Dojo Tactics Tests – A New Way to Assess Your Skills
Kostya Kavutskiy • May 15, 2024
With our 3.0 launch, the Dojo has released a new training tool designed to test tactical skill. Introducing the Dojo Tactics Tests, an attempt to improve on the most popular existing "tactics trainers" (Chess.com, Lichess, ChessTempo, etc.) and provide players a much more realistic “tactics rating.”
This is an exciting feature as we aim to 1) help players improve with a new and useful way to work on tactics/calculation, and 2) allow players to assess their skills in a reliable way (no more 2300+ puzzle ratings!). We’re starting with our version of "tactics tests" and ratings, but in the coming months we plan to expand to positional, endgame, and even opening tests too, with a separate rating for each category.
- There’s always a solution. This seems logical, but as most players have probably experienced, tactics are MUCH easier when you know there’s an answer. But in an actual game, it’s up to the player to figure out if there’s a working tactic or not, and learn how to analyze and come to an objective conclusion. Some lines work, some don’t. With the Dojo Tactics Tests, there won’t always be a flashy puzzle answer. Problems which don't have a "puzzle answer" will accept any decent move in the position, which means no more random guessing at the solution.
- You’re given the opponent’s response. Learning to figure out how the opponent is going to respond to your ideas is a key component of chess. In typical tactics trainers, you’re incentivized to just play the “correct looking move” without actually having to calculate (or visualize) all of the possible consequences. This can lead to bad habits (playing too quickly, missing the opponent’s resources), poor visualization, and can even stagnate one’s chess. In Dojo’s Tactics Tests, you’ll need to input the moves for yourself and for the opponent – full credit requires noticing and including all important resources for the other side.
- You’re not tested on critical defenses. For most puzzle trainers, there has to be “one” solution, and this prevents the machine from showing the most principled defenses, as there might be more than one move that wins against them. With the Dojo Tactics Tests, alternates are scored, and the student is asked to find all the key defenses themselves.
The Dojo hopes to improve on the above issues with the new Tactics Tests, where users are asked to input their own analysis of each problem, which is then graded against the full solution. All of the test puzzles have been hand-picked by a Sensei and have lightly-annotated solutions.
How the Tests Work
- Depending on the level, students are typically given 30-60 minutes to solve 6-12 exercises.
- Every problem has been hand-picked by a human coach.
- For each problem, it is up to the user to input their solution on the board, along with any key defenses and their response to each defense.
- The problems are scored based on how many correct moves the student finds.
- All of the problems have annotated solutions, indicating which moves are scored and which aren’t.
- Only the first moves you input for the solving side count, so players can’t just guess multiple answers and hope one of them is right.
- Puzzles also give full credit for alternates (sometimes two or more moves win, that’s common in chess!) and partial credit for partial solutions.
- Not every puzzle will have a “tactical” answer. This is intentional, to discourage guessing. If you don’t see a clear solution, just input the best move that you can find!
This is a sample puzzle you might be presented with for the 1500-2000 level. It’s Black to play. You can either spend a few minutes and write down your solution to see how you’d be graded, or if you’d like to try the sample for yourself on the board, you can do so here. Remember: if there are multiple defenses, you must include all key variations!
Solution
1...Nxg4! 2.Nxg4 (2.Qxg4 Rxe5-+ Black has won a healthy pawn.) 2…Bd6! 3.Qf3 Rg5 4.h3 f5-+ Winning the piece back with interest.
If you only put down “1…Nxg4”, you would get 1 point for that move. If you put “1…Nxg4 2.Qxg4 Rxe5”, you’d earn 3 points for the puzzle. To get full credit, you’d need to input all of the moves bolded above, as well as the variation 2.Qxg4 Rxe5, which would be worth a total of 9 points. Since players are scored based on what lines they see, we hope this new tool will push players to improve their analytical skills. But there’s a second benefit to the tests as well...
Scoring & Rating
Players will be scored based on how many points they earn on their completed test. The results of each test will also provide players with a “skill rating,” where they are rated based on how well they do compared to others. Our theory is that as more players take the tests, it will give everyone who’s taken the test a more accurate representation of their skill level compared to others.
So for example, a player rated 2000 and higher should aim to score close to 90-100% on a 1500-2000 test. An 1800 should aim for 50-75%, and a 1600 should aim for 25-50%.
Data for the first 1500-2000 test. In this chart, the x-axis is the score on the test and the y-axis is the calculated rating for the test. Different colors represent different cohorts.
We’ve launched a series of tests aimed at the 1500-2000 level and the 2000+ level. As we get feedback on the tests overall, we’ll roll out tests for players rated between 1000-1500 and under 1000 as well.
As mentioned, we will eventually be making positional tests, endgame tests, and even some opening tests! Ultimately these will fit into one’s chess "profile," where you’ll be able to see all of your ELOs and "skill ratings" at a glance.
We often talk about our students or fellow players in terms of their strengths and weaknesses, but we rarely have a way to actually quantify or measure them. Moreover, it’s very hard to tell if one is improving – yes if you gain 50 points in a tournament you can conclude that you’ve improved, but these events are often few and far between. With these tests, we will hopefully give players a clearer sense of what they need to work on and whether they are improving!
Currently free users can access the first of several tests that are available for their rating range. Training Program members get access to all tests.
Time to get solving!