There are various sources of data that can be used with statistical approaches and some reasonable assumptions to get to some acceptable estimates. That would be the start over time the approach and data can be polished and improved.
Example data point:
- Basic onchain data, active addresses, Algo balance, other token balances that can indicate the same or different individuals with a good certainty. Different tokens provide different information.
- CEX hot wallets and transfers to and from them. Exchanges have a separate address per account for deposits and most of them don’t re-use them for any other purpose later.
- For CEXes, I think reaching out to them and seeing if any of them is open to provide any data (e.g., the number of algo holders on their exchange) might work too. Having one CEX data and combining that with other data (algo balances, transfers, etc) can be used as a basis for estimating other CEXes individual accounts.
- Talking about CEXes, a basic useful metric, independent of number of Algo holders, is to know the number of Algos each SEX holds.
- Governance participants in the last few years
- AF has some information about the initial transfers/sales (with KYC) that could be useful for analyzing the later distribution.
I think it is also good to share the methodology with the Algorand community on reddit and ask for feedback. I think you’ll hear many other good ideas/feedback that can help improve the approach.
Please note that this metric is very important and needs to be tracked so we need to start from somewhere. I have been in communication with the CFO in order to include this data in the future transparency reports. I have also exchnged a number of messages with the CMO (Marc) and he plans to use this data as well. You may want to talk to both of them.