You are suffering from data analysis paralysis. don't analyze trivial things. Focus on the important things. day volume of 2.2m is a lot. It is quite usual for the volume to be < 1 million / day.
Interesting question, but ultimately moot. And Yes it is too small because this is a flawed analysis. The premises are suspect. Who says all algos are not retail? Who says 90% is the right number? Seriously when any one of the premises is so suspect, you just stop trying to draw a conclusion and go back the premises. Lookup validity versus soundness. A better, bar napkin, analysis would be to look at the number of executed trades that are under 20. Set your time of sales to exclude blocks over 20. Within that group: under 5 contract, 90% of the trades are probably retail. Most of the 1,2, and 3 are retail. between 5-20 probably 75% are not part of a multiple order that is broken up. Add those two trade totals up for an estimate of retail trades. How many are retail traders? Most retail traders are commission generators and make more than 5 round trips per day. Some MANY more. So divide by 5 and you have an high end ESTIMATE. divide by 15 would give a low end estimate. PS: Why 20? Because professionals are wary of moving the market and clearing the order book. Over 20 is when that starts to happen based on watching the market and checking the volume stats by say 2 second intervals.