Analyzing master games for training

Having wrapped up the last set of my own tournament game analysis with Annotated Game #165, I'll soon be switching to another series of master-level games for my weekly analysis training.  Alternating analyzing your own games and examining relevant master games I feel has been a helpful practice that has added depth and balance to my training program.  Annotating contemporary master-level play I've found to add dimension to a training program in several ways:

• The overall level of play is greater, but there are still key turning points in each game that can be identified for "lessons learned", including tactical oversights and game-changing strategic decisions (good or bad). These turning points are usually more worthy of individual study than those in amateur games, since at the Class level evaluations can often fluctuate throughout the game.
• Seeing how even top-level masters can overlook tactics (and analyzing why) offers a psychological boost for amateur players. Often we improving players despair of never achieving perfect play; there is no such thing, however, so it's best to aspire to play well, rather than to hope to never make a mistake - either by blundering, or not seeing positive opportunities on the board.
• Finding recent "model games" in your opening repertoire can provide great insight into both opening schemes and successful middlegame planning. One of my consistent weaknesses has been the middlegame transition; often I know I have a good position out of the opening, but finding a concrete plan to further improve it is difficult, in the absence of any obvious weaknesses in my opponent's camp. I have established a separate "Model Games" database for these types of games and can also review the database of annotated master games ("commentary games") from this blog (download link in the sidebar).

I've been rather selective regarding the master games I choose, partly because there's no point in accumulating a large backlog of games which I'll never actually get around to analyzing. Other important aspects involved in selecting games are the relevance of the game and how understandable it is; often these elements are closely related.

• Relevance: I don't limit myself to analyzing master games that fall exactly within my opening repertoire, but I do want the games to provide concrete insights related to my knowledge base and play that needs improving. Usually that means having structures and positional themes that I understand reasonably well. Sometimes it may be a particular tactic or strategic theme that catches my interest when looking at the game for the first time.
• Understandable: basically this means that in a roughly two-hour period, with an engine for assistance in evaluating positions and the tactics available, I should be able to understand the game's overall trajectory, including why the players made particular key moves (most moves, in fact). Naturally I do much better in understanding the opening and middlegame phases that are derived from my own opening repertoire and tournament experience. I'll only consciously avoid selecting very technical or specialized games such as Sicilian Dragons or Berlin Defenses, which require a lot of depth to understand many move choices. (Not that I won't go over such games on news sites etc., but I won't select them for master game analysis purposes.)

In practice, I find that I can get a lot of mileage out of games up to around the 2500-2600 level that fall within my general opening knowledge.  This means, for example, I really enjoy looking at the U.S. Chess Championship each year (both open and women's sections).  On the flip side, it's rare that I would select a current World Championship game or the like at the 2700+ level, since that's too bleeding edge for me.  Looking at the current Carlsen-Karjakin match, though, I'm comforted by the fact that many commentators and sometimes the players themselves are also having a hard time understanding the games.