Innings, Balls Faced, Balls Bowled

Volume/Count in the available dataset used to generate the numbers. This may not match the official records perfectly and is meant to provide an indication of how much data was used to arrive at the provided numbers.

Evidence Weighted Stats

When we see more deliveries for delivery level indicators and more innings for innings level indicators, our confidence in the aggregate value of those deliveries is higher. A batter who scores 1300 runs in 1000 deliveries can be projected to bat at a 130 SR more confidently than a batter who scores 13 off 10.
Evidence weighting is a process by which we expect a player who has never played, to be on par with an average player, and pull their value away from this average player gradually as we see more and more evidence of the player's performance differing from that of the average player in a particular direction.

(R/W)PO

Runs or Wickets per Over. The per over normalized rate is used for other metrics like dots and boundaries as well, but it is explicitly called out in those cases.

Addl. <Metric>

This the value of the metric above expectation. The expectation is based on a model that accounts for a variety of factors as described here, or similar. This expectation represents what the outcome of the delivery in terms of runs scored, boundaries, dots and wickets would have been for an average batter and bowler in the same match situation at the same venue and competition. The difference between this expectation and the actual outcome gives the metric.
Note, that the direction of this metric is based on (Actual minus Expectation) regardless of the metric (RPO, WPO, Dots, Boundaries) or who it applies to (batter/bowler) i.e. higher Addl. RPO is better for the batter and worse for the bowler and so on.
I realize this can be a bit confusing, and will likely be fixed in the future at which time I will update this description.

<Metric> Percentile

The proportion of players that have a metric value below this player. See note in the Addl. <Metric> section for the interpretation of direction of goodness e.g. a 5% on Addl. RPO Percentile for a batter implies that only 5% batters do worse, while a 5% on Addl. RPO Percentile for a bowler implies that 95% of bowlers do worse. So, for RPO, a higher percentile is better for batters and worse for bowlers.

WPA

Win Probability Added. A win probability model computes the win probability at the end of every delivery. The win probability added for each delivery is the delta between the win probability at the end of the previous delivery and the end of the current one.

Addl. WPA Proportion Leveraged

For each delivery, there is a range of possible WPA values based on possible outcomes. This is generally bounded by wkt/dot on the lower end and a 6 on the higher end. The expected WPA; the WPA for the expected outcome, is within this range. This gives the maximum and minimum Addl. WPA opportunity available on either side of the expectation. The actual outcome generates a WPA that is on one of side (higher or lower) of the expectation. This metric gives the proportion of available WPA realized by the actual outcome on the side of the expectation that it is on. So, from the batter's perspective, if the outcome is higher than the expectation, this number is positive and represents the actual WPA over expectation as a propotion of the maximum possible WPA over expectation. If the outcome is lower than expectation, this number is -ve and represnts the actual WPA under expectation as a proportion the of the maximum possible WPA under expectation.

Most Freq. Pos. (Batting)

The batting position at which the batter has batted most often.

Avg. Pos. (Batting)

The avg batting position at which the batter has batted weighted by number of appearances. A fractional value can indicate the proportion of times they have batted in adjacent positions if those are the only positions they have batted in.

Util. Rate

Proportion of balls faced/bowled with respect to total balls faced/bowled by the team.

Contrib. Rate

Contribution rate. For batting, this is proportion of runs scored with respect to total runs scored by the team. For bowling, this has two flavors - Proportion of runs conceded & proportion of wickets taken with respect to total runs conceded and wickets taken by the team.
Note, that this metric is directionally opposite for the runs conceded flavor. Higher contribution rate for runs given is worse while higher contribution rate for wickets taken and batter runs scored is better.

Avg Dbl Digit Streak (Batting)

Average number of consecutive innings with >= 10 runs scored.

Dbl Digit Inn Proportion

Propotion of innings with >= 10 runs scored.

Avg Failure Streak

Average number of consecutive innings with <= 3 runs scored.

Avg. (Batting)

The batting average - runs per dismissal.

Survival Factor

Unlike most other metrics here, which are defined per delivery, this metric is defined per inning. This is the probability that an average batter facing deliveries in the same situation, venue and competition would have been dismissed by the time the batter in question played their Last delivery (regardless of whether they were dismissed or not). This is a measure of innings length. A higher value indicates the propensity to play longer innings.

Phases

PP: Overs 1-6
Middle: Overs 7-15
Death: Overs 16-20

Against Partnerships

Bowling performance when the current pair of batters have contributed more than 25 runs to the team total in the present innings.