Tangerine 10 by the numbers
Three studies built from the game's own code: what a random board contains, which play style scores most, and every way to make ten. Each one comes from a simulation script, not from guesswork.
What 100,000 Tangerine 10 boards look like
A fresh desktop board offers 54.65 moves on average, 34.77 of them adjacent pairs. 64.94% of nines start with no 1 beside them.
Pairs or big boxes? Four bots, 120 seconds, real scoring rules
With a fast hand, pairs-first averaged 473.31 points against 334.54 for biggest-box-first. At 3 seconds a move the order flips: 95.42 vs 83.5.
Every way to make ten, and how often each one shows up
29 combinations of two to five numbers make ten. Two random tiles do it 11.1% of the time, five only 0.21%.
How these were made
The script scripts/simulate.mjs copies the line of game code that fills a board — a random whole number from 1 to 9 in every cell — and uses a fixed random seed so the results can be reproduced exactly. It generated 100,000 boards for each board size, and ran four automatic players for 5,000 games under each of three speed assumptions, using the real 120-second timer, two-second combo window and multiplier table. The pages read their numbers straight from the script's output file.
For the rules themselves, see How to Play; for practical habits, the strategy guide.