Let us assume that if either of the two players improves, he bets, and the other player folds.
We can now figure out exactly how far off the numbers would be if we calculated equity based on the cards we saw.
35.72% of the time 8 7 improves and bets.
In this case the other calculation method gives us 61.36% equity instead of 33.33% equity.
30.95% of the time J T improves and bets.
In this case the other calculation method gives us 68.18% equity.
The other 33.33% of the time our real 33.33% equity is used.
Averaging these numbers together, the other calculation method would give you 54.13% equity in the long run. This is almost twice as much equity as you are due (33.33%).
We can only calculate this though because we know all of the cards. In a real world example, we do not know all of the cards, and so have no way of knowing how far off the numbers will be.
When PokerTracker does not know all cards that were in the hand at the time of the all-in, it will not calculate EV for that hand, precisely to avoid the inaccuracies mentioned above. When PokerTracker calculates your equity, it is 100% accurate. We do not accept being accurate 97.5% of the time and inaccurate 2.5%.