Ttyjfyk

There is an interesting new service that compares ROI among bakers, called Bakers Performance Index. It differs from other ways to check this (e.g. see this post), in that they delegate 10 XTZ to a list of bakers for evaluation, and report actual rewards received. Since there is no reason why bakers should pay some tezos owners more than others, this seems to be a very accurate test for effective ROI.

What struck me though is the low ROI. For some, it's actually zero. Consider for instance the first set of columns, which seem to show an average of payment for cycles 34 to 88 (site has no documentation, so I'm just guessing).

The highest ROI is just under 3%, and that baker has an expected ROI of 7.10%. That seems like a massive difference. Same applies to the rest of bakers. What's going on? Why such difference? I am missing something? Or is this just lack of accountability, enabling bakers to underpay?

I am Axel, the developer behind bakersperformanceindex.com and I hope this tool can help you in reviewing the baking services you're using.

The basic idea behind it was:

1. delegate the same amount of xtz to baking services, in our case 10xtz


2. delegate at the very same cycle, in our case at cycle 34


3. collect and display data in a simple and raw manner

So I delegated 10xtz at cycle 34 to 60 bakers.

There are basically 3 major columns:

• cycle 34-->n-2


• cycle n-1


• cycle n

Each column has 3 sub-columns

• Rewards


• ROI


• Balance

Rewards/ROI/Balance are headers for each column.

Let's take an example: Hayek Lab for the column cycle 34 --> cycle 88 (at this time 27th march 2019, cycle 90 in Tezos Time)

• Rewards: 0.268


• ROI: 2.681%


• Balance 10.268

So between cycle 34 & 88, Hayek Lab had distributed 0.268XTZ of rewards for 10XTZ delegated, which makes a ROI of 2.68% for this period of 54 cycles.

If we annualize with the following figures:

• Number of cycles in a year: 365/3 = 121.66


• 121.66/54 = 2.25 (we've baked for 54 cycles in a 121.66 cycles)


• Annualized return for Hayek Lab: 2.25*2.68%=6.03%

I think numbers are consistent

If we take the figures for cycle 89, we have:

• Rewards 0.0058


• ROI: 0.056% (we have a ROI of 0.056% for one cycle)


• Balance:10.274

This is the balance we have after baking during 55 cycles. The annualized ROI with the ROI at cycle 89: 0.0058*121.66= 6.77%

Figures are once again consistent.

If I do not annualize ROI is because, imagine the baker you choose, for one reason or another, does not distribute the rewards, your annualized ROI is not the same.

I just want to rely on proven figures, all the data displayed is the data we have right now and not plans on the future.

After one year or 121 cycles or at cycle 155, we will have one year data so the ROI will be quite exact.

Few more comments:

1. 
   

   If some bakers display 0, is just we did not get any reward from them for various reasons:

   
   
   
   • 10xtz < Minimun Delegated Amount accepted
   
   
   • Anonymous delegation not authorized by baker
   
   
   • baker is a scam, node not working...
   
   



2. I wish I could delegate 1000 xtz to all baking services at cycle 100. I am trying to convince bakers to get on board.




3. Many new features are coming, but their development takes time & resources.




4. You can get in touch with me at bakersperformanceindex@protonmail.com for sharing ideas.

I believe the period for which the data was collected and presented on this page is not sufficiently long to be representative.

The page states that data on returns was measured between cycle 34 and 88, which is 54 cycles, roughly 160 days or 5.25 months.

In your example, the ROI of 7.1% as estimated by the baker is stated as annual returns, i.e. 12 months, and the ROI on the page thus represents only 44% of this timeframe. Therefore, it seems that after 12 months, the baker will likely reach the estimated ROI. Nothing suspicious here.

As for those bakers with a 0% ROI, I assume that they have not yet been included in this project and data is just missing, though I have not checked that.