It's how fast the pool finds blocks compared to its proportion of the global hashrate. For instance:
BTG block time: 10 minutes
Pool hashrate: 5 MH/s
Network hashrate: 50 MH/s
Pool has 10% of total hashrate, therefore should find blocks every 100 minutes on average
if it finds blocks with a 90 minutes average, luck = 90 minutes / 100 minutes = 0.9

Anything slightly below 1 is good, because indeed lucky (if you are superstitious), or because it has a slightly faster server connection or something.
Anything slightly above 1 is bad,
grossly above or below = fishy af

Luck is sometimes worded as "effort". Again, less effort to find blocks = good.

pool.gold ... well... just look effort from the last 20 blocks: 1.58, 1.06, 1.87, 6.14, 1.60, 5.90, 4.75, 0.22, 1.21, 4.17, 2.33, 9.59, 3.38, 2.49, 3.73, 0.42, 7.46, 4.42, 3.43, 0.69

Average luck, last 20 blocks... a whopping 3.32

Accepted shares don't mean much. Suprnova (and many other pools) use Vardiff. Your difficulty increases if your hashrate increase (but the value of your share increases too). This prevents pools from being spammed by low diff shares from huge rigs, and allows very small rigs to submit more than 1 share per hour.

Ultimately, you should get (total BTG created in a day) * (your hashrate / total network hashrate), and then factor in Pool's luck.
Suprnova luck is normal (like most pools). Pool.Gold's bad luck is a statistical aberration.

pool.gold is most definitely a scam. They have had roughly 25% of hashrate since day 1 and barely pull 12-15% of blocks.
Looks like some sort of Selfish Miner attack or Block Withholding attack.

Suprnova works a-ok.

I stopped mining BTG and sold all my coins altogether though. I just do not trust the project anymore, and don't see its added value to the ecosystem.