You are never really up against one hand. You are up against every hand your opponent would play exactly the way they just did. Weak players guess — "he has the flush" — and then marry the guess. Strong players hold the whole range in mind and ask one question: how often does that range beat me? But before you can read a range, you have to read the board cold. So hand reading is really two gears, and they stack.

Gear one: read the board

Given a five-card board and any two hole cards, you should see the best five-card hand instantly — and know which holdings beat which. This is the unglamorous half, and it is where most money leaks.

A pair is beaten by two pair is beaten by trips; ties break on the kicker, then the next card down.
When the board pairs, full houses come alive — and your trips or two pair may now lose to a boat.
A pair on the board can counterfeit your two pair: if the board is K-K-9-9-2 and you hold A-7, you play the board's two pair with an ace kicker, and so does half the table.
Sometimes the board plays itself — a straight or flush sitting in the five community cards that no two cards can beat — and the pot is chopped.

Read this wrong and everything downstream is wrong. You cannot weigh a range you cannot evaluate.

Gear two: read the range

You rarely see their cards, so you swap "their hand" for "their range": the set of holdings consistent with everything they have done so far. The trick is that a range only ever shrinks. Each action crosses hands off.

Preflop sets the starting range — a tight early-position raiser holds a very different set from a button who min-raises everything.
Each later street prunes it. A hand that would not bet, call, or raise the way they did is gone from the range.
What survives is what you are actually up against. Your equity is measured against that whole set, weighted by how often each holding occurs — not against the one scary hand your gut volunteered.

Bet sizing is the strongest filter. Big bets tend to be polarised — strong value or a bluff, rarely the medium hands, which prefer to keep the pot small.

Worked example

A tight player opens from early position: roughly the big pairs, A-K, A-Q, a few suited broadways.

The flop is A♠9♦4♣ and they bet. Now prune: they keep betting aces, sets, and A-K / A-Q; the suited broadways that missed give up. The turn is 4♠ — the board has paired. Full houses are suddenly live for A-A, 9-9, 4-4, and A-4. They barrel again.

The river is a brick and they shove all in. Step back and name the range, not the hand. From a tight player, this line is almost all boats and quads — A-A, 9-9, 4-4, A-4 — plus the rare missed bluff, of which a dry board like this offers few.

You hold A♥J♣: top pair, good kicker. Against that range you beat essentially none of the value and only the odd bluff. You are not folding to "the boat." You are folding to a range that is mostly boats. Fold.

Where people go wrong

Anchoring on one hand. You decide "he has it," stop updating, and play the rest of the hand to a story you wrote on the flop. Hold the range; let each action prune it.
Forgetting the bluffs and the bricked draws. A range is not only value. Weight the hands that took this exact line and missed — sometimes they are a big share of it, sometimes (as above) almost none.
Misreading the board. The cheapest way to misread a range is to misread the board: missing that it paired (boats live), that three to a flush are out, that your two pair got counterfeited. Gear one first, always.
Levelling yourself. Do not invent an exotic holding to justify the call you already wanted to make. Weight hands by how often they really occur.

A rule of thumb that travels: name the range before you name the hand, then ask what fraction of it you beat.

How the hand-reading puzzle trains this

The hand-reading puzzle shows you two complete hands and the full five-card board, and asks which one wins — or whether it chops. That is gear one in its purest form: read the board fast and exactly. Who actually made the best five? Who holds the better kicker? Did the board pair and counterfeit someone, or does it play itself? The closer the two hands rank, the harder the puzzle and the more your rating rides on the fine distinctions. Range reading is built directly on this muscle — once you can evaluate any holding on any board without thinking, the live skill of narrowing a betting line into a range has somewhere solid to stand.

Go deeper

Reading a range is only useful next to your equity against it — the two lessons are halves of the same decision. And when the river is shoved in your face, the question is always the same: of all the hands that play this way, how many do I beat?