Once I hit on the idea of “Custom Model Games”, I knew I had a winner. Model games played by grandmasters are great for a lot of things, but they had some distinct disadvantages. By building model games myself with the assistance of Stockfish, I would get almost all of the benefits with relatively few of the problems. The plan quickly became this:

1. Identify an opening to learn and play.
2. Figure out how many games to create to “cover” this opening, based largely on the amount of time available to study and the importance of the opening in the overall repertoire.
3. Map out the variations these games should cover, based on likelihood they will come up.
4. Build out a game-length variation using Stockfish and my own feelings to include plans I liked and understood, lines that made sense to me, and moves that were inspiring. Spend time with Stockfish as necessary to understand why it preferred a move before making the decision to include it in the line or not.
5. Once the number of games have been completed, put them into a chapter in a personal Chessable course and work on memorizing them.
6. Use the opening in online games, identifying gaps and issues in intuition and understanding.
7. Add additional games to cover new lines or change existing games as necessary to teach important concepts and plans.

As a note, I decided that custom model games are better suited for memorization than they are for solitaire chess. Very simply, custom model games take more time to construct than a “found” model game and can be completely optimized to make them as useful as possible, so the in-depth ingestion method of memorization makes more sense than a more superficial type of exposure.

The biggest disadvantage of custom model games was that the player playing them, "me plus Stockfish", had some weaknesses. While I could guide the game into paths that I understood and patterns I could reproduce, the human part of that player was adding very little in the “quality of moves” department. If Stockfish didn’t really understand a position, then "me plus Stockfish" understood it still less. Probably the best choice for player would have been “me plus Karpov plus Stockfish”, but Karpov wasn’t around to help me. Having a strong coach might have helped cure that problem, but I also kind of liked that I was doing it myself. Stockfish was good enough that it would prevent me from making significant errors while I made the moves I found to be most natural, which was usually good enough for what I was doing. As an aside, because the long-term planning element was usually my job rather than Stockfish’s, sometimes it took several times of playing out various lines for me to get the idea of what was going on. Once I did though, I could intentionally select a line of play that showed the key ideas that I hadn’t understood at first, and after a few times running through the line and memorizing it the ideas were just second nature.

Of course, even if the long-term planning and positional understanding in these games was sometimes a little deficient, these games more than made up for it with amazing tactics. Honestly, the games often reminded me of the slashing games I grew up with that were played by Tal or Morphy or Kasparov, even though mine sometimes came from sedate openings. Because I was choosing the lines, I could pick ones that I liked, and apparently I liked sound positional sacrifices capped by shocking, white-hot tactics. Many of the games I put together still give me goosebumps, even though I’ve run through them many, many times. Training with them makes me remember why I play chess - and I love doing it even though it’s still work.

Now let's run through the basics - what you should prioritize when you are building a custom model game for yourself.