In other online games like csgo and valorant, it's extremely unlikely to reach a certain rank without having beaten (or drawn with) someone of that rank. I guess this is largely because you cannot choose your opponents.

The case is similar in OTB chess where you (usually) can't choose your opponent. So for people peak rated X between, say, 1400 and 2600, they've definitely beaten or drawn with someone with peak rating greater than X. Eg if my peak (FIDE or USCF or whatever) rating is 1488, then I've definitely beaten or drawn with someone whose peak rating is at least 1488.

In online chess like on chesscom and lichess, you can farm rating in choosing opponents lower than you but you wouldn't necessarily be able to compete with people your own rating. Farming is even much more feasible when you do it in a farmbitrage way for chess960 in lichess. It's like this:

  1. Let's say I create a new account and after 400 games, I've been rated about 1326 - 1345 for the last 50 games playing people only rated between (-10,10) of my rating whom I get paired with in public challenges.

  2. Also, I have a peak rating of about 1359.

  3. Being told of the idea of farming, I'll now try private challenging players rated 1000-1149.

  4. Eventually I get a new peak rating of 1360+. But wait let's try to continue farming.

  5. Now I've even reached peak 1580 and have my rating settle about 1537 - 1558 for the last 50 games.

  6. The problem is if I ever go back to the (-10,10), then I'm screwed. I'm about 1537 - 1558, yet I didn't actually beat anyone rated 1500+. I'd be screwed against even a 1480.


What would be some problems if websites like lichess and chesscom were to do something like preventing a user from reaching a rating of, say, 1351 unless they've beaten or drawn with someone already 1351+ (rating at the time of the game as opposed to peak rating. Or you can do peak rating if you want. Let's say at the time of the game) ?

I know it's kinda artificial, but here's how I imagine what could be the case:

  1. Let's say I'm 1347 and the highest rating of a player I've beaten or drawn is 1352.

  2. Now I'm playing against a 1346.

  3. If I win, then I would normally get a +8. But since the most I've beaten is a 1352, then my rating increase for this game is limited to +5 only.

  4. Thus, if I win, then my rating is just 1352.

  5. As long as I keep playing players lower rated than me, I should not have a rating more than 1352.


  1. You can do peak rating instead of rating at the time of the game. Eg I haven't beaten someone rated 1351 at the time of the game, but I've beaten someone who was previously rated 1351 (but was eg 1337 at the time of the game).

  2. Of course this has to be up until a certain point. In OTB FIDE ratings, Magnus Carlsen is peak 2882 but hasn't beaten or drawn with players rated 2852+...except vacuously because there are no such players. So I guess the maximum for this could be like 3000 or something.

  3. It doesn't have to be exactly 1351 to be allowed to reach 1351. It could be that I need to have beaten / drawn with a 1301 to be allowed to reach 1351.

  4. Consider that if FIDE, csgo or valorant implemented such a rule, then virtually nothing would change.


  1. My purpose in this question and my recent farming / rating questions is not really to beat the system or whatever, but to know if I've hit a plateau s.t. I realise 'oh ok time to do more puzzles or study more endgames.' Like if I can reach a rating X without further studying, then why not? Otherwise, how else would I decide 'oh I need to study more' ?

In games like csgo and valorant, I can know exactly this. If after 100 games I haven't ranked up, then I'll be able to realise that if I want to rank up, then I'll have to practice a lot more.

  1. I think also for like sub battles or something. If I'm doing a sub battle, then how would I know the rating of anyone who applies for my team is not obtained primarily by farming? Like there's no asterisk or anything that says 'rating obtained primarily by farming'. So they could've farmed in order to get to a higher rating bracket to join. (I could check their highest rated win, but still.)

In games like csgo and valorant (do they have sub battles? Idk lol), I can know exactly this. If someone is rated 1600 in csgo or valorant (I know they're more on ranks / rating groups than ratings, but you know Gasai what I mean), then I know they've already beaten (or drawn with) someone of that rating. Like they didn't reach 1600 from just beating 1300 players or something.

1 Answer 1


Well, for one thing, you wouldn't be able to start things out this way. Imagine you have a new website and you start people at 1200. Nobody would ever be able to go above 1200 because nobody is there in the first place.

Even if you have an established player base, sometimes two new players will play each other. Under your proposal the winner wouldn't be able to gain any rating points. Indeed, anyone only playing people at their rating would have a difficult time gaining any points.

And this manipulation of the ratings would have a detrimental effect on the accuracy of the ratings as a whole. If you can beat 12 players in a row who are rated 1000-1149, your rating is almost certainly not 1149 or below. Telling the rating system that your rating is actually only 1149 is just going to make you underrated, and that has negative effects on your opponents' ratings as well.

OK, much of this can be mitigated with your stipulation that it doesn't have to be quite so strict and being within 50 points is enough. But then you're postulating a person with a strength of around 1326-1345 who can beat someone rated 1000-1149 more or less every time, but can't get a single win or draw against someone rated 1530 so they can legitimize that 1580 rating. Ratings aren't so absolute that a strength difference of 200 points will ensure a win for the stronger side.

  • Thanks 1 - 'Nobody would ever be able to go above 1200 because nobody is there in the first place.' - ok fine so why don't you just start it after some time? 2 - 'Indeed, anyone only playing people at their rating would have a difficult time gaining any points.' --> what do you mean at their rating? just play someone 2 points higher than you is that 'at their rating'? and so what if they don't rank up? to be a GM, you must beat a GM (unless you were part of the 1st generation of GMs). why can't similar apply here?
    – BCLC
    Commented Jan 26, 2023 at 13:50
  • 3 - 'you can beat 12 players in a row who are rated 1000-1149, your rating is almost certainly not 1149 or below.' - Well if you're genuinely like a 1310 then you should have no problem beating/drawing like a 1360 even just once. Note that they key requirement is not consistently beating 1360's but beating or drawing just once. 4 - '1000-1149 more or less every time' - ok so make it 900-1049 ?
    – BCLC
    Commented Jan 26, 2023 at 13:50
  • 1
    My advice for you is: (1) Stochastics I at university, (2) a good nap and (3) to bury the stupid idea of bolstering your ELO, as what really matters is your chess and not some number representing it (with obvious drawbacks). Honestly, why this obsession?
    – Pit
    Commented Jan 26, 2023 at 20:08
  • 1
    @bclc If you start at 1200 and play +-10 and win every game but can't get more rating than your best opponent, you're eventually going to get to 1400, but it will take 20 games minimum, and probably more like 76 games on average. That is a very long time to be underrated, and it's going to mess with everyone else's ratings in the meantime. In OTB, if a GM-strength player can't be a GM because they lack norms at least they're still rated 2500. I can't speak to csgo or valorant since I've never played them.
    – D M
    Commented Jan 26, 2023 at 21:59
  • 1
    @BCLC Also, barring something like that 960 exploit, it's really not that easy. You have an ability of 1350 and you want to get to 1580 by playing people rated around 980? Can you really win 27 games for every loss and 13 for every draw? Sheesh, the single draw against the 1580 player that your proposal would require would be the easy part. (That's another weakness of your proposal that I sort of hinted at in the last paragraph of my answer: the hurdle is too easily cleared to even matter.)
    – D M
    Commented Jan 27, 2023 at 0:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.