Why Tactic Trainer's rating system needs fixing
Apart from letting me play for free on their site ;), I have one real grudge with chess.com, and that's the Tactics Trainer rating system.
I have voiced my discontent before, and proposed an easy solution to the problem.
Recently I've entered in another discussion (wasn't the instigator;) on a TT puzzle comment section, and rolled some numbers in the process.
I'm comparing two hypothetical players, one (player A) taking his time making sure he gets it correct, the other one (player B) faster but more error prone because he needs to guess sometimes. I'm assuming linear interpolation between -4 for right/slow and +8 for right/fast, -8 for wrong, which should be close enough to reality.
Correctness (percentage of puzzles correct):
- Player A: 80% (phenomenal strength)
- Player B: 55% (merely average)
Speed (compared to other solvers):
- Player A: 25% (pretty slow, but still faster than 25% of people)
- Player B: 90% (fast)
So player A is getting 4 out of 5 puzzles right every day, but he's quite slow.
Player B is getting only between 2-3 of 5 right every day, but does it fast.
What would you expect for their respective point gain/loss after 100 problems?
Answer:
- player A: -240 points
- player B: 14 points
A rather clear confirmation that the current rating system ridiculously encourages guessing versus making sure you get it right.