Stunning Statistic in Chess.com Opening Explorer

In the following line standard line:  1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 Nc6 the piece hanging move 6. Bd3 (allows 6...Nxd4, winning a knight for seemingly nothing) has been played 170 times according the Chess.com Opening Explorer.  And even more strange is apparently the hanging knight on d4 has only been captured one time!

This is classified as the Sicilian Defense:  Open, Classical, Yates Variation.  There does not seem to be much information regarding this line.  And due to not being able to see the "soundness" of 6. Bd3 by looking at the position I bit the bullet and checked position with the Chess.com engine (Stockfish 16) which evaluates the knight capture 6...Nxd4 at -2.26 at a depth of 20, a clear advantage for black. 

Can anyone shed any light on this statistical oddity?  And by the way I am not checking this (via engine or this forum) due to a game of mine or anyone else's in progress nor did I check this (via engine) while a game was in progress.  I literally am asking for a friend.

Thanks!

It's almost certainly a move transposition ending in the resultant position. Explorer is not move dependent. If you check any of the games you'll see that the move order to reach the position is different.

For whatever it is worth the move order 1. e4 c5 2. Nf3 Nc6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 d6 6. Bd3 (black moves #2 and #5 were transposed from the line in the original post) yield the exact same results. I would provide a screenshot of both but I gather there is not much even you could do to explain this.

There are only 2 games in the database where that exact position exists:

• https://www.chess.com/games/view/15355551 - here the knight was taken
• https://www.chess.com/games/view/633047 - here it wasn't but it's possible that the game score has an error in it because the knight is shown hanging for a number of moves.

In the lines in Explorer giving 6. ... e6, it's ones where the previous move was Nc6 and white moves the knight after it's threatened:

• https://www.chess.com/games/view/16143506
• https://www.chess.com/games/view/16085928

The same appears to be true for the other options as well. The Explorer database is not move aware dependent. It is only looking at positions where the exact position exists after the given "move" has all the pieces on the board in the correct positions.

As far as I'm aware a lot of chess databases work the exact same way. They are concerned with finding matching positions because it's very useful to see all games that reached that position, regardless of the move order that resulted in the position.

You do know that you just bumped a 2-year-old thread about this?

Yes, because a friend sent me a link to it.

Since there are not too many threads on this I figured I would use that forum as well to reach out.

However anyone chooses to spin what exactly the database is showing or trying to show the bottom line is I read 6...Nxd4, winning a free piece, only played 1 out of 170 opportunities, and even for two different move orders. So it is a good thing I rarely rely on the Opening Explorer and know to go over positions in opening databases and not solely rely on statistics. Hopefully everybody else does the same.

It was only played once because it wasn't possible in any other games (other than that weird second game which is probably wrong anyway). If you don't believe me, just take a stroll through all the games from any of the resultant positions.

why??????

i mean, the "all games with a given position are shown" thing makes sense, when applied to the position on the board

but when it is applied to the "popularity" of a move...

we get stuff like after e4 f5, exf5 (the obvious best move) is way down

showing e4 f5 d4 games when searching for d4 f5 e4 games makes sense

showing d4 as being played 100ish times when most of those are d4 f5 e4 games makes little sense

The Explorer database is just a percentage of times the position was reached for any games in the database. That is the position on the board with the addition of the next move.

In a database a lot of options are going to get heavily skewed to old games, so popularity may or may not be a good description. If you use something like ChessBase on a large database, it will let you know the same information but will also let you know if the positions are played anymore/more recently.

Yes, it makes perfect sense. The positions are the exact same so seeing the continuations. The important part is understanding what the database is actually showing you. Once you realize it's about positions, not lines, you know what to look out for.

Most databases work in a very similar way, unless they have a specific option to only look at move order, which can result in missing out on some interesting games that reach the same position from a different move order.

No. Ignore the move order. Just look at positions. The database is not move order aware, just positions. It's a position database.

So, in that example, it's showing all the positions that match from a different move order, though in many cases it can also be from the given mine order as well.

misunderstanding me

i am TOTALLY FINE when we input e4 f5 d4 and get d4 f5 e4 games

where is disagree is, that the chess.com database shows d4 as having been played 100ish times after e4 f5 (and this has to be the only reasonable assumption here) and this is the problem op gives

I'm not sure I'm explaining this correctly. The list on the right with moves are not necessarily moves.

Those are positions. So when you have this position:

Explorer shows there are 842 games where where this position exists in the database when there is a pawn on d4 as well

If you look at those games, you'll find that almost all, if not all of the games, came from the move order 1. d4 f5 2. e4 ... Again, the Explorer database is just showing positions and does not look at the move order, at all.

that to me, and i assume many others, looks like d4 was played 842 times

no idea why a database would do this

Because it's a database doing something different than what you think it is. It's only purpose it to find all positions where the current position plus the change based on the given move matches.

That's it. It's a positional database. It will also have positions reached by a given line of play, but that is not all it can do.