Skip to content

v1.6.2 - RP reliability fixes

Choose a tag to compare

@yzRobo yzRobo released this 22 Jul 17:49
· 8 commits to main since this release

v1.6.2 — RP reliability fixes

Diagnosed from a week of live data: some tracked players got a 0 out of
nowhere
, one row had an impossible −243, a fast back-to-back pair both got
the same −65, and a couple of rows carried literal 0 RP totals. Three
root causes, three fixes:

Fixed

  • Bogus ALS readings are now rejected. EA/ALS occasionally serves
    rankScore: 0 for a player who very much has RP. That glitch used to poison
    the baseline, fake a "season reset", and fabricate deltas. It now counts as a
    failed poll — the tracker keeps the last good reading.
  • Baselines are chained across matches, durably (new rp_chain.json). The
    next match's baseline is corrected against the previous match's observed
    end value. This fixes two real bugs from the data:
    • fast back-to-back games no longer attribute the same RP movement to both
      matches (the second game's snapshot froze before the first game's RP had
      propagated);
    • a squadmate who dies and leaves early no longer gets a false 0 (their RP
      posts before the summary screen is read, so the old baseline already
      contained their gain).
      The chain survives restarts and even a cleared pending queue.
  • A 0 is only recorded with pre-match evidence. If the tracker never
    observed the baseline before the match ended (e.g. it was started on the
    summary screen), a flat reading proves nothing — the match resolves as an
    honest blank instead of a fabricated 0. This kills the
    "first game of the night is always 0" bug.
  • Last-ditch sanity: nothing beyond a believable single-game loss and no
    zero/negative RP totals can ever be written, whatever upstream produced them.

Install / update

Click Update now in the app, or download ApexTracker_share.zip, unzip
anywhere, and run Start Tracker.bat. Settings and match history are kept.