What are some metrics to collect to test transposition table efficiency?

[likeawizard]

Since the introduction of TT in my code I have had suspicions that it works poorly. It does seem to work but it seems to work strangely - it seems the TT barely makes a difference until very late game when a mate can be found or very few pieces present.

So I was wondering what kind of statistics would be interesting to collect from my search to peek into and improve the TT.
And what would be a good testing approach to see how the thing works and when changes are made if/how it improves/deteriorates?

[brianr]

Fine70 test is a good one:
8/k7/3p4/p2P1p2/P2P1P2/8/8/K7 w - - 0 1

should find a wining score not a draw (without tablebases)

[Ciekce]

SPRT, SPRT, SPRT