This guide assumes you are playing 4-player games with no top-secret token.
Setup: Choosing Combination for Opponent
1) Never give your opponent 3 cards of the same color.
2) Avoid giving your opponent 2 cards of the same color unless you have no other option.
3) Maximize the possibility of giving out false information with the two cards you keep in your hand (e.g., if you have the option of keeping 2 outdoor cards hidden or 1 outdoor and 1 indoor card, keep 1 outdoor and 1 indoor because now you give false information on an extra category compared to before).
First Turn of Game Tip
If you are the very first player to go in the game, always flip a secret informant card instead of asking a question because asking a question will give out an additional clue to an opponent.
1) It's rarely a good idea to ask an opponent a question when they don't have 0 magnifying glasses - avoid giving opponents extra turns unless you truly need that player's answer to a question!
2) Be careful not to ask repeat questions (same category + same witness) - I've seen even elite players make this mistake sometimes, always slow down and check the questions that have already been asked.
3) If you have a lot of question marks on your sheet and few X's, consider flipping a secret informant on your turn instead of asking another question.
4) Avoid flipping secret informants as soon as 2-3 or more of the possible cards on your sheet are X'd out. Flipping secret informants gets riskier the more X's you have on your sheet since it's more likely you will learn information you already know, effectively wasting your turn.
5) All else being equal, give preference to asking an opponent about a category that matches a card in front of them. For instance, if you are curious about red cards and Opponent C is the only one you can see that has a red card, asking them about red will help you without tipping off Opponent C that they also have a red card.
6) Always double check how many magnifying glasses you have at the start of your turn - even if you have more than one, you lose them all if you accuse during your turn. Ask questions with any extra before accusing!
1) If everyone is flipping secret informants, keep flipping secret informants too (asking questions instead will leave some players with more info than you since they'll have your question to reference as well as the secret informants they've looked at).
2) If people are asking questions right out of the gate, using secret informants becomes dependent on what you can gather from the clues that have been asked. If you have a lot of question marks on your sheet, consider flipping informants. If you have several X's on your sheet already, it's probably better to ask questions instead.
Have a Strategy When Formulating Your Questions
You Only Get 2 or 3 Clues: In the competitive 13 Clues scene, in most games you will have the opportunity to ask 2, maybe 3 clues before it is time to start trying accusations - it is important to strategically use the few clues that you have. There are two main strategies you should keep in mind when formulating a question:
Strategy #1 - Maximizing Clue Coverage: The general idea is to ask questions that cover several unknowns on your sheet, as opposed to questions that cover a single card. For example, if all 3 orange cards are unknowns on your sheet, asking an orange question could be a good idea, especially since the answer could be 0 (this happens surprisingly often for the color clues). Asking horizontal categories (sex/place/weapon) can also cover more ground, so don't neglect them.
Strategy #2 - Following a Lead / Trying to Solve a Nearly Complete Row: In some ways, this strategy contradicts Strategy #1 because it isn't concerned with asking broad questions. Its focus is on obtaining 100% certainty for a single row (sex/place/weapon) on your sheet. These questions are best used when you strongly suspect you have a specific card and certifying that you have it would cross off two or more other cards on your sheet.
Example: You hold the blowgun (yellow ranged) as a dead card. The rifle (blue ranged) is in front of Opponent A and the gun (orange ranged) is in front of Opponent B. Someone asks Opponent C about ranged weapons and C responds "3". Since there are only 4 ranged weapons in the game, the 3rd card Opponent C sees must be the crossbow (pink ranged). Thus, the crossbow is either a dead card that C holds, or it is in front of you. Following the lead of C's "3" response, you ask about ranged to either Opponent A or B. Although it is disappointing if A or B responds "1," there is a huge payout for an answer of "2" because the latter confirms with 100% certainty that you have the crossbow. This allows you to cross off all of the up close weapons off your sheet, and potentially other pink cards depending on the previous clues that were given.
Utilizing Meta Information from Clues
1) People will only ask about a card if it is possible for them to have a card from that category. Sometimes you can cross out cards on your sheet just based on who is asking a specific question. Example: Opponent A asks you about blue cards, but Opponent B and C each have 1 blue card in front of them. You can cross off the third blue card on your sheet because Opponent A would never have asked about blue if that card was in front of you since the other 2 blues are in play.
2) More common is the scenario where a person seems interested in a category that you are also interested in. Don't X out cards in this case, just keep this information in mind when you are trying a chance accusation because you can prioritize accusing on the categories people seemed less interested in.
Example: Opponent A asked a blue question, and no one has asked about yellow this game. You are uncertain whether you have the maid (blue female) or the nurse (yellow female). All else being equal, you should prioritize guessing the nurse in an accusation because Opponent A may have thought they also had the maid since they asked on blue earlier. This is not a guarantee, just a guideline for pushing the odds in your favor.
Cards in Peoples' Hands
1) Most games you can deduce 1 or 2 dead cards people are holding. Take note of these (e.g., "Opponent C holds the knife").
2) Take notes about categories of dead cards people are holding, even if you don't know the exact card (e.g., "Opponent B holds a dead outdoor card"). This information could prove useful later.
3) Rarely, you will be able to deduce both of the cards a single person is holding. Even more rarely, you will have time to ask that person a question right before an accusation. Since you know the exact misinformation that person could give, you'll be able to interpret their response with 100% certainty.
Using Opponent False Accusations
1) Immediately cross off any cards on your sheet that were in an opponent's false accusation.
2) After each false accusation, consider each of the 3 cards that appear in it. Can you combine the knowledge of previous clues asked with the knowledge that your opponent thought they had that card in front of them?
Example: Opponent A gave a false accusation that contained the countess (green female), and opponent A doesn't have the countess in front of them. During a previous clue, Opponent B said they saw 1 green card in play. If Opponent A thought the single green card in play was the countess in front of them, you can eliminate ALL the other unknown green cards on your sheet.
When and How to Accuse
1) Keep in mind that accurate accusations, on average, tend to come out around the 10th clue of the game. If you have a single magnifying glass and there are 9 or 10 clues in play already, it makes more sense to make an accusation instead of asking another question unless there are a lot of unknowns on your sheet.
2) Although rare, sometimes by around the 10th clue you will have the correct solution with 100% certainty.
3) More common is deducing down to a 50/50 chance; always accuse with 50/50 odds instead of asking questions unless you have more than one magnifying glasses, in which case accuse after asking a question.
4) Going for a 33/33/33 chance is not unreasonable either, especially if the clue count is at or above 10.
5) It's always better to throw out chance accusations when you are 100% certain about 2 or even 1 row on your sheet compared to being uncertain about every row - even if your accusation is false, if two of the cards in the accusation are correct, you give out minimal information your opponents can exploit. This increases your chances of being able to make another accusation.
6) If your odds aren't great, consider how far above the 10 clue threshold the game is at. You may be able to ask one more question and hope the turns cycle back to you if the game is within the 9-12 clue range, but waiting for your accusation gets riskier as the clues in play grow.
7) Also consider if your opponents seem close to solving. If Opponent C threw out an accusation with 2 correct cards and 1 incorrect card, there's a good chance they'll correctly accuse on their next turn so you should always accuse instead of asking a question, regardless of the odds, because it's not likely your turn will come around again.
These are some of the essential strategies, but it is by no means comprehensive. I'm still learning new things about this game, and there are also things I know are possible but just choose not to do. For example, you could go even deeper by carefully considering if a question would help others more than it helps you, although I have done fine without going into deep thought about what my opponents might be looking for (well, at least beyond General Tip #5, since that one is pretty easy to adhere to).
A theorycraft idea I have (but haven't executed on due to the time investment) is to do statistical analysis to see if there is any sort of bias in which cards people choose for the beginning combination (such as preferring the left card when they have to choose between two cards, or if players prefer to give cards of a certain color or other category). This data could be used by a competitive player to inform their accusation choices to gain an edge. This statistical analysis could even be done to develop profiles to target specific players' biases and habits.