fitztrev
commented
9 months ago Hi, can you reformat into markdown and update your comment with the content directly? Thanks
Open SocialSophia opened 9 months ago
fitztrev
commented
9 months ago Hi, can you reformat into markdown and update your comment with the content directly? Thanks
SocialSophia
commented
9 months ago How would I do that?
On Fri, Feb 9, 2024, 12:07 AM Trevor Fitzgerald @.***> wrote:
Hi, can you reformat into markdown and update your comment with the content directly? Thanks
— Reply to this email directly, view it on GitHub https://github.com/lichess-org/lila/issues/14632#issuecomment-1935072135, or unsubscribe https://github.com/notifications/unsubscribe-auth/A274VZO77K6C5BDSUT7NBS3YSVK7TAVCNFSM6AAAAABC6AXBGKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSMZVGA3TEMJTGU . You are receiving this because you authored the thread.Message ID: @.***>
NameLessGO
commented
9 months ago Lichess Tutor SocialSophia
The Lichess Tutor should be organized into a simple to follow menu that basically organizes the Lichess Insights into a simple and easy to read graph that shows insights on where players can improve and where their strengths are. Much like the Puzzle Dashboard (see below) gives players a quick glance at their puzzle strengths and weaknesses this will be the Lichess Insights equivalent.
Lichess planned on releasing this in as early as Q4 2022, so this would be an appreciated feature from the community to the devs. Their last update (shown below) had a lot of potential and just needed some refining tweaks that may have made it simpler to implement.
For the categories to track they should be:
The reason for choosing these categories is most sports games use similar metrics and these happen to be the best at quantifiable evaluation. Similarly to the Lichess Insights having two tabs (see below), the Lichess Tutor should have two tabs as well one for time with the options:
And the other for the time format:
To make the graphics fit perfectly and to use round numbers everything will be based off a 100 rating scale depicted as ten separated blocks. Thus each block represents 10 points, so a player with perfect SPEED would have 100/100 which would be a full bar with all 10 slots filled. All numbers are rounded up to the nearest integer (example 73.8 → 74). The Lichess colour system is nice, so let's expand upon it. Depending on the percentage the bars colour will be different:
Meaning For instance SPEED is a combination of 'clock time usage' and 'flagging skill', for SPEED the formula could be average remaining time (seconds) ÷ time format = 100. Time format would always be equal to 100. For all time format increments would be ignored and the minutes covered to seconds for easier conversion. For instance 5+3 would be 5 × 60 = 300 seconds so divide the time by 3 for ANY 5 minute format.
Example: If a player had 136.4 second average in the last five games for 5+3 time format then you would divide by 3 to get the 100 average which is a 45.3 since most players don't and shouldn't finish games with 90%+ of their time remaining to adjust for this we can utilize Lichess Insights for move time and moves per game this allowing a more accurate representation of how long players spend on moves.
A simple equation could balance out these 6 variables so that they each contribute equally to 100%, such as weighing them at 16.666. Based on averages in the example below we can give 31.2% which we add to the original percentage: 45.3 + 31.2 = 76.5 SPEED feel free to adjust the weight of the Lichess Insights move times. The players stats should follow a bell curve.
Next let's do OFFENSE, this calculates how aggressive the player is and there's several ways to calculate this a combination method like we used for SPEED will serve us best. First we can calculate average material imbalance from Lichess Insights (see below) yet many strong players sac pieces and pawns in order to gain their offensive advantage. Secondly, it would be based off how the player acquires their wins and what's their win percentage. The third metric uses Lichess 'threats' system, based on the number of threats the player exhibits per game average it would raise their OFFENSE rating. And lastly bulk of the weight for this rating could be based off of the evaluations, the more the player is in a winning position the higher this rating would be
For DEFENSE we could use the Accuracy × Variant = 100. DEFENSE directly correlates to accuracy so this one is pretty straightforward.
For the OPENING, MIDDLEGAME, and ENDGAME there could be a few ways to calculate this from comparing the 'game phase' versus others at a similar rating also using average centipawn loss and accuracy of the different game phases (shown below)
Lastly, TACTICS has its own tab in the Lichess Insights, so using those we can precisely measure the tactical awareness per game mode and variance.
To recap for the player 'SophiaSocial' would have an estimated rating of:
The breakdown would look something like this, only it would be segmented and with the correct colors assigned.
Post Game Analysis could include a new section for the rating of the player in the respective 7 categories for Lichess Tutor (shown below is a mock-up).
Along with this change another helpful addition to many players would be the addition of ‘brilliant’ and ‘good’ moves. Which would perfectly contrast the existing mistakes and blunders.
SocialSophia
commented
9 months ago @NameLessGO thanks
Lichess Tutor by SocialSophia.docx A combination of various ideas that could benefit Lichess.