Again, I'm a total newbie here just trying to reconcile different data sources before I begin to form any conclusions!

Well, the chart of the values of your gtr1 link looks basically the same as your P123 chart.

18.95% is essentially the same as 20.475%, within the margin of error. That could easily be explained by minor differences in the assumed prices of the trades.

So I would say that probably gtr1 backtest is confirmed by this P123 backtest.

Assuming so, then since QQQ inception (3/10/1999),
QQQ: 10.08%
top10: 18.63%
top100: 9.14% (equal weight QQQ)

-------- later -------------
Okay, I grabbed daily QQQ quotes from yahoo and made a chart showing QQQ and the gtr1 top10 daily prices, normalized for the same initial value.

With log scale, the slope of the top10 was steeped than QQQ, except for the last few years. Since about mid 2017 the slopes (i.e., returns) are about the same.

The charts: https://ibb.co/4GcQn1V
https://ibb.co/VWYKPPHb

As many people have said, there are periods where every good strategy doesn't work well. For this one, it isn't actually bad, it just doesn't outperform. Although with much greater volatility and drawdowns.