I found the ADS-B data and graphs related to MU5735 on a public tracking site ( cathaypacific8747.github.io/mu5735/ ). While I'm not completely sure about the accuracy of the data, it still provides interesting insights.

One thing I noticed is that the heading remained close to 100 degrees, even during the descent. This makes me believe that, up until the final moments before the crash, the pilot may have still been trying to recover the aircraft. However, during parts of the descent, the heading fluctuated significantly — jumping from around 100° to 140°, and even down to 80°. That kind of unstable behaviour doesn’t make much sense if we were considering a deliberate, calm pilot suicide — especially with such erratic heading changes, which might suggest loss of control or conflict in the cockpit.

As for airspeed, it's generally expected that during a rapid descent, airspeed increases unless the pilot reduces thrust or deploys speed brakes. In the 737-800, during normal descents, pilots typically use either Level Change (LVL CHG) or Vertical Speed (VS) mode, and they rarely disconnect autopilot unless there’s an emergency.

Based on the data, it seems like the aircraft initially maintained airspeed, which suggests the autopilot may have still been engaged, likely in Level Change mode — this would explain the stable airspeed and typical descent profile at first. However, later on, the airspeed began to decrease, which is one of the most mysterious parts to me. This could have happened because the pilot manually reduced the speed in the MCP (speed window), or because the autopilot was disengaged.

Then comes another strange moment — the vertical speed briefly went positive, meaning the aircraft climbed slightly after the initial dive. I have two possible scenarios for this:

1. If the autopilot was still engaged, and the pilot reduced the target speed in the MCP, then the aircraft might reduce descent rate (or even climb slightly) to bleed off airspeed — this would be a logical autopilot reaction under LVL CHG mode.
   

1. If the autopilot was already disengaged, then everything was under manual control — this supports the suicide theory, or possibly a human factor like confusion or error.
   

Either way, after that brief anomaly, the aircraft entered a massive uncontrolled descent, reaching over –10,000 feet per minute. At one point it dropped below 10,000 feet, which is extremely dangerous in that mountainous region. Strangely, the aircraft’s vertical speed then reversed and became positive again — around +10,000 feet per minute, which is an extremely aggressive pitch-up maneuver.

This is, in my opinion, the craziest moment of the entire flight path. Why would the pilot pitch up so violently? My best guess is that the pilot was still trying to regain control or survive, possibly under extreme psychological or technical pressure. It’s impossible to fully understand what happened in the cockpit without the cockpit voice recorder (CVR).

Finally, in terms of airspeed management, I believe someone was actively managing the speed throughout the flight — maybe not perfectly, but it didn’t look like a fully uncontrolled scenario. The recorded speed stayed around 450 knots, which, in ADS-B terms, likely represents ground speed (not indicated airspeed). This adds even more mystery, because if the aircraft was in a steep descent and the autopilot was off, we might have expected that speed to fluctuate more wildly. Furthermore, once the aircraft descended below 10,000 feet, we would normally expect the pilot to set the target speed to 250 knots or lower in the MCP, following standard ATC and operational procedures. In Boeing 737-800s, especially under Level Change mode, the aircraft would then pitch up or down to maintain that selected speed. If the ADS-B data shows the aircraft was still flying significantly above 250 knots below FL100, that further raises questions — either the pilot was no longer managing speed normally, or there was a deeper issue preventing proper descent configuration.