It is not straightforward to define a metric like this. What counts as an "English word" ? There are books full of the scientific names of plant and animal species, which usually come from Latin; do these count as "English words" ? What's the cutoff?

IMO, a much better metric is frequency-weighted; that is, taking some corpus of real English and counting the words in it, rather than weighting "every English word" with the value 1.

If you do this frequency-weighted analysis, Old English is far ahead of French and Latin combined (especially in colloquial speech; they're closer in formal writing).