The win rates they report are bonkers off. For example, I’m playing protoss rogue at top legend and according to donkey hs the winrate at that bracket for that deck is 48 percent, making a deck you should steadily lose rank with.
Well I’m not losing rank and steadily heading for top 100 with it, beating already several top 100 players on the journey.
How is this discrepancy possible, how can I be doing so well with this deck when hs donkey is calling it dog water? Could it be differences in my list, of which there are maybe 5 or 6 different cards?
It’s the only deck I play anymore as it and maybe ashmane rogue are the only decks with skill caps higher than “alt tab for your opponents turn, play the obvious card on your turn, repeat”.
In fighting games there are low tier and high tier characters. When you take the average of players on those characters you get a distinct win/loss rating that determines their tier. But, sometimes you come across people called “specialists”. These are people who are able to lift a character beyond its tier/win rate through practice and match up knowledge.
In your post you admit this is a deck you are essentially exclusively playing. You have become a specialist.
are a lot, such changes can make a deck reach much more higher win rates if you are know what you are doing because meta decks contain usually about 4-5 bad cards as far as I saw. These are in just to fill some empty slots. IF you manage to fix these, then you can increase their win rates from about 50% -to about 70%. of your deck. Atleast thats how I reach it nowdays since everything is dead which is not meta.
Guessing its winrates across ALL players, not just individual players?
An average score does not mean that is the chance for all players, but the statistical total sum of all participating players games. (all players combined wins divided by the amount of games recorded = average)
Some players might produce results of 100% winning with it, some players might produce results of 0% losing with it, and there you get a 50% average.
You could just be on a winning streak. You have games numbering in the hundreds compared to Data sights with millions of games. Also it takes players of all skill levels in the data sets. You could just be better than average using that deck. You don’t have scrub data bringing the average down.
Major edit: time to stop posting essays every single time, it’s disgusting.
Yes, 5-6 cards can be too much, see if you can find the correct version of your deck under Decks tab (watch out for how many games you put in a filter, it may be too much)
but also, top 100-200 players’ sample is too low to drastically impact the winrate of the top 2k category, so you can call that tainted data. We would need the category of top 200 but it would be useless because of too little games played.
what TBC said. Becoming a specialist on the deck will do that to ya. Btw, if you’re doing that better on the deck than top 2k, it’s a significant sign the deck is bad, tier 3. Now you know how I felt Sludging most of the time
Just goes to show, IMO, how important it is to find a deck you can resonate with, as the difference in winrate it can give you just for specializing it is bigger than the differences in winrates of two tier 1 or tier 2 decks.
EDIT: ofc there’s another reason, but it might hurt someone’s ego here. If you’re looking at the sample size on hsguru, it’s probably a lot bigger than yours, which explains the lower winrate. If you have 51-50, that’s 50.49% winrate. If you then win one, and then lose one, it’s now 50.48. That’s the beauty of the sample size. The more you play, the worse you get, according to statistics.
EDIT 2: Nevermind, I just can’t stop posting essays. If Scr0tie was here, we’d be now having a discussion about the formula they’re using to calculate winrates and he would then “prove” that the formula hsguru uses doesn’t divide with the total number of games, so this statistical rule doesn’t apply to it, but that’s nonsense. If a formula doesn’t do that, it’s not a correct formula, and what I can’t prove by mathematics, I can prove with experience, as I used to rock 70+ winrates on Sludgelock and still get stuck in 100-s, seldom highrolling my way into top 70. This means that those who were above me, consistently kept a winrate even higher than that for a longer period of time, whenever that might might have been (not necessarily the same patch, but before in their career).
If your sample is lower than 60 or 80 games, your statistics are very bad. There is high Skewness in a game like this so 20 or 40 samples are very little to show any major pattern and I include the basic win rate itself.
In case you wonder how that number is proven, 60 or 80 games is when the win rates start stabilizing.
Top 1K win rates fluctuate so often that it’s hard to use that as an idea for what to expect there. As one deck gains in win rate, other decks pop up to combat it.
When you look at those win rates up there, you’re likely a day behind. Unless a deck is consistently winning up there, the win rate won’t be as accurate as it would be in the Diamond - Legend bracket.
This is why most people use D-Legend bracket for overall deck win rates.
It’s very sensitive what to use because it heavily relies on the age of the month but also of the patch. If both the patch/cards and month is new than you can relatively reliably judge that anything above diamond having a high win rate isn’t a bad deck but right now that the patch is old and the month not new either then the 5K and sometimes even the top 1K is a horrible indicator because plenty of people who play badly but play 24/7 can go top 1K or at least 5K.
The top 100 is a bit unstable and misleading because certain singular people counter certain singular people with specific tech cards but it’s almost always reliable to show high skill cap decks that are not low in winrate. A problem is that the stats sites rarely group the top 100 (they do up to 1K) so your best bet is the streamer data of D0nkey but those are often too low sample sizes (they rarely have the needed 60 to 80 samples).
