You're looking at a super small sample size of 150 games here, which is not conclusive data at all. The data from these small numbers of games swings wildly. For example 3v3 top 200 with 277 games from 16-06 to 18-07 shows the complete opposite with Allies on 60-64% wr. The data only starts being reliable at sample sizes of like 3000-5000 games minimum.
No, it was 568 games for that time period. Also, if you want a good measure of the variability of data, you can take a bunch of individual samples and get an estimate of how many you would need for statistical significance. It isn't just some set number like 3000. I did a small portion of that, just as a reality check. I went with 1-week samples since that would also show trends over time. Here are the weekly results (except 13 July which is only 5 days):
Games OST Win % OKW Win %
15-Jun 524 51.8 59.8
22-Jun 420 58 58.2
29-Jun 384 60.1 70.8
6-Jul 288 60.8 61.4
13-Jul 172 64.7 69.2
Stdev 4.73 5.72
There really aren't any wild swings. The average is more than two standard deviations away from 50%. As time has passed, the win percentages for both OST and OKW are on a slight upward trend. The number of games has been decreasing, but I don't know if that has as much to do with win rate as it does people trying a new patch and then going back to whatever else they were playing.
In general, I think the balance team has done good work for 1v1 through 3v3. There's just some things that scale really differently into larger games that are causing problems. Those problems aren't getting better over time.
(PS - sorry about formatting, it always seems to compress and make a table difficult to read)