So, how did I know that Balch was a bad player looking at his dice stats? Well, the key is in the bottom half of that image--the Battle Outcomes stats. (Sidenote: the Dice Outcomes stats are useless. You should really never be looking at them. It doesn't matter what you roll, it matters whether that is higher than what the opponent rolls.)
First, a little refresher on how the dice work in relation to these statistics. a 3v2 roll is going to be the most common. That's when you have a 4-stack or bigger and you're rolling against a 2-stack or bigger. You know that every time one of these types of rolls happens, 2 troops are going to die. You don't know whether they will be your opponent's or yours, but you do know that 2 troops will be taken off the board. So we can figure out how many total rolls a player has made by totaling up his kills and losses and dividing it by 2. The same can be done for the rest, but if any player is only rolling 1, then the division is unnecessary.
Now, you don't have any control over what kinds of decisions that your opponent is making, so let's forget about the second row of the second table, which is the defensive stats. And let's not worry about the total either. Only look at the first row of the second table: the Assault row. I'm sure a lot of you figured out instantly where I was going with this when I posted the table. This guy is making a seemingly large number of high-risk attacks, by which I mean attacks where he has less than a 4-stack and therefore less than 3 dice to roll. The numbers seemed very high to me, so that's when I made my comment in the forum.
But then I got to thinking, what is a "good" distribution of attacks? So, I took a somewhat random sample of the following players (their scoreboard ranks are in parentheses):
Kaskavel (1); 100mates (2); Chariot of Fire (20); Jippd (26); Pirlo (190); Agentcom (191)
I then compared the results to balch who was the complainer in previous post.
I've presented the results in graphic form below. There are some interesting differences that get obscured by the scale, but you can get the big picture by looking at it.
Notice that there are some differences among the good players. CoF apparently really likes the 3v2 roll (more on this below). Jippd is the opposite and seems to seek out the 3v1 roll (which, remember, is rolling a stack of at least 4 against a single). But the biggest relative difference by far is in balch's chart. The disadvantageous rolls that barely even show up in the charts of the other players are clearly being depended on by this player. He is rolling less than a 4-stack a whopping 18.6% of the time that he clicks the "assault" button. The average for the other 7 players, by comparison, is between 5% and 6%. And you can see that it's hurting his game because he ends up with relatively fewer opportunities to roll 3 dice against the defender.
I mentioned that I'd come back to CoF and his high percentage of 3v2 rolls. I think this is reflective of lots of large-map, large-team, no spoils games. Instead of trying to take territs (like rolling 3 dice against 1 die in order to card), he is oftentimes looking to inflict the most damage. That is often done by creating lots of 4-stacks and trimming everything that you can rather than taking any territs. The other players' game choices also probably affect their stats.
I may come back to this at some point and look at a bigger sample or a sample more appropriate to players of team games, but I just thought I'd mention it here. And perhaps some of you may want to look at your dice stats. If you're making disadvantageous rolls more than 6 or 7% of the time, you are outside this range and might want to re-evaluate your strategy. And out of that amount, over half of those disadvantageous attempts should be 2 dice versus 1 die (i.e. an attack from a territ with 3 troops to a territ with 1 troop).