDIT) just to be sure: it's coincidence that both RI2 are at 46. They're based on different data, from their respective trend.


(EDIT) adding one more time scale...

even larger trend (since May 19) = up

RI1(2) = 33
RI2(1) = 19

comment: In other words, the trend that took us from ~$450 to where we are now shows no signs of reversing (by the RI metric).


(edit) Here's a (qualitative) summary of the above (quantitative) results:

The currently most robust trend, from May 19, appears to be 'up'. It depends now which time frame's reversal is confirmed first, i.e. whether the June 1 downtrend RI2 crosses 50 first (then: likely upwards breakout) or the June 26 uptrend RI2 crosses 50 first (then: likely downwards breakout). Given the current data, the upwards breakout is the more likely scenario (higher RI1 value, and embedded in the larger May uptrend).