The preponderance of evidence suggests that rapid tactics training is a sound method for developing important pattern recognition even if one has no plans to play blitz chess (I'm too old, too slow, for that).
- My basic solution, thanks to some forum assistance, is to use custom looped sets that drop problems I get right fast enough (I'm thinking 15 seconds), but consider problems solutions even within 15 seconds as wrong if there is rating loss.
- I saw a suggestion here to compose such sets starting with a rating around 400 points below one's "normal" blitz tactics. If it's too easy, I will work the set down to nothing relatively rapidly. If too hard, I will adjust and try again.
- I did some solving to establish a base blitz rating: ~1300
Set A (Mates) Features- Rating type: blitz
- Rating range: 900-1000
- Number of moves: 1-2
- Total attempts for all users: 20 or more
- Problem type: winning problems
- Outcome type: only mates
- Quality rating: 4-5
- Color to move: any
- Number of pieces: 1-32 (default)
- Game move number: 1-300 (default)
Set A Previous Attempts- All problem sets / All Problems
- Match against most recent attempt only: False (not sure about this one)
- Treat loss of rating points due to time as wrong: True
- Shortest solve time: 0-15 seconds
- Include unsolved problems: True
- Use most recent time: False (not sure about this one)
- Days since last seen: 0-370 (default)
Set A Options- Rating adjustment: blitz
- Selection method: Sorted (looping)
- Sort direction: Ascending
- Sort by: problem rating
(until there is an option for random sorting!)
Set B (Non-Mates)I read a related suggestion that made sense to me: mix up mating problems with non-mating problems in a 75/25 ratio. So I created a second set, Set B, that is exactly the same as Set A except the Outcome Type is set to "Only Non-Mates." That resulted in a set of about 700 problems.
Mixed Rapid Training Set (merged)I used the Merge Sets feature to create a merged set with Set A at 75 percent and Set B at 25 percent.