Record · Understand · Practice
Team shot statistics
Learn what 0–4 grades mean, track your team’s shots and turn the numbers into a practice plan.
Start your team shot logA clearer plan. A calmer call.
Skip’s School
You don’t need a speech for every shot. You need a plan your team understands.
Every position · Every level
The Mental Game of Curling
Reset after a miss, work with nerves and keep team communication useful when a game gets tense.
Explore the mental-game guideBuild the team. Learn the sheet.
Teamwork, ice reading and rock matching
Building a Curling Team
Choose people to grow with, agree on a season, and build clear team communication.
Download printable PDFReading Curling Ice
Compare paths, spot changes and record what your team sees.
Download printable PDFRock Matching and Your Rock Book
Compare stones carefully, explore fictional sample records and keep your own notes.
Download printable PDFNew to skipping, or ready to think a little deeper? Start with the score and hammer. Work toward shot choices, useful misses and the skills your team needs. These are club learning tools, not an official coaching qualification.
Make the call work together. Explore beginner and advanced sweeping, stopwatch timing and team communication.
Try the Directional Sweeping Lab.
Eight interactive challenges: read the turn, move the brush, choose a goal and compare practice records.
Play the sweeping gamePrint it. Bring your team.
The Skip’s School Club Workbook
Work through practice situations, choose a call, sketch a plan and discuss what happened. Every practice page includes a blank curling house for your team’s notes and stones.
Includes drills, reusable worksheets and facilitator discussion notes. Print the pages you need or use the whole workbook with your club.
Download the club workbook · PDFMake your ice time count
Build Your Own Practice
Choose a skill, two, four or eight players, and 30, 60 or 90 minutes. Get a printable plan with drills, rotating roles and room for notes.
Build a practice planYour call. Then the next stone.
Build an End
Follow all 16 stones in one worked example. Make a call at three key moments and compare the reasoning with Granite.
Step through an endStart here
Build your skipping toolkit
What does a skip do?
Read the house, choose a realistic shot and give your teammates a clear target. A good call includes what happens if it is a little off.
Read the basic skipping guide →With hammer. Without hammer.
Do you want to build a score, force one or create a steal? Start with hammer, then adjust for the score, end and stones in play.
Explore hammer decisions →Give the next shot a purpose
Center guards, corner guards and open paths create different opportunities. Think about the reply as well as your own stone.
See two starting patterns →The same shot can mean something different
Protecting a lead and chasing points often call for different risks. Decide what result helps you before choosing a favorite shot.
Look at late-game choices →Count useful outcomes
A partial make can still win. Compare the results of two calls, then ask how often your team gets each result.
Try the decision calculator →The shot is missing. What can we still save?
Spot a useful alternative while the stone moves, make a clear call, and recognize when the whole end needs a new goal.
Learn Plan B and bailing out →Talk through a shot
Use the simulator’s illustrated examples to discuss targets, contact and the result you want. It is a learning aid, not a prediction of real ice.
Open the shot simulator →06 · Read the shot as it happens
When to Bail Out and Switch to Plan B
Granite: “The original shot may be gone. Look for the useful shot that’s still available.”
The shot is missing. What can we still save?
A stone is light, heavy or off line. Recognize the change, find a useful result that is still reachable, and tell the team quickly. This is a team skill: the sweepers judge weight while the skip or acting vice watches line and possible contacts.
Before delivery: agree on the target, a useful miss and the danger. “We want to draw behind that guard. If we’re short, a guard here can still help.” The backup must fit the score and stones remaining. A guard that helps early may achieve nothing on your final stone.
- Hear the weight. Sweepers report the likely finish: “Short of the house,” “Back eight,” or “Through.” Update it as the stone travels. The thrower’s release feedback is an early clue.
- Read the remaining path. Can realistic sweeping still get the original shot? If not, check the available curl, nearby stones and possible contacts. Find a useful finish on the path the stone can actually reach.
- Name the new job. Say “Plan B” and a specific result: “Keep the guard,” “Get it into the rings,” or “Clear that stone.”
- Give one clear instruction. Call “Sweep for weight,” “Sweep for line,” or “Off,” as needed. Sweepers keep reporting weight; the house resolves competing priorities. Watch until the stones stop.
A quick conversation
Sweeper: “Too light for the house.”
House: “Plan B, guard this side. Off.”
Sweeper: “Guard weight.”
House: “Good. Keep watching.”
This is one example. It does not mean every light stone should be left unswept.
Three moving-stone examples
These are fictional teaching situations. Choose the call for the actual path, weight, ice and team.
1. A light draw becomes a useful guard
You are building a steal with more stones to come. The intended draw cannot reach the house, even with realistic sweeping. Its line can still protect a useful route.
New job: finish as a guard that helps the next shot.
Possible call: “Plan B, guard. Off.” If it needs help reaching the chosen guard zone, call for weight instead.
Check: What does it protect? Does it block your own next shot? Don’t abandon a draw the sweepers can still reasonably bring in.
2. A heavy corner guard becomes a stone in the rings
With hammer, you wanted a corner guard while building two. The stone has enough weight to enter the house. There is an open route to a useful position on the side.
New job: keep a potential scoring stone without creating a damaging contact.
Possible call: “Plan B, keep it on the side. Sweep for line.” Use this only if holding the line helps and the projected weight still lets the stone remain in the house.
Check: Sweeping for line can add distance. “Heavy” does not automatically mean “Off,” but don’t sweep a useful stone through the back.
3. A wide come-around becomes an exposed counter
You wanted to draw behind a guard. The stone is wide and a little light. The buried target is no longer realistic, but sweeping could bring it into the open side of the house.
New job: keep a usable stone in the rings, accepting that it will be exposed.
Possible call: “Plan B, open side. Sweep for weight.”
Check: Compare that with a short finish. An exposed counter may help, but it could also give the opponent backing or an easy hit. Curling Canada’s coaching manual discusses this line-versus-weight tradeoff.
What sweeping cannot fix
Sweeping can influence distance and curl. It cannot send the stone to any target you choose. You cannot undo earlier sweeping, and stopping does not apply brakes. A fast takeout that completely misses its target may have no useful recovery.
Use permitted techniques. World Curling prohibits techniques intended to slow a stone. Don’t improvise a braking method or assume one sweeping side always produces the same result.
Sometimes the best Plan B prevents a worse contact. Sometimes there is no useful save. Keep the call calm, then plan from the actual result.
Practise spotting the alternative
With a club instructor, use comfortable draw-weight shots and agree on two possible finishing zones. The caller identifies when the original target becomes unrealistic, names the backup and gives one clear instruction.
- Did we recognize the change early?
- Was the backup reachable and useful?
- Did everyone understand the new job?
- What did the stone actually do?
Judge the communication and decision as well as the finish. Review team communication and directional sweeping.
After the shot: does the end plan need to change too?
An in-motion backup changes this shot. Bailing out changes the result you are working toward for the end. Reassess after every delivery, especially when a miss or strong reply closes a route. There is no universal “bail after this stone” rule.
- Check score, end and hammer. What result still helps?
- Count both teams’ remaining stones. Do you have enough shots for the required jobs, allowing for their replies?
- Look for danger and access. Are their counters protected? Is your scoring route disappearing?
- Use this team’s realistic shots. Compare likely useful outcomes and damaging misses on today’s ice.
- Say the new goal. Make the next shot serve it.
With hammer: two becomes one, or a realistic blank
If getting two now requires a difficult sequence and risks your last scoring route, one may become the priority. A blank is worth considering only if clearing all scoring stones is realistic and retaining hammer helps under your competition’s rules. A crowded house does not become easy to blank just because you want out.
Team phrase: “Two is getting expensive. We’re working for one. Keep that draw route available.”
Without hammer: a steal becomes a force or damage control
If your protected stone is gone and the opponents have separated counters, another guard may only help them. Compare removing a threat with drawing or freezing closer than their second counter. Try to make them work for one; if that is no longer realistic, limit the larger threat.
Team phrase: “The steal has faded. Watch their second counter. This shot’s job is to reduce the damage.”
The scoreboard can overrule the safer-looking choice. Tied in the final end without hammer, forcing one loses. Down two in the final end with hammer, scoring one loses. A difficult attacking chance may be necessary when the lower-risk result cannot keep you in the game.
A simple reset: “Our new goal is ____. This shot’s job is ____. We can accept ____. We must avoid ____.” Confirm target, turn, weight and sweeping plan.
Coaching sources and rules
- Curling Canada: Don’t skip out on skipping — looking for alternatives while a stone is moving.
- Curling Canada: Discover Curling, PDF pages 51–53 — useful alternatives, weight reporting and line/weight tradeoffs. Used for enduring coaching principles; some rule examples in this manual are historical.
- Curling Canada: Long-Term Curler Development Guide — end contingencies and bail points.
- World Curling: Sweeping technique policy.
- World Curling rules, effective September 1, 2026; revised October 6. End examples use standard four-player curling. Check your competition’s rules, including protected guards; mixed doubles and other formats can differ.
Learn the game. Love the game.
Your quarterback-style reminder
A quick glance before the call
Score. End. Hammer. What do we need? What is the useful miss? How much time is left?
Skip’s Sideline Card
One printable page with two panels to cut or fold. Keep it beside your practice sheet, in your bag or laminate it for the season.
Download the quick card · PDFLast Rock Decision Cards
Three illustrated situations: tied, down one and up one. Includes worked examples, team observation sheets and a plan for checking your estimates.
Download the 12-page pack · PDFA curling “Moneyball” pilot
Which call gives you more ways to win?
This calculator explores the first card in the pack. Change the estimates and see when the preferred call changes.
Teaching example, not measured performance. Every starting percentage is invented. This is a simplified final-shot comparison, not a validated recommendation for your team.
The situation: tied, with the last stone
Standard four-player curling. This is the last stone of the scheduled final end; the opponent has no reply. Our A1 stone is four feet right of the tee. Their O1 stone is two feet in front of the tee and is shot stone. There are no guards or other stones.
Call A: draw
Full: finish within one foot of the tee, clear of O1. We score one and win.
Useful partial: miss that target but still finish closer than O1. We score one and win.
Miss: O1 stays closest. We lose.
Call B: hit
Full: remove O1, keep the shooter in the house and leave A1. We score two and win.
Useful partial: O1 and the shooter go out; A1 stays. We score one and win.
Miss: O1 stays shot stone. We lose.
Model limits: draw branches assume no contact; hit branches preserve A1. Unexpected contacts, blanks or tied measures need additional branches. Record them rather than forcing them into a convenient label. “Full” and “partial” describe these particular targets, not official shooting grades.
Check your thinking
Why can a lower full-make rate still be the better call?
Because a useful partial can still win. Compare every outcome and its value, not just perfect shots.
Does this prove that hits are better than draws?
No. It only follows from these invented inputs and this simplified house. Change the estimates or the possible outcomes and the answer can change.
What if the estimates are almost the same?
Small samples and changing ice can easily outweigh a small modeled difference. Treat it as uncertainty, not a precise ranking.
A second decision: value the result
Down one: win now or tie?
An easier shot can keep you playing. A harder shot can win right now. Compare the chance of winning the whole game, including what happens after a tie.
Fictional teaching estimates. Every starting percentage and player profile below is invented. This is an explainable practice tool, not a tested prediction for your team.
The situation
Standard four-player curling. You trail by one and are delivering the final stone of the final regulation end. The opponent has no reply in this end. Score one and you go to an extra end without hammer. Score two and you win immediately.
A conditional outcome model: assume a coach has checked that both calls are available on the actual layout and that the branches below describe their possible results. This tool does not simulate stone paths or guarantee that a particular house offers these shots. Blanks, tied measures, other scores or unexpected contacts need different branches. Do not force those results into this example.
Call A: draw for one
- Score one: tie the game and play an extra end without hammer.
- Concede one: lose the game.
Call B: double for two
- Score two: win the game now.
- Score one: a useful partial; tie and play an extra end without hammer.
- Concede one: lose the game.
Read the result, then test the assumptions
Where does the 16.25% break-even rate come from?
With the starting estimates, the draw wins the game 90% × 25% = 22.5% of the time. The double’s tie branch adds 25% × 25% = 6.25%. Its outright-win branch therefore needs another 22.5% − 6.25% = 16.25% to draw level.
Holding the draw’s tie rate, the double’s tie rate and the extra-end estimate fixed: break-even score-two chance = (draw tie chance − double tie chance) × extra-end win chance. Multiply probabilities, so 90% × 25% means 0.90 × 0.25.
Changing a player’s technique may also change the partial and miss rates. Holding one branch fixed is a way to explore the decision, not a promise about real attempts.
Why can the draw average more points but win less often?
In the starting example, the draw’s expected signed end score is +0.80 and the double’s is +0.55. But one point only ties this game. Two points win it. The harder call can have more game-winning value even while its average end score is lower.
Expected points and chance of winning answer different questions. Here we are trying to win an ordinary win/loss game.
What should our team record?
Use comparable situations. Agree on the target and outcome branches before delivery. Record every attempt, including unusual outcomes, and keep sample size, player, turn, intended and delivered weight, ice conditions and sweeping context.
Compare your estimates with later attempts. Small samples, changing ice and different opponents can change the answer. A good call can still miss, and one successful shot does not prove a call was best.
If the opponent still has a stone to play, you must include their plausible replies. This final-shot example has none within the end; it includes the opponent’s extra-end strength only through your extra-end win estimate.
Start with the coaches
Playing the Numbers Game (Curling Canada, 2022, pages 44–45, PDF) is the article behind the “harder winner or easier tie?” discussion. It describes how coaches and teams use analytics; it is an article, not a controlled trial.
Renée Sonnenberg’s 2023 coach handout: Win Probability, Shot Difficulty & Video (PDF) gives practical questions and a recording framework. Its historical teaching rates are not this team’s estimates. Curling Canada’s 2023 training-camp report shows how teams combine situation-specific analysis and video.
Nerd out: the research
These papers study different parts of the problem. Real match records, real stone layouts, simulated shots and robot matches are different kinds of evidence. None supplies the fictional percentages used in this calculator.
- Kevin Palmer, Gerry Geurts & Jason Gunnlaugson: The Evolution of Curling Analytics (2019)
Conference paper · practical analytics. A bridge from team statistics to situational scouting, game-winning probability and decisions. The official conference page links the paper. - Brenzel, Shock & Yang: win probability from score, end and hammer (published online 2018; journal issue 2019)
Real-match data · end-level model. Uses 1,199 Canadian men’s championship games from 1998–2014 to value game states. It does not inspect stone layouts or estimate a specific player’s shot execution. - Fry, Austin & Fanzon: Elementary Econometric and Strategic Analysis of Curling Matches (2025)
Team-strength model. Accounts for differences between the teams’ offensive and defensive strengths. Useful for thinking about the matchup behind a game-state estimate; it remains an end-level approach. - Ahmad, Holte & Bowling: Action Selection for Hammer Shots in Curling (2016, PDF)
Simulation · final-stone decisions. Compares calls while allowing imperfect execution. The final-stone setting removes any opponent reply within the end. The Olympic comparison is not an on-ice randomized trial. - Yee, Lisý & Bowling: Monte Carlo tree search with execution uncertainty (2016, PDF)
Simulation · opponent replies. Studies sequential decisions against an opponent when execution is imperfect. This is a basis for looking beyond your own next shot; the gains were demonstrated in simulated curling. - Chiba and colleagues: valuing curling game states from real boards (2026)
Real boards + simulated outcome labels. Uses configurations from 1,752 games. Final-shot outcomes are labeled through simulation; earlier values are rolled backward from the next model. Real board data do not make the alternative outcomes observed human results. - Oberlin and colleagues: backward induction and reinforcement learning for curling (2026)
Preprint · reduced simulation. Explores strategy in a four-rock setting. Training and live-support uses are proposed applications, rather than a validated full-game assistant for human skips. - Won, Müller & Lee: the curling robot Curly (2020)
Real-ice robot experiment. Reports three wins in four matches against expert human opponents while adapting to changing ice. This is an important feasibility result, not proof that a recommendation app improves a human team’s win rate.
Keep the rules era and format in view. Older game data predate today’s five-rock and no-tick rules. Historical rates need checking before reuse. Four-player curling and mixed doubles also need different assumptions, including hammer transitions. Publication date alone does not make a model’s estimates suitable for your team.
Use this in practice or after a game. World Curling rule R11(c) forbids electronic devices that provide information to players on the athlete field of play during games; time-only stopwatches are an exception. Check your event’s rules and permissions before any live use. Read the official rulebook, R11(c), page 17 (PDF).
Download the club experiment workbook
Six printable pages: set a repeatable draw-versus-hit test, record every attempt, and compare results cautiously. Blank houses included on each practice page.
Turn estimates into evidence
Build a team profile, one useful skill at a time
- Pick comparable situations and agree on full, useful partial and miss before throwing.
- Record the turn, target, weight, difficulty, ice conditions and who swept. Keep the actual final score separate from the shot grade.
- Keep all attempts, including unusual results. Write down sample sizes; a few good shots do not establish a reliable rate.
- Use the patterns to choose practice priorities. Check your estimates against later games before relying on them.
A lineup is more than four shooting percentages. Sweeping, communication, ice reading and the combination of skills matter too. This pilot helps you ask better questions; it does not identify an “optimal” roster.
Research behind the idea
- Action Selection for Hammer Shots in Curling: combining execution uncertainty with the value of the resulting position.
- World Curling: Solid numbers: why graded shooting percentages need context.
- Win Probability, Shot Difficulty & Video: coaching notes on tracking outcomes and using them in practice.
HouseCall teaching pilot. Examples use hypothetical data. Source review: October 2026.
Learn the game. Love the game.
Make the plan clear. Give your teammates a shot they believe in.
More ways to develop your game
Curling with a delivery stick
Understand the equipment, develop a repeatable delivery and take a practice sheet to the club.
Download printable PDFWheelchair curling strategy
Four-player decisions, no-sweeping tradeoffs, four fictional positions and a team worksheet.
Download printable PDFOne problem. One useful next step.
Granite’s Curling Troubleshooter
Heavy draws? Missed line? Mixed-up sweeping calls? Pick a problem for possible causes, one small drill and a lesson to explore.
Find help with a curling problem →