Looking at this example, {F3905194}, either algorithms are not implemented as intended, or algorithms are giving un-useful results and should be tuned.
1) Nominal backlog growth and pessimistic backlog growth are identical, and should not be.
2) Velocity forecast is zero; optimistic and nominal should probably both be greater than zero.
More generally:
[ ] Backlog growth and velocity growth algorithms look reasonable in all of these test cases
[ ] Continuous project: http://phlogiston-dev.wmflabs.org/ve_tranche13_burnup_points.png
[ ] Finite project, status unclear: http://phlogiston-dev.wmflabs.org/and_tranche3_burnup_count.png
[ ] Finite project, almost done: http://phlogiston-dev.wmflabs.org/cot_tranche16_burnup_count.png
[ ] Finite project, almost done, high volume: http://phlogiston-dev.wmflabs.org/ios_tranche1_burnup_count.png
[ ] Finite project, almost done, low volume: http://phlogiston-dev.wmflabs.org/ios.html#tab3
[ ] Finite project, really done, low volume: http://phlogiston-dev.wmflabs.org/ios_tranche16_burnup_count.png