In short, anything can happen. Fortunately by understanding price behavior, we can determine what is more likely to happen given defined technical conditions.
I want to take a look at 3 Fib charts based on Friday's ES action. Some people say cherry picking but as I look at the chart these three are about the only possible Fibs that could be drawn because the market action was either in a strong uptrend or flat. Here is the first one based on the early pre-market low. The rejection of the 23.6% provided an entry to enjoy the rest of the day.
Here is the 2nd one. This time the Fib is drawn starting at the low at about the Cash Market opening. Strangely it shows the same entry point but at the 38.2% level based on a new starting point. How about that! The same point worked on both fib drawings but at a the next fib level down.
The final chart I find amazing and it is drawn from the swing low entry point to the interim high. The market moved down and churned flat for hours but the new 38.2% retrace acted as a support level for hours before the final run-up.
And one more chart. After the final upmove started it offered one more chance to get in at the 23.6% retracement. See Chart below
Well I was going to post a chart backing up your first chart but I guess I'll go for an eye exam instead.
Honestly, there is too much bias in these examples. Its super easy to do this on a day where price just goes straight up. Of course if you have 3 fibs to choose from, the 23, 38, 50, one of them will hit nicely for a retracement entry. When doing this on a non trading day, the results will be far, far worse. Also, you're picking swing point that you know are the tops or bottoms, but in real time, you don't know this. Let me provide a counter example. Because we don't know about this being a super trend, shouldn't we assume that we also need to look for counter-trend trades? After the high at 4126, we put in lower high and lower lows, so it makes sense that it could be rolling over now. If we use the hindsight low at 4108 and draw our fib to join the downside, we see that the 23, 38, 50, 61.8, they all fail to act as a good short entry. Sure you can question why short in a strong trend, but most days do not trend like this, where it takes off from the open and never looks back. I would say this type of day is less than 10% of the days. The trend are never as clean. Furthermore, lets not forget that the low at 4108 is quite hindsight. In real time, perhaps we could have thought that the prior low at about 4113 looked like a low to use for our fib. Out chart would look like this below. Its zoomed in so we can see it well. Maybe this is too fast of a chart to use for fibs, but we see that its only the 50% level that gives a precise entry. The 23 and 38 fail. So my point is that anytime you say look, the 50% worked, it means that the 23 and 38 failed. You pointed out many times how the 23 worked in your example, so that means you're always going to take the 23, but naturally this has to fail in order to get to the 38, 50 and 61.8. So when you add up all the failures along the way, and all the different swing points you can use, do you really have a statistically valid system that will provide entries on all types of days? This day once again was exceptional since it was a relentless trend all day.