Elephant in the room

[Jersey #10]

I am reminded of this disastrous thread of Rett http://indiancricketfans.com/showthread.php?t=261363

[Jersey #10]

If you need a complicated system to do that be my guest :winky: Obvioulsy, there are various factors but when we talk about proven performers, it's not difficult to say, for example, if Gavaskar had played 150 tests, he would have got 12k runsThose who don't like Gavaskar would probably say what if he failed to go past 50 runs in any test after is 125th test. That's a probablilty too .... But we have our senses to know which is more probable, i.e. Gavaskar reaching 12k runs

It's very very difficult to say that! That's like saying, if Abhishek had lasted a few minutes more or less, Aish would have had a son. There is no concrete back up for such foolish asumptions.

[Jersey #10]

I won't even read the rett's recent essay since Outsider is on the case!

[ganeshran]

If you need a complicated system to do that be my guest :winky: Obvioulsy, there are various factors but when we talk about proven performers, it's not difficult to say, for example, if Gavaskar had played 150 tests, he would have got 12k runs Those who don't like Gavaskar would probably say what if he failed to go past 50 runs in any test after is 125th test. That's a probablilty too .... But we have our senses to know which is more probable, i.e. Gavaskar reaching 12k runs

Extrapolation of data to make a prediction in the future is not a simplistic product of the runs per test * the number of matches to be played in the future. More the number of tests, more the margin of error will be. In a short range like 5-10 tests, the actuals might match the predicted score, but on higher ranges like 50 tests, such predictions can be grossly inaccurate. Till his 110th test match, Ricky Ponting had scored 9368 runs at a runs per test of 85. After 45 test matches. He should have reached a total of 13200 in his 155th match. He is currently at 12495 runs which is a lesser by 700 runs than his predicted total. There would be 100s of such examples provided you apply your metric to existing data which can be verified, as opposed to making a prediction in the future which cant be verified.

[Magneto]

no' date=' not 'rpt'.[/quote'] That was exaggeration on purpose. :headshake:

[akshayxyz]

That was exaggeration on purpose. :headshake:

ok...:embarrassed:

[zen]

Extrapolation of data to make a prediction in the future is not a simplistic product of the runs per test * the number of matches to be played in the future. More the number of tests, more the margin of error will be. In a short range like 5-10 tests, the actuals might match the predicted score, but on higher ranges like 50 tests, such predictions can be grossly inaccurate. Till his 110th test match, Ricky Ponting had scored 9368 runs at a runs per test of 85. After 45 test matches. He should have reached a total of 13200 in his 155th match. He is currently at 12495 runs which is a lesser by 700 runs than his predicted total. There would be 100s of such examples provided you apply your metric to existing data which can be verified, as opposed to making a prediction in the future which cant be verified.

As far as I see, rpt did a good job of telling where Ricky would be at 155th test, when he had just played his 85th .... it would say that such small variances are acceptable and everyone understands what 'forecasts' stand for It's comical to see 'forecasting' being tried to turned into an exact science. We all know forecasting cannot be always 100%. That doesn't mean we don't forecast. We have to whether it is in business, life, weather, etc For example, someone could forecast his annual income, is there any guarantee that he will make that. There could be an act of God, or some recession or whatever. But doesn't mean we don't forecast. We are treading on moot points now

[Jersey #10]

If you need a complicated system to do that be my guest :winky: Obvioulsy, there are various factors but when we talk about proven performers, it's not difficult to say, for example, if Gavaskar had played 150 tests, he would have got 12k runs Those who don't like Gavaskar would probably say what if he failed to go past 50 runs in any test after is 125th test. That's a probablilty too .... But we have our senses to know which is more probable, i.e. Gavaskar reaching 12k runs

there are no complicated systems, they are either right ways to do it or wrong ways to do it and then there are better/faster ways of doing the right stuff. You've got it all wrong so don't worry about the complicated systems, continue with your knowledge of 4th grade math.

[Jersey #10]

As far as I see, rpt did a good job of telling where Ricky would be at 155th test, when he had just played his 85th .... it would say that such small variances are acceptable and everyone understands what 'forecasts' stand for It's comical to see 'forecasting' being tried to turned into an exact science. We all know forecasting cannot be always 100%. That doesn't mean we don't forecast. We have to whether it is in business, life, weather, etc For example, someone could forecast his annual income, is there any guarantee that he will make that. There could be an act of God, or some recession or whatever. But doesn't mean we don't forecast. We are treading on moot points now

ok how about this? can you tell me where Ricky Ponting will be after 157 tests? Your error window is +/- 2 runs since it's such a small window of matches.

[ganeshran]

As far as I see' date=' rpt did a good job of telling where Ricky would be at 155th test, when he had just played his 85th .... [b']it would say that such small variances are acceptable and everyone understands what 'forecasts' stand for It's comical to see 'forecasting' being tried to turned into an exact science. We all know forecasting cannot be always 100%. That doesn't mean we don't forecast. We have to whether it is in business, life, weather, etc For example, someone could forecast his annual income, is there any guarantee that he will make that. There could be an act of God, or some recession or whatever. But doesn't mean we don't forecast. We are treading on moot points now

The variances are not small, they can be huge and the probability of error keeps on increasing as you increase the number of matches. Lara's 72nd match runs per test suggests that he would have scored 14500 runs had he played 180 tests. It also suggests that he would have scored only 10600 runs in 131 tests which is also grossly inaccurate. Forecasting has never been an exact science. The problem is that you are using a forecast based on runs per test (with its inherent inaccuracies) as a verifiable fact to assert that Lara would have outscored SRT had he played 180 tests. The probability of Lara outscoring SRT depends on the prior probability of Lara's rpt prediction being true which itself is uncertain.

[zen]

The variances are not small, they can be huge and the probability of error keeps on increasing as you increase the number of matches. Lara's 72nd match runs per test suggests that he would have scored 14500 runs had he played 180 tests. It also suggests that he would have scored only 10600 runs in 131 tests which is also grossly inaccurate. Forecasting has never been an exact science. The problem is that you are using a forecast based on runs per test (with its inherent inaccuracies) as a verifiable fact to assert that Lara would have outscored SRT had he played 180 tests. The probability of Lara outscoring SRT depends on the prior probability of Lara's rpt prediction being true which itself is uncertain.

Again the variance is acceptable. 14,500 runs is close to what Tendulkar has now. The point is that Tendulkar has more runs than others because he played more tests (and having 15k runs in itself doesn't make him the greatest) still stands. Like you pointed out, even a number that you think is off suggests that a player of similar calibre would have got to that number if he had got the 'opportunity' to play that many games .... and since nobody actually gets paid for devising a forecasting system, the range that rpt gives is more than acceptable All the digging up that you are doing only leads to the conclusion that I gave a long time ago. It is just that you are keen to making the journey by yourself to get there rather than use 'short cuts' (things I said a long time ago) Thanks!

[akshayxyz]

I am reminded of this disastrous thread of Rett http://indiancricketfans.com/showthread.php?t=261363

Good find. Looks like it is natural for readers to start thinking about more parameters as 'Ram' did here http://indiancricketfans.com/showpost.php?p=1315218&postcount=114 and I did in this thread. http://www.indiancricketfans.com/showpost.php?p=1706488&postcount=61 btw - here is another genius, who could predict w/o even using any data related to cricket...using some complex mathematics, that he can not teach, others can not understand. http://indiancricketfans.com/showthread.php?t=272367&highlight=numerology

[akshayxyz]

Again the variance is acceptable. 14' date='500 runs is close to what Tendulkar has now. The point is that Tendulkar has more runs than others because he played more tests (and having 15k runs in itself doesn't make him the greatest) still stands. Like you pointed out, even a number that you think is off suggests that a player of similar calibre would have got to that number if he had got the 'opportunity' to play that many games .... [b']and since nobody actually gets paid for devising a forecasting system, the range that rpt gives is more than acceptable All the digging up that you are doing only leads to the conclusion that I gave a long time ago. It is just that you are keen to making the journey by yourself to get there rather than use 'short cuts' (things I said a long time ago) Thanks!

What kind of logic is that? IFFFFFF you can devise a forecasting program - for any popular events/sports of public interest, that can predict within some 'acceptable' margins of error - you would be millionaire the next day - given that it should be acceptable to people who have the money.

[zen]

Appears as if we have two parallel tracks going 1. rett commenting on his opinion 2. some folks commenting on rett (for some "unknown" reason :P ) and trying to console each other on how right they are :giggle:

[zen]

What kind of logic is that? IFFFFFF you can devise a forecasting program - for any popular events/sports of public interest, that can predict within some 'acceptable' margins of error - you would be millionaire the next day - given that it should be acceptable to people who have the money.

^reading comprehension problems What was implied was that for a free system, rpt is pretty good .... so no need to treat it like some advance system for which people are getting paid for

[maniac]

Whoa....this mindless discussion is still going on....

[ganeshran]

Again the variance is acceptable. 14' date=500 runs is close to what Tendulkar has now. The point is that Tendulkar has more runs than others because he played more tests still stands. Like you pointed out, even a number that you think is off suggests that a player of similar calibre would have got to that number if he had got the 'opportunity' to play that many games .... and since nobody actually gets paid for devising a forecasting system, the range that rpt gives is more than acceptable Pompous statements like these are very amusing.

That is a logical fallacy. You are comparing Lara's projected score to what Tendulkar has currently and concluding that the variance in the forecasting system is small. That should be compared to actual data - the runs scored by Lara in 131 tests - which is an error of 1300 runs from the predicted data when the time period is in the range of 60 tests. Its pretty obvious that runs/test is useless for both comparing as well as forecasting a player's future performance and not too different from a poster on ICF who said Harbhajan Singh is better than Akram because he has a higher number of wickets per test. Saying that SRT has more runs than everyone else because he played more tests, is very different from saying that anyone who gets to play that many number of tests would get to SRT's tally. To prove the latter you again need to first prove the fact that the rpt metric is infallible or atleast has low error rates which has clearly not been proved yet. In other words you are basing your conclusion on an original premise which itself is not watertight.

[Yoda-esque]

Ganesh & Akshay : I thought you guys would know better :headshake: Rett is WUMMINg around now and you guys are falling for his bait.

[akshayxyz]

Whoa....this mindless discussion is still going on....

There is no discussion. Rett is slaughtering ICFers one by one. Honestly, I am converted to rpt club.

[akshayxyz]

Ganesh & Akshay : I thought you guys would know better :headshake: Rett is WUMMINg around now and you guys are falling for his bait.

Thanks for the word - now I know the word for what I myself do sometimes :P But it is hard to believe, esp, after he wrote that loooooong post on rpt, and then further replied to Outy's reply on that.