From the comment thread of this post, reader rg suggests one way to settle this theoretical contest.
I think the better metric is how much a leader improved the position of their party, relative to before their tenure as leader. That kind of metric could capture Layton’s 2011 gains. Harper is less impressive on that metric – in 2000 37.7% of Canadians voted for one of the constituent parties of the CPC. Harper fell below that mark in 2004 and 2006, but edged it slightly in 2008 and 2011. Of course Harper looks better if you take the poll position of the Alliance and PC’s at the point when he became leader.
Lets contrast that to Charest. The swing to Charest in 1997 was a modest 2.8% from historic lows. In 1998 he actually won a lower share of the vote than Daniel Johnson, and even in his 2003 victory he edged Johnson’s 1994 result by less than 2%. Both his 2007 and 2008 victories were well below that bar. Part of the reason for this is the success of the ADQ through those years, but it is less clear to me that Charest is necessarily a great politician.
I think there is a better case for Danny Williams. The man led a party that was not traditionally Newfoundland’s party of government. As of 2003, Tories had been premier only 17 of the 54 years since confederation. Williams won in 2003 with a 17 point swing, and then gained another 11 points in the next election. Look at Dunderdale’s standing in the polls, despite a prosperous Newfoundland economy, as a strong indicator of the personal political abilities of Williams.