Just Learning: PE Confusion

[Just Learning]

I'm a bit confused about PE (Playing Efficiency). It seems systems can score very high on BC and IC, but PE seems to lag far behind. If a system has a BC of 98% and a PE of 60%, does that mean you are betting correctly 98% of the time and playing correctly only 60% of the time?

That seems like a tough way to win...

[SammyBoy]

> I'm a bit confused about PE (Playing
> Efficiency). It seems systems can score very
> high on BC and IC, but PE seems to lag far
> behind. If a system has a BC of 98% and a PE
> of 60%, does that mean you are betting
> correctly 98% of the time and playing
> correctly only 60% of the time?

> That seems like a tough way to win...

I believe the highest possible PE score is 70, so I don't think your statement above is correct.

[Parker]

> I'm a bit confused about PE (Playing
> Efficiency). It seems systems can score very
> high on BC and IC, but PE seems to lag far
> behind. If a system has a BC of 98% and a PE
> of 60%, does that mean you are betting
> correctly 98% of the time and playing
> correctly only 60% of the time?

> That seems like a tough way to win...

That's not quite correct. BC is simply telling you when you have the advantage and should bet more. Most counts are pretty good at that.

PE is much more complicated, since it is a measure of how well the system tells us to play each hand. Suppose that you were using a computer that was tracking every single card played as a unique value, constantly analysing the results, and telling us how to play our hand based on that data. That "system" would have a PE of 100.

So, in order to get that we would keep to have a side count for every single card value. For example, if we are holding a 16, 5's are very good and 6's are very bad, but most systems count them the same. Or, if we have a 13, the 8 suddenly becomes very important, yet most systems don't count it at all.

This is why, as SammyBoy mentioned, the best that any single-parameter system can do is around 70.

Also note that the number is merely a comparison to computer-perfect play. It has nothing to do with hands won or lost.

[Just Learning]

> I believe the highest possible PE score is
> 70, so I don't think your statement above is
> correct.

Now I'm more confused. I thought I asked a question.

I've seen that multi-level single-parameter systems have PE's approaching 70% -- with multi-parameter systems going further. But my question is, does that mean you're playing your hands incorrectly 30%+ of the time?

[Just Learning]

> That's not quite correct. BC is simply
> telling you when you have the advantage and
> should bet more. Most counts are pretty good
> at that.

> PE is much more complicated, since it is a
> measure of how well the system tells us to
> play each hand. Suppose that you were using
> a computer that was tracking every single
> card played as a unique value, constantly
> analysing the results, and telling us how to
> play our hand based on that data. That
> "system" would have a PE of 100.

> So, in order to get that we would keep to
> have a side count for every single card
> value. For example, if we are holding a 16,
> 5's are very good and 6's are very bad, but
> most systems count them the same. Or, if we
> have a 13, the 8 suddenly becomes very
> important, yet most systems don't count it
> at all.

> This is why, as SammyBoy mentioned, the best
> that any single-parameter system can do is
> around 70.

> Also note that the number is merely a
> comparison to computer-perfect play. It has
> nothing to do with hands won or lost.

Thanks Parker.

That makes perfect sense. It also sheds some light on my second question about why PE is more important with fewer decks. Per your explanation, when holding a 16, the effects of removing 1 "5" from the deck is far more significant with only 4 of them in a single deck vs. 24 in a 6-deck shoe. Am I on the right track?

Thanks again for your help.

[Magician]

> I've seen that multi-level single-parameter
> systems have PE's approaching 70% -- with
> multi-parameter systems going further. But
> my question is, does that mean you're
> playing your hands incorrectly 30%+ of the
> time?

No, it means you are getting 70% of the possible gains from deviating from BS. It's not just a measure of whether or not you played a hand correctly but of how much expectation you gained (or lossed) by deviating when you did. Sometimes the correct play is only better by a tiny fraction, other times the difference is much greater.

I have not seen data on how often a counting system determines the correct play (i.e. the play with maximum EV based on cards unseen) but I think perhaps BS alone would be correct more than 70% of the time.