Don't get too focused on volatility. I am just a concerned by the teams with the lowest volatility (that may be gaming the system) as those with high volatility (that may be overcommitting/underestimating.) You may want to understand how the teams at each end are getting to their rating.
There are likely multiple factors applying to any one team's volatility. Volatility can come from a number of causes such as:
- Poorly defined or understood use cases. (Teams may have better business knowledge and get better clarification. Teams may be getting use cases for with an unclear solution or using new technologies.)
- Teams may be working on use case with differing levels of uncertainty. (Requirements without a certain solution can require experimentation to find a solution.)
- Uneven skill levels between or within teams.
- Teams may be assigned use cases for which the team does not already have the required skills. (This is good for team growth, but increases volatility.)
- Team location may invite more or less interruptions.
- Teams cohesion and effectiveness may differ.
- Teams may have different estimation skills.
- Teams may fill time working on technical debt and/or learning new skill. (This will reduce measured volatility.)
On average people in IT tend not to have the best interpersonal skills. I certainly have far less skill than my wife. Forming an effective team involves interacting on many levels. How well a team functions is likely to impact its volatility.
Try looking at your data:
- Sorted by velocity
- Sorted by volatility
- Calculate velocity/volatility
- Calculate volatility/velocity
It appears you may have two teams that are having difficulties. This could be for any number of reasons, some of which may be outside the team.
You may have four teams that are gaming your measure. They may be making good use of time made available when they overestimate a sprint.
You don't have trend data which may also provide good information. Generally you would expect velocity to increase and volatility to decrease (to a point). Changing assignments can break trends.