Why tracking goes wrong
Keeping a record of your games sounds like the objective part of getting better. In practice, the log is written by a person who remembers some games more vividly than others and who often wants a particular answer. Most problems with personal data are not arithmetic errors. They come from what gets recorded, what gets skipped, and when the analysis happens.
Logging only some games
The most common error is selective recording. People tend to log a session when it feels notable: an unusually good run, a frustrating loss, or a new strategy being tried. Quiet, average games go unrecorded. The log then looks more extreme than reality in whichever direction you tend to remember.
The fix is a rule fixed in advance, such as “every ranked game, logged right after it ends.” A rule that does not depend on how the game went is what keeps the sample representative.
Deleting or excluding “unusual” games
It is tempting to drop a game because the opponent quit, the connection was bad, or you were distracted. Some exclusions are legitimate, but a flexible rule becomes a quiet way to remove losses. If you plan to exclude certain games, write down the criteria before you look at results and apply them evenly to wins and losses alike. When in doubt, keep the game and add a note.
Changing categories midway
Suppose you track win rate by deck, then decide halfway that two similar decks should be merged, or you redefine what counts as an “aggressive” start. Your earlier and later entries now measure different things. Keep category definitions stable, or record the date of any change so the data can be split at that point.
Stopping at a flattering moment
If you check your results constantly and stop tracking when the numbers look good, the stopping point is chosen by luck. Random results wander above and below their long-run level, so any record can be caught at a high point. Decide the sample size or end date beforehand, such as “the next 200 games,” and look at the result then.
Judging too early
Even a clean log needs enough games to say anything. Small samples swing widely, which is why early conclusions about player statistics are so often reversed by later ones. A 7 and 3 start tells you very little, and the size of the uncertainty is the subject of confidence intervals. A good habit is to write down how many games you need before you will draw a conclusion, and stick to it.
Comparing unlike conditions
Wins against beginners and wins against strong players are not the same quantity. Neither are results from a casual evening and results from a high-stakes tournament. If your opponents, formats or stakes vary, either track them as separate columns or keep the results in separate groups. Otherwise an apparent improvement may just be an easier schedule.
Hindsight and memory
Writing results from memory at the end of a week introduces gaps and flattering rounding. The same goes for recording why you lost: after the fact, people often assign bad luck to losses and skill to wins. This pattern is a familiar example of confirmation bias, because the tidy story fits what you hoped to see. Log the result immediately and record the reason in a separate note, ideally before you know how the game ends if you are tracking a decision.
Mixing up luck and skill
A win does not prove a good decision and a loss does not prove a bad one. Tracking only outcomes makes a risky play that happened to work look smart. If you want to learn from your records, log the decisions and the circumstances, then review them against what the odds said at the time. Separating the quality of a choice from its result is the heart of how luck and skill are told apart.
Missing the denominator
“I won five games with this build” means little without the number of games you played with it. Always record both attempts and successes, and compute rates from them. The same goes for rare events: noticing three lucky streaks says nothing if you did not also count the many sessions with no streak.
Poor structure
- Free-text notes only. Without consistent fields, nothing can be counted later.
- No date or version. Patches and rule changes make older entries incomparable.
- Duplicates and typos. Occasional double entries or misspelled categories split one group into several.
- No backup. A spreadsheet that gets overwritten once can erase months of data.
A simple starting layout
A minimal sheet with one row per game is enough: date, opponent or opponent type, format, your choice (deck, character or strategy), result, and a notes column. Add columns only when you know how you will use them. Keep the rules for what counts as a game written at the top of the sheet.
None of this guarantees a clear conclusion. Some questions need more games than any one person plays, and an honest result of “not enough data to tell” is a perfectly good outcome. What good habits do give you is a record you can trust enough to know what it can and cannot say.