less votes than normal: it is because more priority is given to higher steem power now. I don't mean to be rude but if you didn't cash out 90% to bittrex it might be better for the pool, so I would say this is not actually a problem in terms of how the system is operating. Trust me that I understand the tactics of moving SBD into BTC right now and buying steem later to multiply it, so if that is what you are doing then great, but an algorithm has no way to differentiate between intentions of behaviors from individual people and must be structured in a way that benefits the most people on average and in many cases where critical abuse scenarios are possible a system must assume the worst of people by default to protect the others involved, and if someone can drain the pool into bittrex using a bunch of tiny accounts and never power up we are in a very bad position for long term growth (I am not saying that is what you are doing, but that the system itself should assume that you are by way of factoring in SP and the rate of your SP increasing over time).

If anyone is able to collect a large chunk of the rewards but never increases their SP we are not accelerating properly. You have powered up about 16SP in all of available history outside of post amounts, but have sent ~115SBD to bittrex, and I don't judge you for that but from a system standpoint it can and should look like someone draining the pool because there is no way for the system to assess you as a person or determine if what you are doing is legitimate or not, it just has a bunch of numbers to work with unless things start to get really advanced. So just imagine if all the high-rank 15SP accounts wake up and replicate the pattern, they could farm our pool and send it to bittrex and we will all be voting into dumping the price of SBD at their profit for people that have their rank simply from old idle accounts that have no voting power to offer anything in return, and instead of increasing their 15SP they just cash out, so the system must learn to understand SP and the value it holds. Again, I don't mean it personally but it just so happens you are in a position that facilitates my expression of this serious concern and one of the primary reasons I push for SP-weight (among so many reasons, being one of the primary is significant) People attempting to drain the pool in one way or another will always be the frontmost issue we find ourselves dealing with, but from a system standpoint one of the best ways to prevent this kind of abuse is to place more value onto accounts having higher SP, because those are the kind of people most likely to keep powering up and are already offering the most in return.

I think something your issue highlights is the gradient decline of number of votes doesn't extend beyond the FVL voters range, but rather singular votes are concentrated and have some cutoff range, so you get a black and white difference between a small range of rank at that specific threshold going from 500 votes immediately down to 50, but those 50 do come from the highest SP holders so you are still getting quite a lot in return compared to most people in the pool, just looking at your caster page here your posts get about half of their total value from the shadow pool, somewhere in the range of 2.5 added to payout, and using this tool to calculate you are giving about 0.07 to the pool from a vote. One problem here is that posting frequency is not yet factored into the equation other than limitations for anti-spam, so the opposite is not part of the weight you are given and an infrequent posting time will not earn any extra to make up for time idle. I am not sure it should considering the dead accounts that have high rank just from their age, but assuming you posted once per day you would currently be in the range of gaining 350% what upvoting yourself would gain and that is assuming you allow shadowbot 100% voting, so at your current status it is actually something in the range of 700%, which is far beyond any other offering on steemit, so despite being less than the peak from a month ago it is still quite extreme, especially considering you still have the option to move up even more. At one point your return was much higher than this already extreme ratio, and the change to factoring in SP has reduced it, but at another point you had way less than you do now so from that perspective you are still way ahead. I think the next step is POWER UP! Everyone in the pool should be encouraged to power up really, no matter what rank they have. All it will take is a couple people from those old micro accounts to start draining the pool to show the current weighting of SP is actually very lenient and still in favor of smaller accounts and account age.

Anyway, as a game designer your kind of problem is the kind I prefer to solve even if I don't agree it actually is one. I contemplate the issue and it seems to mostly derive from the cliff of queue depths where non FVL voters never reach down beyond a certain range except when those higher rank than that range are inactive. This means you could be getting less votes simply from others jumping ahead of you in rank, or you could be getting less votes from inactive accounts waking up and posting, but in either case the number of votes decreases by a large margin as soon as you fall outside the range of non-FVL voting.

I think to resolve this problem shadowbot will need some gradient decline of the smaller votes so that people can notice their votes increasing over time rather than in big chunks when they hit a specific range. To do this there could be some kind of overflow mechanic where a post reaches a certain number of votes and then starts diverting some of them to the rank below, and the overflow amount could be based on rank as well. Supposing the threshold is a static amount it would automatically scale alongside the total size of the pool. Ex: 1000vote overflow threshold would behave differently now than it would once there are 2000 voters. This might be tricky to avoid problems and open the door to some abuse, but mitigated by priority being isolated away from rank according to SP. I am sure there are other solutions, but that is my random idea for it.