Partial answer
Let's start with:
Pawns
There are:
All of them, 8 for each side. No pawn was lost nor promoted.
Hence:
-
The black pawn from G7 captured something in order to land in F6. So gxf6
was a black move.
-
The white pawn from E2 captured something in order to land on F3. So exf3
was a white move.
-
The white pawn from H2 captured something in order to land on G4. I'll detail that later below.
-
The black pawn from B7 captured something in order to land on A2. It also walked a lot and outmanuevered the white pawns somehow, which is only possible if the capture happened before the B white pawns advanced but after the A white pawn advanced.
So, the possibilities are:
The first two or three black moves for that pawn are b6
(maybe), b5
and b4
.
And:
The last two were b3
and bxa2
or bxa3
and a2
.
That is:
Four or five black moves.
Also:
White did a4
or perhaps a3
and a4
before that black pawn captured something. One or two white moves here.
-
The white pawn from C2 captured something. It is either on B3 or B5 now, the other being the white pawn from B2. One of them also walked a few squares.
So, the possibilities are:
A: cxb3
, b4
, b5
, plus a b3
somewhere after the b4
. Total of 4 moves.
B: c3
, cxb4
, b5
, plus a b3
somewhere. Total of 4 moves.
C: c3
, c4
, cxb5
, plus a b3
somewhere. Total of 4 moves.
D: c4
, cxb5
, plus a b3
somewhere. Total of 3 moves.
-
The black pawn on D4 is easy: It is either d3
, d4
and d5
or d4
and d5
. No captures. Two or three moves.
Anyway:
There were 5 pawns capturing something (3 white capturing black and 2 black capturing white).
And what is missing:
-
Both black knights, the black queen, the white bishop on light squares, one white knight - 5 pieces.
Hence:
Each one of those pieces was captured by a pawn. No pawn captured twice. No other captures happened.
Further:
Each one of those must necessarily have been moved at least once in order to be captured. Otherwise, no pawn could reach them.
Also:
The black pawn on F6 couldn't capture the white bishop due to square's color mismatch. So it captured a knight, that needed at least three moves to get there. The other black pawn must have captured the bishop and this means that its last two moves were bxa3
and a2
, also that bishop must have moved at least twice.
Let's see more moves:
The black queen must had at least two moves before dieing. Further, at least two moves for the king-side black knight and at least three for the queen-side black knight.
There are some strangely placed pieces:
-
The black bishop on G1.
Why?
It could only come from H2 before it was blocked by a lot of white pieces. Hence Bg1
was a black move. Sometime before that, it also needed Bh2
and before that either Bh6
and Bf4
or Bg7
and Be5
. At least 4 moves (possible more because it could move around somewhere else before going to where it went).
-
The white knight.
Why?
It could only come from G3. Hence Ng3
was a white move and Nh1
after that. And there must be at least one other move before (Ne2
if it was exactly one).
-
The black bishop on H3.
Why?
It could only reach that before the white pawn on G4 reached that place. So, Bh3
was a black move before g4
was a white move and hxg3
was another white move before.
-
The white king.
So:
How the heck it got there? It would need at least 4 king moves to get there. But in fact, it was actually at least 5 since it needed to go around the pawn at F3.
-
The white rooks.
Which is:
At least 4 moves.
-
Other black pieces.
Which is:
3 king moves and 3 rook moves if there was no castling or castling and a bunch of other moves.
-
The white bishop.
So:
It is in the same diagonal from its home square, but there is an unmoved pawn blocking the way. So it must have moved several times around it to reach this place. At least 4 moves.
So far we have:
29 confirmed white moves and 25 confirmed black moves.
A curious note:
Those two rooks on G6 and G7 are on the opposite sides than what would be expected. Putting them there without mating the kings is complicated.
This is not enough to fully reconstruct what is going on, but I think that it is at least a good partial answer.
Hopefully this graph will be helpful:
That is...
A partial topological sort of which move must happen before some other move. By adding all the moves there, we then would need to just find which order from the starting nodes going down to the moves on the end nodes that works.