Open Side Menu Go to the Top
Register
quantifying/detecting tilt quantifying/detecting tilt

06-05-2015 , 01:32 PM
What measures might one use if you wanted to quantify/detect tilt in a poker player utilizing any of the stats found in typical poker software? I was thinking one easy way would be to try to correlate agression factor, % of hands played with recent up/down swings.

I was also thinking that one might be able to tell the degree of tilt solely based on money won according to time. Downswings should be longer and more dramatic than is statistically likely.
quantifying/detecting tilt Quote
06-05-2015 , 03:10 PM
Quote:
Originally Posted by imfatandugly
What measures might one use if you wanted to quantify/detect tilt in a poker player utilizing any of the stats found in typical poker software? I was thinking one easy way would be to try to correlate agression factor, % of hands played with recent up/down swings.

I was also thinking that one might be able to tell the degree of tilt solely based on money won according to time. Downswings should be longer and more dramatic than is statistically likely.
This probably won't work.

* There's probably too much noise to even detect a change in the first place. At any particular moment, you know your EV to maybe +-1 BB/100 (based on table factors and such). Tilting may change your EV by less than that amount.
* Tilt is not a binary on/off event.
* Tilting may actually *increase* your EV in the short term if your opponents aren't adjusting to your change of style.
* Not all tilt is the same. Your aggression may go up or it may go down.
quantifying/detecting tilt Quote
06-05-2015 , 03:24 PM
This is why I play live. It's so much easier to "feel" other people going on tilt.
quantifying/detecting tilt Quote
06-05-2015 , 03:33 PM
Quote:
Originally Posted by imfatandugly
... I was thinking one easy way would be to try to correlate agression factor, % of hands played with recent up/down swings.
...
If you try this approach, you have to take great pains to ensure that hands that contributed to "recent winnings" don't contribute to aggression.

If you look at the AF of players stratified by hands they win or lose, I would completely expect there to be a statistically valid relationship. There is a causal relationship between hand strength and AF, and a causal relationship between hand strength and winnings.

(e.g., on the river, the player with the nuts is always aggressive)

That being said, I think you will find something interesting, but I don't think it will be tilt.
quantifying/detecting tilt Quote
06-06-2015 , 12:45 PM
Quote:
Originally Posted by Aaron W.
This probably won't work.

* There's probably too much noise to even detect a change in the first place. At any particular moment, you know your EV to maybe +-1 BB/100 (based on table factors and such). Tilting may change your EV by less than that amount.
* Tilt is not a binary on/off event.
* Tilting may actually *increase* your EV in the short term if your opponents aren't adjusting to your change of style.
* Not all tilt is the same. Your aggression may go up or it may go down.
standard deviation will probably change as well. I believe this converges much more quickly than winrate.
quantifying/detecting tilt Quote
06-06-2015 , 12:48 PM
Quote:
Originally Posted by FoldnDark
This is why I play live. It's so much easier to "feel" other people going on tilt.
I can often tell if someone is tilting online. I think the biggest give-aways are increased speed of play and repeated exclamations of the hope I die in some horrible fashion.
quantifying/detecting tilt Quote

      
m