Road to 1500 elo

2023-12-29

About two months ago I started a challenge for myself: achieve a 1500 elo rating in chess. I have no training or history in chess whatsoever. I knew the common rules before I started, but not the nitty gritty about castling and en passant. Let alone the 50 move rule. Or the myriad ways of running into stalemate.

Meta game


I realize that many perceive chess as this dusty boring ancient game. I've never really considered it like that, even though I hadn't invested any time into the game myself either.

Having invested a bit of time now I can only conclude that chess actually has parallels to Magic the Gathering or Hearthstone (or anything else). The meta just takes 10 years rather than 10 weeks. And "expansions" take form of dusty books with hard to read cryptic codes, sorry, no new pieces.

No but for real. In any deckbuilding game you have to know the meta to play well. You must know what kinds of decks are popular and likely to be played against you in order to know how to counter them, or how to play around certain types of attack setups.

In chess, a deck is comparable to a "chess opening". The first two or three moves may seem irrelevant in a game that has a gazzilionmillion different possible ways of being played, but they can tell you so much about the strategy your opponent intends to take.

And much like any modern game, there is a meta in chess. The meta periods just take much longer and are more subtle to the untrained eye (although the latter point could be argued about any game). There have been years where the Indian opening was a popular chess opening, and "d4", and the English, etc. These openings have certain counters, or if nothing else counter moves that give you the best edge according to statistics. The caveat, of course, that these statistics are often assuming very high level play and not like hanging minor or major pieces ;)

Beyond the openings, there's also some interesting depth in the pieces themselves. I once heard a streamer talk about the pawns and how they defend the king, form a line, and a whole bunch of theory on pawn structure and order and what not. I thought it was interesting to think of the game in those lines.

So, there's more to the game of chess than meets the eye. But it certainly doesn't have flashy graphics (actually, I take that back) and requires the consumption of some often dry book material. I think it's easier to consume that through videos of charismatic people. But I find it very hard to make my way through the ocean of online content and find the relevant content for your level.

Average


If you know anything about chess, and I don't blame you if you don't, you'll realize that a 1500 elo rating isn't the end of the world but it's still quite ambitious for somebody like me to just want to get there. For example, you become a grandmaster (GM) at 2500 (casually ignoring additional requirements) while a 1500 rating is apparently considered "an average club player".

I figured aiming for "average" might be somewhat attainable and only requires about 120 years of chess to catch up with. I'm not interested in a formal title, winning a tournament, or anything like that. In fact, I mostly set this challenge so I could justify myself buying this awesome looking Evo Chessnut chess board, which at this point I don't even want anymore.

So let's nuance my challenge a bit;
I want to attain a 1500 rating in "rapid" chess on chess.com.

Nuance


There's three things to note about this:

- I'm only playing online games for this. No "over the board" games, which have a slightly different feel. Online games also give (me at least) a slight advantage in that I can draw arrows to plan my attack. Over the board, you obviously can't paint arrows. You probably have to pay a fine for ruining the board and all that. It gets messy quickly.

- The chess.com elo rating appears to be "worth less" than the "more official" FIDE elo rating. Translation: it ought to be easier to attain 1500 on chess.com :) Beyond that, it's like playing other ladders. The consistency includes freebies and uncontrollable losses (like getting disconnected or something). Whatever an "elo" is worth on chess.com, it's at least relatively consistent on that platform. Just a number to chase. I'm very good at that :)

- "Rapid chess" means each player gets 10 minutes on the clock and that's it. It slightly skews how you play chess. More on that below.

In other words, even if I do attain 1500 in rapid on chess.com, I probably still won't be able to rank well in an over-the-board tournament of roughly equal rating. But that's fine, that's not my goal :) The number is a bit arbitrary and at this point seems like an ambitious enough goal that may still be attainable.

Rapid chess


There are a bunch of different formats in chess. Well you can do any timing format that combines time on the clock and time added when making a move.

For example, the FIDE world championship rules state: " The time control for each game shall be: 100 minutes for the first 40 moves, 50 minutes for the next 20 moves and then 15 minutes for the rest of the game plus an additional 30 seconds per move starting from move 1.". That's way too long, who has time for that.

The default on chess.com is "rapid chess", which is 10 minutes on each clock and no additional time. So after 20 minutes, the game will be over one way or another. Often sooner than that. This means I can timebox my games properly, at the cost of not always having time to properly think things through. I think I would fare better with a 30 minute clock but part of that may also just be lack of experience.

The timing aspect has definitely been relevant in a bunch of games. The rule is simple: you lose when you run out of time, no matter how good or bad your position is on the board. I've probably lost more games on time than I've won on time, so far. But I think it's become less of a problem with more recent games.

Although even as recently as last night I had a game where I lost on time, when my opponent had 1 second left, and I could have won with a checkmate had I put my queen one cell lower than I did. It can be a bit depressing (:

There also was one game where I knew how to close it but still had to promote and achieve ladder mate with about 20 seconds on my clock (and them having 5 minutes). And I literally won that game with 0.0s left on the clock, making three moves in the last second of the clock. Didn't even think that was feasible. Or that zero would still lead to a win. But, yes, it does.

There are also faster chess game modes like 3 minutes ("blitz"), 1 minute ("bullet"), and even a 30 second "hyperbullet" variant. I've heard playing bullet chess is like putting your brain in a blender. Maybe I should try that some day. I also heard it deteriorates your tactical gameplay while improving your time skills, which makes sense. Faster decision making is key in any time game.

chess.com


From what I can tell there are two main online platforms for chess right now; chess.com and lichess.org, the .com is basically paid and the .org is open source.

Despite not paying for it (yet?), I have to admit that I prefer the chess.com environment over lichess, right now. I think it's also a slightly more popular platform in general and its rating seems to be slightly closer to FIDE than the lichess rating.

I dislike the hard pushed upsells for anything. Analysis, puzzles, trainings, everything cuts off after a low number of freebies a day. At this point I'm using the service enough that arguably it might be worth the 8$ or 11$ /mo. Plus, I understand they need to make money. There's a tradeof between slick interface and money.

I do like its analysis, even if it feels off at times. But it seems I can make do with once a day. And I use chess compass for stockfish analysis of games afterwards.

Free resources


For now, I make do with several free services instead.

- chess compass (stockfish analysis of games)
- braimax puzzle rush
- blitztactics (puzzles)
- chessbook (openings training)
- 365chess (openings database)
- openingtree (chess openings)
- chess endgame trainer
- lichess board editor (to create FENs)

Road to 1500


So where am I at with this road to 1500 elo? Well... roughly in the middle, apparently. Although I'm not sure elo offsets at zero, heh.

I started about two months ago. My initial chess.com rapid elo was about 600. That held for a bit and then I tumbled down to about 450 before bouncing back up to about 800 right now.

I've also learned that there are a few key factors to train to improve the skill level:

- Puzzles
- Openings
- Endgames

I want to stress that while I did enjoy The Queens Gambit, like last year or shortly after whenever it was released, it actually did not inspire any of this. Sorry :)

Puzzles


I still find it ironically but puzzles are apparently the way to improve your chess game.

The idea is that you'll recognize patterns which help you to recognize certain mate-in-x situations in a real game faster. It also helps you to recognize certain attacks, both for you to attack your opponent and attacks coming to yourself, and how to counter them.

This makes sense to me. Even though actually doing puzzles don't feel like you're working on improving your chess game, I do still hang my queen far too frequently which is flat-out costing me games (well, at my level that is not a lost game per se but...).

Since chess.com puzzles are paywalled after the first few, those are out. I don't like the Lichess puzzles because they're toooo random in terms of topic. Many chess puzzle websites out there seem to still be living in 1999, with their tiny boards and horrible interface. Kind of strange, but probably catering to a very particular audience.

I'm currently using braimax for puzzle rushes. And while I got to "level 3" fairly quickly, I've not been able to get passed level 3, so there's that. Still, it does feel like good practice.

Openings


There's a million different way to start the game. As white you basically determine the main line of the initial half of the game. As black you have to follow suit and respond to whatever white is playing.

At this point in time, I think openings are ironically the least relevant to learn. Some basic theory on openings may be nice to have but for yourself you should find an opening that works well for you and stick to it.

I started with the English opening (a "system" opening that typically ends you with a pyramid of pawns). But that wasn't as consistent for me, probably because people in my elo range just know how to work that (or attack that).

So now I'm playing the Hungarian opening, which starts with a pawn from g2-g3, bishop f1-g2 (called "fianchettoing your bishop"), and king side castling. It's not considered the best position for white but I find it has three advantages:

- It forces me to castle early, which means I don't have to worry about it later
- It prevents many cases of back rank mate because the g2 pawn has already moved
- It puts a decent attack on the a8 rook
- It prevents common aggressive opening attacks from black
- It confuses the heck out of opponents at my level as you don't see this opening a lot

I think that gives me an advantage. I think I have a 70% win rate or so with that opening, although I'm not sure if chess.com properly recognizes the whole opening as such (it always considers me to make a "mistake" in the first few moves, even though they're part of the opening).

I'm not super comfortable as black yet. The d4 opening is very common at my level, but the d5 counter play doesn't feel comfortable to me. I've been trying to do the "Scandanavian defense", which means moving the pawn to e5. But so far that's been a mixed bag for me. Still trying that though.

A good free tool for this is chessbook.com, which gives you a nice interface to practice common openings and gives you stats on what GM's commonly play in that position.

There's also openingtree.com to explore openings and keep practicing them. And you can use this service to search for openings, ignoring the 90's look and feel.

Endgames


The last one is the endgame. There's no absolute rule for when the endgame starts, but you can consider an endgame when it's more about getting a pawn promotion than anything else. Commonly when there's one or two rooks and a knight or bishop on each side and a bunch of pawns.

I think there's two kinds of endgame, actually. There's the pawn endgame, which is mostly a king, its pawns, and maybe one or two other pieces. And there's the king versus king plus other piece.

I think the pawn endgame is a matter of practice. There are so many different ways this can play out. Doing puzzles and using tools like chess-endgame-trainer will up your endgame.

At least, you should practice the ladder mate (two rooks or rook and queen) and queen king mates. You don't want to be figuring out king positioning with 20 seconds on the clock. And the worst is triggering a stalemate when you could have just won.

Computers


Somewhat tangentially, I've been toying with teaching a neural network to play chess. I'm not hoping to achieve anything significant here to be honest. Mostly inspired by this project "Training AI to Play Pokemon with Reinforcement Learning". I love that stuff and it sniped me to try and do this with chess.

Ultimately that led me to realize that I would need to validate moves made by a computer. So for that I wrote an engine. And then I wrote a UI on top of that to validate it. And so... well, yesterday I released Yacge, which is the resulting engine / chess board.

The network training is a work in progress. Maybe more on that later.

Conclusion


I'm not sure if I'll actually reach my target elo. It's going to take quite some time to stop hanging my pieces, stop walking into traps I've seen before, stop being too aggressive in my attacks. But I still think there's a fair chance of actually getting there. Who knows.