Suppose I think that a gambling script which wins, say, 1000 times in a row should be possible. Of course! Not only is it possible: It is trivial. The problem is that depending on the odds of the game, it will need to lose an astronomical number of times to hit an unbroken streak of 1000 wins in a row. (Rather like after trying many billions of losing keys, I got one with a corresponding address which spells segwit.)
No. That's not how it works, gambling or hashing. The next bet or hash has the exact same probability of being a winning bet or a valid block hash or a super cool vanity address as any other regardless of any bets or hashes preceding it.
You mentioned someone pointing you to github and that might help, but for a proper context you also might want to read some posts here in the gambling section. There are some brilliant folks here who know how this shit works and learned some of it the hard way and don't hesitate to share stuff that you won't find in source code.
However I remember that she said her ''strategy'' was based on a vulnerability on a particular casino because the casino itself had some sort of script that would let new accounts win at first or something like that. It wasn't really a normal idiotic ''strategy'' at least that's what she said. I personally wouldn't vouch for something like that even if it was legit since it involves some sort of cheating.