How early can you tell if a game will be a draw?

Is there a way to know with reasonable certainty at some point that a game will be a draw (given unlimited time for play like in daily)?

I know there are obvious examples of known draws but if you are in a less obvious endgame with still a couple of pawns each and a rook and a minor piece each then is there any way to know that trying to play out a conclusion (which could take weeks) is a waste of time?

Like assuming perfect play from both sides how early can it be known that a game will be a draw?

I think this can be repositioned as a general question: how to evaluate an end game? How to know whether an end game is advantageous to one side or another or level and how to know whether there is little chance of either side forcing a win. 

For me, it is hard to imagine being able to play good chess without being able to do this. Imagine you are towards the end of a middle game and you see the opportunity to exchange a few pieces and move into an endgame. Should you do it? It is impossible to make a rational decision without being able to evaluate an end game position. Against weaker players than me I win many games by going into endgame (often with the help of my opponent) that are so much in my favour. And I guess sometimes when I am playing stronger players the same thing happens to me where perhaps I do not understand the endgame sufficiently well. 

Over the years I have read several books about the endgame. I remember one called Test Your Endgame Ability by Speelman and Silman's Complete Endgame Course. Nowadays there are lots of youtube videos and even great videos on chess.com that focus on the endgame. 

In short, there is no magic answer to your question. Sometimes material is equal and the endgame is drawn and other times it is clear win for one side. At other times one is way down on material but with an easy win. The only way to get the knowledge that you want is to study endgames. My suggestion would be a combination of theory and practice. 

Fair enough. I guess it is too nebulous and complex a question to be hoping for an easy answer.

I will crack the books (or the youtube videos)

I read some interesting article recently about how Draw is the the most common result in top level play now (over 50%) and it is over 80% in top level correspondence.

I guess the better the level of play becomes the tougher it is to squeeze out a win.

Thanks for the thoughts guys.

I guess technical engine analysis is only half the game.

Here is the game that got me to thinking about this.

I felt I had a small advantage with an extra pawn (even with pawns equal before move 48 his were stacked) and bishop vs a knight

The good thing I found is that it seems that the engine will tell you afterwards at what point the game reached a position where no win could be forced.

Engine says my endgame advantage peaked around move 46 at +3.71 and then tapered away.

My fatal move seems to have been 51 Rh7+ which took me down from +1.34 to almost 0

The engine is great for endgame analysis I guess since the tree search is much smaller so the results are more conclusive.

Still not sure how I could have played it better from move 46 but will give it some thought.

I just discovered that you can put the PGN into analysis mode go to move 46 and then follow the engines recommended moves for each player from that point on.

That will help a lot.

Some people get more draws than others, likely based on several factors (opponents they play, openings used, [and perhaps most importantly] how "determined" they are to continue versus accepting a draw offer. Personally, I consider my "draw margin" somewhat high. Back to the forum though: there are some ways to "anticipate" a high draw likelihood in advance. The key is familiarizing yourself with endgame essentials. Even simple pawn endings can become complicated; familiarize yourself with a "winning" position versus a "losing one" or a "drawn one." Even subtleties such as King activity, can turn into a game-winning position versus a lost cause. Extremely far in advance, usually the pawn structure can look "drawish", or maybe opposite colored Bishops ending, and so on. Like anything else in chess, it is pattern-recognition, and practice. 

^Thanks KSK. Food for thought there and some good pointers for my research.

Clearly a draw.

Black has no weaknesses. If white tries to press with his one tempo advantage black should have enough counter play to equalise.

^Well given enough time the engine may prove you right there Pyotr but I think that technical conclusion will not spoil the game of chess for us mere humans.

You got all wrong conclusion with engine analysis, i am afraid. I checked with my stockfish

1.  47.Both Bd3 and Bb3 leads to win (there is no mistake)

2. 48. bxf7 was major mistake(taking useless pawn that lead to instant draw---> from win)

3. 51. Rh7 or Bd5 both leads to dead draw, no matter what engine evaluations said( you can analyse in deeper line there is no progress, so Bd5 showing1.34+ something is useless). The position is already dead draw.

There is a long standing running joke, that everyone knows.  If you get paired with me in a 1 day tournament, and its the round right before lunch?  If you offer me a draw EARLY, im going to accept it 99.99% of the time.

the start position is a draw if both players are the same

Thanks DRMBR

I also looked at it with stockfish and there seemed to be a clear path to getting the rook and a decent +6.75 lead

47. Bd3 Rd4 48. Bb5 Nf8 49. Re2+ Kd6 50. Kb6 Rd1 51. a6 Ra1 52. a7 Rxa7 53. Kxa7

My bishop move on 47 was the mistake.

I thought the engine doing the full game analysis was also stockfish. Must be using simpler parameters though for the quick analysis as it steps through the moves quicker than stockfish in manual analysis mode.