Reading the Board: A Mid-Game Evaluation Checklist

Most losses in dots and boxes don't come from a single bad move. They come from a player who never actually stopped to evaluate the board — who kept playing move by move, reacting to whatever was in front of them, until suddenly the endgame arrived and the position was already lost. By the time the chains are obvious, it's usually too late to do anything about their number or their length.

Strong players evaluate. Not once, but repeatedly, at specific checkpoints — usually every few moves once the board starts filling in — running through the same short mental checklist every time. It takes fifteen seconds once you know what you're looking for. This post is that checklist, in the order that actually matters.

Why "Just Play the Safe Move" Isn't a Plan

Early in a game, almost any move is safe — there are no three-sided boxes to avoid completing, so any line is fine. That safety is temporary, and it lulls a lot of players into a false rhythm: play a reasonable-looking line, see what the opponent does, repeat. This works fine for the first third of the game and fails badly in the middle third, because the mid-game is where the board's chain structure actually gets decided, even though no chains exist yet.

The mistake is treating the mid-game like an extension of the opening. It isn't. The opening is about not giving anything away. The mid-game is about actively shaping how many chains the board will eventually split into, and how long each one will be — because that split, more than any single tactical trick, determines who wins the endgame. If you don't evaluate deliberately during this phase, you're not making a plan. You're finding out what plan the board handed you, usually too late to change it.

The Checklist

Run through these five checks, in this order, every three or four moves once the board is roughly a third full. Each one takes seconds once it's a habit.

1. Count the Open Regions

Look at the board as a set of disconnected or loosely connected areas rather than as individual lines. How many separate regions currently exist, or are clearly forming? A region here means a connected area of boxes that will eventually become one chain, one loop, or one small cluster once fully divided by lines.

This is the single most load-bearing number in the entire evaluation, because the chain rule hinges directly on it: the player who wants to control the endgame needs the total number of long chains, plus double-crosses, to come out to the right parity. You cannot apply the chain rule if you haven't counted the regions it applies to.

2. Estimate Chain Length for Each Region

For each region you just counted, make a rough estimate: is this shaping up to be a short chain (1–2 boxes), a medium chain (3–5), or a long chain (6+)? You don't need precision here — a rough bucket is enough to reason about, and precision usually isn't even possible yet since the region isn't fully divided.

The reason this matters: long chains are worth fighting over, short chains mostly aren't. A 2-box chain barely moves the score either direction. An 8-box chain decides the game. If you find yourself investing careful thought into the shape of a 2-box region while ignoring an emerging 8-box one, your attention is misallocated. Re-read the endgame guide on loops and chains if any of this terminology is unfamiliar — it's the vocabulary the rest of this checklist assumes.

3. Check Parity Against the Chain Rule

Once you have a region count and a rough sense of which regions are long, check parity: do you currently want an even or odd number of long chains (plus loops counted appropriately), given the board's total box count and who moved first? This is the actual output of parity counting during live games, and it's worth running as its own explicit step rather than folding it silently into the region count, because parity is where players most often make silent arithmetic errors under pressure.

If the parity currently favors your opponent, this is the single most useful thing your evaluation can tell you, because it changes your entire plan for the rest of the mid-game: you now need to actively change the number of regions on the board — merging two developing regions into one, or splitting one into two — rather than just playing "reasonable" moves and hoping the split works out.

4. Inventory Remaining Safe Moves

Count how many genuinely safe moves are left — lines that don't create a three-sided box for either player. This number shrinks steadily through the mid-game and its exhaustion is what triggers the endgame. Two things matter about this count:

• Roughly how many safe moves are left, for each player, before someone is forced to open a chain. Whoever runs out of safe moves first is the one forced to open, and being forced to open first is usually the losing position.
• Whether any of the remaining "safe" moves are actually disguised sacrifices. A move can technically not create a three-sided box while still being a bad idea, because it commits a region to a shape that hurts you two moves later. Safe and good are not the same thing — see the marginal value of sacrifices for how to weigh this properly.

If you can't answer "who runs out of safe moves first" at any point past the opening, you're not evaluating the board — you're just reacting to it.

5. Identify Your One Actionable Lever

The first four checks are diagnostic. This last one converts diagnosis into a plan: given everything above, what is the single move available to you right now that most improves your position — either by nudging parity in your favor, by keeping a long chain from splitting into two shorter ones (or vice versa, if that split favors you), or by preserving a safe move you'll need later?

This doesn't mean calculating the whole game out. It means picking one concrete lever and pulling it, rather than playing a locally-reasonable move that ignores everything the first four checks just told you.

A Worked Example

Picture a mid-game board on a standard grid, roughly 40% of lines drawn. Running the checklist:

1. Regions: three areas are starting to separate — one in the top-left, one along the right edge, one in the bottom-center. Call it three regions, tentatively.
2. Chain length estimate: the top-left looks like it's shaping into a short 2-box cluster. The right edge is ambiguous — it could become one long 7-box chain or split into two medium ones depending on the next few moves. The bottom-center looks like a clean 5-box chain.
3. Parity check: with three regions on this board size and this player order, you calculate that you want an odd number of long chains. Currently, if the right edge splits into two, you'd have two long chains (the two halves plus the bottom-center) — even, which is bad for you. If it stays merged as one long chain, you'd have two long chains total (bottom-center plus the merged right edge) — still even. Wait, that means you actually need the top-left short region to somehow not count, or you need a fourth region to appear. This is exactly the kind of thing the checklist catches that gut instinct misses.
4. Safe moves remaining: you count roughly six genuinely safe lines left, three of which are in the ambiguous right-edge region — meaning you have direct influence over whether it merges or splits.
5. Lever: since parity currently favors your opponent, your best move is one of those three ambiguous lines, played specifically to keep the right edge merged into one chain rather than splitting it — turning your effective long-chain count into a shape you can work with, or forcing a fourth small region to appear elsewhere that flips the parity back.

None of this required looking more than a few moves ahead. It required systematically asking five questions instead of playing on instinct.

When to Run the Checklist

Not every move needs a full evaluation — that would be exhausting and unnecessary early on, when the board is too open for the numbers to mean much yet. The useful checkpoints are:

• The first time any region on the board becomes clearly separated from the rest.
• Every few moves after that, as more regions form.
• Immediately before you're forced to make a move that will likely open a chain or otherwise commit a region's shape.
• Any time your opponent's move surprises you — a good evaluation habit catches the moment the board's shape just shifted underneath your existing plan.

This is a habit worth building deliberately rather than hoping it emerges naturally; see solo training drills for structured practice, and consider running the checklist explicitly, out loud or on paper, for your first twenty or so games until it becomes automatic.

Summary

The mid-game is not a waiting room between the opening and the endgame — it's the phase where the endgame gets decided, and the players who win consistently are the ones who evaluate deliberately rather than react move by move.

Count the regions, estimate their lengths, check parity against the chain rule, inventory your safe moves, and pull one lever — do this every few moves, and the endgame will stop feeling like something that happens to you.