I'd say draw for the top 5, lower the the percent to 20% for the 1st pick and go down from there, and do something like this: (Let me know if anyone actually understands this math, lol).
If a team got a top 5 pick the year before, their lottery chances would drop as follows:
5th: 5% (Worst team would have a 19% chance instead of 20%)
4th: 10% (Worst team would have a 18% chance instead of 20%)
3rd: 15% (Worst team would have a 17% chance instead of 20%)
2nd: 20% (Worst team would have a 16% chance instead of 20%)
1st: 25% (Worst team would have a 15% chance instead of 20%)
If having multiple top 5 picks in successive years, it would add onto each other. For example: the Lakers had the 2nd pick the past two years, therefore if they were the worst team this year (I know they don't have their pick), they'd have a 9% chance at #1 (A Reduction of 55% [20% + 20% + 15%]). It would add until a teams lottery chances were reduced by a maximum of 75% (5% lottery odds for worst team, etc.)
The extra ping pong balls would go to teams that didn't have a top 5 pick last year. They would be reallocated according to a curve (worse teams would get more).
Extra rule: If a team trades a pick over a year before it's respective draft, they receive no deduction if the other team receives that pick. (Essentially, the Nets wouldn't have reduced odds next year if the Cavs receive their pick in the top 5 this year. Same for the Lakers and Celtics).
This would decrease the incentive for tanking in multiple ways:
1.) The longer a team "tanks", the less rewarding it is. The Lakers and 76ers wouldn't have been as rewarded for tanking as they were.
2.) The reallocation and lesser curve makes the difference between the top of the draft and bottom less so, meaning less incentive to rise in the draft.
3.) The 6th pick might be better than 3rd, depending on players available. This is because the team with the 6th pick has no reduction in draft chances the following year, and would get reallocated ping pong balls in the following draft.