A few thoughts/ questions on O Hagan''s law here. (And why not?)
The first version I heard was (something like): "If you have nine market losers
and no terrible rolls, then double."
Then somehow, this got translated into "If the probability of losing your market during the next exchange is >= 25% then double (assuming that
there aren't horror sequences to cancel out the market losers)."
Is there any basis for this translation? Assuming that O' Hagan said
the remark about "nine market losers" (and I'm not familiar with the original source), isn't it possible that he meant what he said?
Presumably the law is derived from experience. The sample set
where you have >= 9 market losers is different to the sample
set where you have 25% market losing sequences.
From experience with XG, I think that the figures of 9 or 25% are way
too high. 6 or 17% crushing threats is usually enough.
Another point is that this animal never exists in practice. You just don't get positions where 9 rolls are massive market losers and the other
27 are so uneventful, they resemble the experience of queuing up
at your nearest supermarket to buy toilet rolls, shampoo, carrots and spinach.
As he's an intellectually curious mathematician, the bg world was
interested but not exactly shocked or stunned when Tim questioned
the O'Hagan figure, and wondered whether it had a solid scientific basis. (As if anything anyone says is likely to be verifiable or scientific.)
But it shouldn't be particularly difficult to arrive at a meaningful O'Hagan type statistic. So I'll outline that as follows. The answers wll probably
be different depending on whether it's a recube or not.
First decide how to represent the "market losers". In practice, this is likely
to mean obviously crushing sequences or rolls.
So let's assume that these rolls give us an equity of 90% on the upturned cube but are certain to lead to D/P if we don't cube. The other rolls are assumed to be nullo rolls (in Robertie-speak).
However, the opponent will have the cube so we need to give a value to that. Let's assume that the cube is worth 0.3 to the opponent.
Let's also assume that the cube started out in the middle.
Let's also try out the traditional O'Hagan rule so assume 25% market losers. The non-cube decision gives us an equity of 0 * 75% + 25% * 1 = 0.25.
The cube decision gives us an equity of 2 * - 0.3 * 75% + 2 * 0.9 * 25% =
0.
So don't double.
Paul
On Saturday, August 28, 2021 at 8:48:18 AM UTC-4, peps...@gmail.com wrote:us). In fact, a lot of times it would be something we didn't even think about when creating the roll because it's a non issue (to us). So over time things do get ironed out and worded more accurately for everyone.
A few thoughts/ questions on O Hagan''s law here. (And why not?)
The first version I heard was (something like): "If you have nine market losers
and no terrible rolls, then double."
Then somehow, this got translated into "If the probability of losing your market during the next exchange is >= 25% then double (assuming that
there aren't horror sequences to cancel out the market losers)."
Is there any basis for this translation? Assuming that O' Hagan said
the remark about "nine market losers" (and I'm not familiar with the original source), isn't it possible that he meant what he said?
Presumably the law is derived from experience. The sample set
where you have >= 9 market losers is different to the sample
set where you have 25% market losing sequences.
From experience with XG, I think that the figures of 9 or 25% are way
too high. 6 or 17% crushing threats is usually enough.
Another point is that this animal never exists in practice. You just don't get positions where 9 rolls are massive market losers and the other
27 are so uneventful, they resemble the experience of queuing up
at your nearest supermarket to buy toilet rolls, shampoo, carrots and spinach.
As he's an intellectually curious mathematician, the bg world was interested but not exactly shocked or stunned when Tim questioned
the O'Hagan figure, and wondered whether it had a solid scientific basis. (As if anything anyone says is likely to be verifiable or scientific.)
But it shouldn't be particularly difficult to arrive at a meaningful O'Hagan
type statistic. So I'll outline that as follows. The answers wll probably be different depending on whether it's a recube or not.
First decide how to represent the "market losers". In practice, this is likely
to mean obviously crushing sequences or rolls.
So let's assume that these rolls give us an equity of 90% on the upturned cube but are certain to lead to D/P if we don't cube. The other rolls are assumed to be nullo rolls (in Robertie-speak).
However, the opponent will have the cube so we need to give a value to that.
Let's assume that the cube is worth 0.3 to the opponent.
Let's also assume that the cube started out in the middle.
Let's also try out the traditional O'Hagan rule so assume 25% market losers.
The non-cube decision gives us an equity of 0 * 75% + 25% * 1 = 0.25.
The cube decision gives us an equity of 2 * - 0.3 * 75% + 2 * 0.9 * 25% = 0.
So don't double.
PaulI don't exactly follow all of what you're saying/asking and one inherent issue with these 'rules' is that better players tend to know what the rule means at its root so if it isn't worded precisely or with every little deviation, it doesn't matter (to
I did a group lesson once upon a time on O'Hagan's Law and it's a very useful tool. Without going back and referencing my own material I'd say O'Hagan's Law is knowing that you have a double on a centered cube when you lose your market a fourth of thetime on the upcoming sequence. (Your turn + your opponent's) For a redouble you need 11/12 MLS. You should also note and adjust accordingly if there are any market crashers. (if you have 1 market crasher you need 1 more MLS ffor eg). The size of the MLS
If you search for O'Hagan's Law on my forums you'll likely find a lot of examples to test yourself with and also perhaps a nugget of genius such as:
http://www.bgonline.org/johno.wmv
On Sunday, August 29, 2021 at 12:55:59 AM UTC+1, J R wrote:to us). In fact, a lot of times it would be something we didn't even think about when creating the roll because it's a non issue (to us). So over time things do get ironed out and worded more accurately for everyone.
On Saturday, August 28, 2021 at 8:48:18 AM UTC-4, peps...@gmail.com wrote:
A few thoughts/ questions on O Hagan''s law here. (And why not?)
The first version I heard was (something like): "If you have nine market losers
and no terrible rolls, then double."
Then somehow, this got translated into "If the probability of losing your
market during the next exchange is >= 25% then double (assuming that there aren't horror sequences to cancel out the market losers)."
Is there any basis for this translation? Assuming that O' Hagan said
the remark about "nine market losers" (and I'm not familiar with the original source), isn't it possible that he meant what he said? Presumably the law is derived from experience. The sample set
where you have >= 9 market losers is different to the sample
set where you have 25% market losing sequences.
From experience with XG, I think that the figures of 9 or 25% are way too high. 6 or 17% crushing threats is usually enough.
Another point is that this animal never exists in practice. You just don't
get positions where 9 rolls are massive market losers and the other
27 are so uneventful, they resemble the experience of queuing up
at your nearest supermarket to buy toilet rolls, shampoo, carrots and spinach.
As he's an intellectually curious mathematician, the bg world was interested but not exactly shocked or stunned when Tim questioned
the O'Hagan figure, and wondered whether it had a solid scientific basis.
(As if anything anyone says is likely to be verifiable or scientific.)
But it shouldn't be particularly difficult to arrive at a meaningful O'Hagan
type statistic. So I'll outline that as follows. The answers wll probably
be different depending on whether it's a recube or not.
First decide how to represent the "market losers". In practice, this is likely
to mean obviously crushing sequences or rolls.
So let's assume that these rolls give us an equity of 90% on the upturned
cube but are certain to lead to D/P if we don't cube. The other rolls are assumed to be nullo rolls (in Robertie-speak).
However, the opponent will have the cube so we need to give a value to that.
Let's assume that the cube is worth 0.3 to the opponent.
Let's also assume that the cube started out in the middle.
Let's also try out the traditional O'Hagan rule so assume 25% market losers.
The non-cube decision gives us an equity of 0 * 75% + 25% * 1 = 0.25. The cube decision gives us an equity of 2 * - 0.3 * 75% + 2 * 0.9 * 25% =
0.
So don't double.
PaulI don't exactly follow all of what you're saying/asking and one inherent issue with these 'rules' is that better players tend to know what the rule means at its root so if it isn't worded precisely or with every little deviation, it doesn't matter (
the time on the upcoming sequence. (Your turn + your opponent's) For a redouble you need 11/12 MLS. You should also note and adjust accordingly if there are any market crashers. (if you have 1 market crasher you need 1 more MLS ffor eg). The size of theI did a group lesson once upon a time on O'Hagan's Law and it's a very useful tool. Without going back and referencing my own material I'd say O'Hagan's Law is knowing that you have a double on a centered cube when you lose your market a fourth of
If you search for O'Hagan's Law on my forums you'll likely find a lot of examples to test yourself with and also perhaps a nugget of genius such as:Thanks for your thoughts. From my experience of XG, I'm sure that with blitzing threats in the opening, you need far fewer than 9 MLS
http://www.bgonline.org/johno.wmv
or 25%, presumably because of the large number of gammons.
Paul
Sysop: | Keyop |
---|---|
Location: | Huddersfield, West Yorkshire, UK |
Users: | 292 |
Nodes: | 16 (2 / 14) |
Uptime: | 205:47:24 |
Calls: | 6,618 |
Calls today: | 2 |
Files: | 12,168 |
Messages: | 5,316,743 |