What the heck do individual game analysis ratings mean?!

I would really appreciate any insight I can get into the ratings that the analysis engine gives for individual games.  I am feeling extremely frustrated because I am losing nearly every (rapid) matchmaking game that I get paired into, and when I dutifully go to analyze my losses, each and every one of these 1020-60 rated people I am paired with are turning in games rated 1300-1400.  Meanwhile, consistent with my own overall rating, my gameplay usually gets scored 1000-1050.  I would think that if I am playing against people in the 1000-1050 range, an in-game skill of 1000-1050 should be good enough for a win at least some percentage of the time, right?

I have read posts saying "oh, there are cheaters and sandbaggers operating around this rating level" and I get that, but it can't be every single person I get matched with right?  I am seriously considering to set my custom game filter to exclude anyone ranked at all higher than me so I have more chance of playing 900 rated players against whom I might have a chance, but somehow that move felt unsporting, so I wanted to post this and get feedback first.  I am not a 1300 level player and I am not likely to become one, so I just want to play at an appropriate, sporting level and not get destroyed every time.  I hate how once I drop into the 1020-30 range, suddenly I encounter a run of terrible players who resign early and suddenly my rating boosts again up to 1040 and again the cycle of being killed by people suddenly with 1400 games begins again.

That value is just an estimate of the play based on the player ratings and their associated accuracies. It's not really accurate and should be taken with a grain of salt.

Thanks for confirming this, so in this case finding the actual rating level where my skill is at and playing folks in that pool sounds like the way to go, because my actual rating feels way inflated and seems like I'll need to manually go accept games only with lower rated players where the challenge for me will be more equal. I just want to play within my "class" so to speak, and I feel the automatic matchmaking has just gotten worse and worse the higher my rating gets. Like I'm rated 1050ish and it has been at least 3 weeks if not a month since I got matched with someone below 1000, even though I set my filter to -200 as the lower range of rating.

If you're playing rated games your actual rating will reflect your performance in the pool. Most of the time the system is going try to pair you with someone close you your rating but shouldn't exceed the rating parameters.

Yeah man I wouldn't fret too hard about it. It's a fictional ELO that Chess.com made that is meant to rate your play in a game based on likelihood of winning loosing or drawing. The winner will always have a higher ranked game rating. 
For example: my game rating for a win I just had was 1400. My elo is ~610 and my opponents was ~620 lol. Its just some made up formula.

The winning player won't always have the higher score. It's happened to me quite a few times, including last night, both on the winning and losing side.

I think this rating shows more the quality of a particular game than the actual level of the player. If there were fewer gross errors and more logical moves in the game, the number would be higher. But one successful or disastrous game can easily distort the result. I noticed this after reviewing several games in a row. Sometimes the system feels generous, sometimes very strict. I once read about performance metrics in cаsino game at https://casino-portugal.pt/pt/ and it reminded me how single events can skew averages. In chess, consistency matters more than one clean game. That is why I focus more on mistakes and ideas than on the final rating number.

The game ratings are actually very accurate, but they are a reflection of your moves vs the moves that a computer would make in the same situation. What they dont really take into account is the rating of your opponent. For instance, if your opponent makes terrible moves and consistently leaves pieces hanging, and you take those pieces, you will be making very "accurate" moves, moves that even the best chess algorithm would also make. The game rating algorithm is only evaluating your performance at picking the best moves in that particular game so this rating will vary widely based on your opponent and what moves they happen to be playing. if you know a lot of book moves and opening theory, book moves obviously are the best moves and will contribute to your accuracy, and if you combine that with an opponent that keeps making obvious mistakes, it will be easier for you to attain a higher game rating. But its not something to scoff at and say, "oh it doesnt mean anything, its inaccurate"... it's actually extremely accurate... at telling you how well you responded to your opponents play. The better your opponent is, the harder it is to play accurately is one way to think about it.

Why do you think they're accurate?

I would mostly ignore it. It is an algorithm playing "guess the elo".

Accuracy of play highly depends on the position. For example, in some lines of the French exchange, on my level, me and my opponents often get 90+ accuracy and 2400 estimates, just because the position is so simple that few mistakes are being made.

On the other hand, in complex positions it often happens that both get only like 55 accuracy and an estimate of 1000.

So don't get deluded by the overestimation, don't get discouraged by the underestimation... your rating itself (with its fluctuations, usually +-100 pts) is the best estimation of your skill. It is based on all your games, not just one. GL