8 and 21 day moving averages FTW

Phase 1. Added 8 and 21 day EMAs to the BTC daily chart . Started looking from ~10 months ago. Used these rules:

Strong trend reversal signs (open a position; close any opposing positions):
• The 8 day and 21 day EMAs cross.
Crosses are marked with a vertical line on the chart. Red for crossing down; green for crossing up. The open long or short is marked with a green or red dashed horizontal line that extends until it's closed.

Weak trend reversal signs (decrease your position only; marked with a short solid horizontal green line for buy and red line for sell):
• Price tests the 21 day EMA twice in a row (candles both apt. colour)
• Price crosses the 21 day EMA and closes beyond it

Trend continuation signs (increase your position only; marked with a short green/red horizontal line):
• Price bounces/is rejected from the 21 day EMA

You can see that the above rules produced much more profit than loss.
Problem: It did produce losing positions. Also, waiting for the cross means waiting until the daily close, which can be quite a big candle, taking price in the direction you don't want.

Phase 2. Can we remove the losers?
Insight: Price always retests the 21 day EMA after the 8 day and 21 day cross. Either it bounces from it or it crosses again.

Extra rules:
Wait for the cross, then monitor for a maximum of 6 days. You want price to back off a bit from the 21 day EMA and then approach again.
Try to get an entry as close as you can to the 21 day EMA within 6 days of the cross. Price is unlikely to actually touch the EMA again if the cross is sustained. It's not like retesting horizontal support. The distance from the EMA isn’t important, just a noticeable movement away and towards again.
Set stop loss just the other side of the 21 day EMA and adjust daily.
If price crosses the 21 day EMA , your setup has failed. But the distance and the loss will be small.
If price bounces from the EMA , your setup is confirmed AND you got a good price.

These extra rules eliminated ALMOST ALL the losses in this particular backtest. They also gave slightly better entry prices in all cases (depending how strict you are with waiting for the daily candle to close).

Crosses that failed the above rules and would not result in opening/closing a position are marked with a purple oval in the chart and require no action.
Places where price "retested" the slower EMA are circled in green.

Results:
6 positions were taken in 10 months (304 days), an average of one every 51 days. Additionally, there were 21 trades that increased or decreased existing positions, an average of one every 2 weeks. There was one unprofitable position, where the loss was roughly equal to the risk capital minus the amount taken profit on in one event (e.g. if you take a conservative 50% profit on each reduction signal, the loss would be 50% of your risked-amount-per-trade).
The other positions were all closed in good profit; the open short is in good profit.
No other indicators were used.