2 questions on ratings and evaluations

Wondering about a couple of things:

1) If someone is 100 ELO higher than you, what's the statistical likelihood they will beat you? e.g. 3:1 in their favour.

2) When an engine makes an analysis of a position, does it also factor in the time on the players' clocks? Time is a key factor in chess. Someone could be up in material or have a better position but have only 30 seconds left to make their moves in.

Thanks.

How would a chess engine know how much time you have on a clock?

I mean, say you're playing a timed game against an engine and you ask it to evalute the position. Then it would know.

Also you can have it analyse games that are time-stamped.

200 points is about 3-1

I don't know the specifics, and I haven't verified the below mathematically, but I believe that the score you should get against an opponent should follow a normal (Poisson) distribution around your rating.

If you know anything about statistics and probabilities then I think that a rating difference of about 400 points is supposed to approximate three standard deviations.  In other words, with a difference of 400 points, there should be about a 99.7% chance the higher rated player will win.

So that would make one standard deviation be around 130 pts.

What that means is that if you play a large number of games against a person rated 130 points lower than you, you should score roughly 68% against them.  Let's say in a 100 game match, your total score against them, adding up all wins as 1 pt and draws as 0.5 pts, should be around 68.

Engines don't factor in the clock situation.  If the engine is playing, you can tell it to take as much or as little time as you want, but that's about it

There is literally no engine evaluation that correlates positional and time analysis.  The engine is impartial and will evaluate the position based on the factors that are present on the board, same as how psychological play is frowned upon by the computer.  Only if the time given to analyze the position is decreased will the accuracy of the engine analysis be reduced.

This may explain #1

http://www.3dkingdoms.com/chess/elo.htm

This is interesting.

0 rating difference each player scores about 1/2
100 rating lower, you should score about 1/3
200 rating lower, you should score about 1/4
300 rating lower, you should score about 1/6

Is an easy way to remember.

Are you sure?  I'd think I was having a bad day if I lost 1 in 6 against a 1300 player.

Drawing 2 is counted the same as losing 1 and winning one.

Those fractions were just a quick way to remember. Here's more specific numbers:

So we see you're expected to score 5 out of 6 against someone rated ~270 points below you. Against someone rated 300 below after 10 games, you're expected to lose one and draw one (or draw 3). So your intuition is right that vs someone 300 points lower, losing 1 in 6 would be under performing for you.

I have got draws against people 400/ 300 Elo above me it's how you play if your mindset is I'm going to lose you will lose.

Very interesting, thanks guys