EMA(2) line reverses from below the candles to above a downward gap candle.

The ema(2) you are using on this chart is actually ema(2,-2) - in other words, the plot is of the ema is offset, or moved back two bars. So, the ema value that appears to occur on Dec 20, for instance, doesn't actually occur until Dec 22. (Also, note that logically, the ema cannot turn up before the price does and price doesn't turn up until Dec 21. Any MA is a lagging, not a leading indicator.)

Likewise, the instances where the ema(2) appears to be falling "under" the down bars (e.g. the streak of down bars in September before your first circle), is not "real"- it's because of the offset. In "real time", the ema(2) intersects most or all of the bars (as you would expect from a very short ema length).

Nevertheless, I think you have caught an interesting pattern, but I would test for it this way: first test for a falling ema, say, yesterday and the day before, then test for today's ema above yesterday's ema (the ema turns up), and then test for a close above today's ema. So, something like this:

// test for falling ema recently; you could extend this a couple more days maybe
and [1 day ago ema(2,close) < 2 days ago ema(2,close)]
and [2 days ago ema(2,close) < 3 days ago ema(2,close)]

// test for ema up tick today
and [ema(2,close) > 1 day ago ema(2,close)]

// test for close today above ema today
and [close > ema(2, close)]



I didn't include the gap, but if you want it:

// test for gap between yesterday and the day before

and [1 day ago high < 2 days ago low]

Note that on a stock like SONY, which trades on other foreign exchanges while US exchanges are closed, the gaps probably do not have technical significance - they are due to (probably) normal price changes on those other exchanges while US markets are closed. They do not necessarily represent buyer or seller capitulation, which is what a gap represents in a one market (or simultaneous market) symbol.

Sorry to hear about your loss. Only time heals. God bless.