Trying to understand pool performance

The testnet is running now for some days and I thought, the list of pools would become more stable with time but this seems not to be the case. I know that there were issues with forked chains and not everything is running as it should be. That’s fine, this is what the testnet is for. But I would like to get a better understanding of what happens here.

Take the pool performance: At the rewards calculator you can read: “… stake pool performance, which is the number of blocks a stake pool produces in a given epoch versus the number it was expected to produce.” Given that I would expect that most of the pools have a performance of 100% and only the ones with technical difficulties, internet connetion issues or similar problems have lower values.
When I look at the ranking of https://pooltool.io/pools, there is at the moment one single pool which reaches 100% (or 1.0 ).


Within the best 20 out of 500 pools most are in between 0.1 and 0.2 With such low values the number of blocks could not have been written. At the moment, there are between 30% and 50% of the blocks written.
.

Back to the pool list: If there was no incentive for a certain pool size, I would expect that the percentage of produced blocks by a pool runs linear with the percentage of staked ADAs at this pool. A pool with 1% of staked ADAs would produce around 1% of the blocks, a pool with 10% stake would produce around 10% of the blocks.

In the upper part (the pools with the highest pool performance) there are a few very small pools which have by sheer luck catched a block and produced it. And not even they have all a performance of 1.0

Then there are the huge pools of IOHK. Here “luck” is no more relevant. These pools are so huge, they should be well above the saturation limit. As I read in the explanation: “To help prevent any single stake pool from growing too large, the Incentivized Testnet has a mechanism to ensure pools saturate at a certain percentage. If any stake pool exceeds this percentage (of total stake in the network), then the rewards for that stake pool will no longer increase but will continue to be shared among delegators, making the pool less desirable. This saturation point is configurable and may be increased as the number of healthy pools grows.”

In my understanding up to now, they would get less blocks per stake which makes them less attractive. When you look at the pools IOHK2 and IOHK4 then you see that with around 4% stake they create around 6-7% of the blocks so this seems to be not the case. These two huge pools seem to be very attractive instead. So maybe not the probability to catch a block is reduced at the saturation point but the reward per block? Unfortunately here I can’t see the actual rewards, only the liftime rewards which have no meaning here.

Does anyone know axactly how the performance is calculated? As it is now, it doesn’t seem to be very helpful as it differs greatly from “Epoch blocks %” divided by “Live stake %”. Any good ideas?

1 Like

It seems to me (I dont have the exact formula but its my experience) that performance metric is a short term performance metric and missing a block penalize you quite a lot compared with overall epoch performance. I think also users should ask themselves how many blocks did a pool produce compare with its stake. If you have 1% of the overall stake and there are 4320 up for grab a good target is 43 blocks that period and thats your statistical average in the lottery. Somewhere around that number with more margin of error to smaller sample size pools (those producing less blocks) could in my opinion be valuable for delegators to consider.

1 Like

My impression is that the performance figure is pretty meaningless at the moment due to the network problems, pools are missing blocks through no fault of their own.

2 Likes