Which Statistical Data Could Shed The Most Light

[Willie Pep 229]

- - Primary motive then was to stay out of jail just like it for any cognitive citizen today.

Pretty rough engaging in the most natural sport in the world only to have the authorities, Holy Rollers, and intellectuals want to nail em up in a Jail cell with a lucrative fine as the get out of jail card.

Take away the incentive (money) and they will stop doing it.

The reformers saw boxing as male exploitation as prostitution was to women. By any means necessary is always the way of reformers.

Need an ********? Supply about to become limited; expect a price hike.

[The Old LefHook]

The real winner of all those NWS?
How many of the bouts between the great black fighters of yesteryear were truly 100% contested?

This might be impossible!

[The Old LefHook]

Ive weighed in on this before Lefty. I would like to know probabilities. As many as one could program in... based on every sanctioned bout ever (a computer could do this) what is the difference in probable success for a puncher during a 15 round fight versus a 12 round fight? Things like that... We could then tell when a fighter truly does something remarkable... and we can have a knowledge base of information that might help us actually predict 100% one day, and if we could... we could back test and actually maybe answer questions about "who would win?"

Imagine having a computer that could tell you who would win if Louis, in his prime, had a ham and egg sandwich yesterday, and fought Frazier in his prime... lol.

Probabilities are what I am interested in too. I don't know if what I settle on can be most conveniently cast in that form. I am thinking of something simple, like recording a bunch of KOs at random and seeing which round they occurred in, noting the standard deviation and watching for groupings. One could get a lot of information from that. Unfortunately, it seems like useless information to me. I want something I feel can be instructive. I also want something I can do without many BoxRec clicks, because they now have me intimidated. In fact, I haven't even been back to them yet to see if I CAN get onto their site since they unclicked me.

* * * * *

Now let us suppose a man was going to tackle your problem. What will those who know something about statistics accept as "randomly collected," and how large a sample space do they demand before taking it seriously? These are cogent questions.

Also, I need the help of Margie or someone else determining exactly which dates demarcate the two categories, since there have been many championship fights that were scheduled for fewer rounds than 12. In our own era I believe I can simply begin at Hagler/Leonard for 12 round championship fights. But I am not sure where to begin so that every championship fight up to Hagler/Leonard is guaranteed to be scheduled for 15 rounds.

If someone can answer that simple question, I will see if I can get onto BoxRec.

Is there a list of championship fights, rather than doing it piecemeal on BoxRec. I am not sure how many clicks I can get out of BoxRec anymore. I got hacked from or around Mexico City a few paces back. Things suddenly became difficult, even registering for sites that required registration was almost impossible. Still not out of those woods, but did manage to get a government agency to take my case individually. They will not fix me though. They are trying to fix the hackers. Ahem!

In fact, BoxRec seems like a poor if not impossible way to do this job. I need a preexisting list or lists to select randomly from. Compiling the list from BoxRec and fighters' records would certainly require more clicks than an unregistered individual can receive. I tried to register right after they unclicked me, but could not get registered. I should start writing down passwords.

I suppose that if anyone had such a list of championship fights though, they would simply do the job themselves. If you had such a list, you would already have done the job, wouldn't you?

[billeau2]

Probabilities are what I am interested in too. I don't know if what I settle on can be most conveniently cast in that form. I am thinking of something simple, like recording a bunch of KOs at random and seeing which round they occurred in, noting the standard deviation and watching for groupings. One could get a lot of information from that. Unfortunately, it seems like useless information to me. I want something I feel can be instructive. I also want something I can do without many BoxRec clicks, because they now have me intimidated. In fact, I haven't even been back to them yet to see if I CAN get onto their site since they unclicked me.

* * * * *

Now let us suppose a man was going to tackle your problem. What will those who know something about statistics accept as "randomly collected," and how large a sample space do they demand before taking it seriously? These are cogent questions.

Also, I need the help of Margie or someone else determining exactly which dates demarcate the two categories, since there have been many championship fights that were scheduled for fewer rounds than 12. In our own era I believe I can simply begin at Hagler/Leonard for 12 round championship fights. But I am not sure where to begin so that every championship fight up to Hagler/Leonard is guaranteed to be scheduled for 15 rounds.

If someone can answer that simple question, I will see if I can get onto BoxRec.

Is there a list of championship fights, rather than doing it piecemeal on BoxRec. I am not sure how many clicks I can get out of BoxRec anymore. I got hacked from or around Mexico City a few paces back. Things suddenly became difficult, even registering for sites that required registration was almost impossible. Still not out of those woods, but did manage to get a government agency to take my case individually. They will not fix me though. They are trying to fix the hackers. Ahem!

In fact, BoxRec seems like a poor if not impossible way to do this job. I need a preexisting list or lists to select randomly from. Compiling the list from BoxRec and fighters' records would certainly require more clicks than an unregistered individual can receive. I tried to register right after they unclicked me, but could not get registered. I should start writing down passwords.

I suppose that if anyone had such a list of championship fights though, they would simply do the job themselves. If you had such a list, you would already have done the job, wouldn't you?

The information and data is out there. AS you intimate, the trick is how it is organized.

[Ivich]

This might be impossible!

I should think it is but that wasn't the point.

[Ivich]

The information and data is out there. AS you intimate, the trick is how it is organized.

It needs a far more up to speed techno than this old fella to collate such info, I can barely turn the pc on.

[billeau2]

It needs a far more up to speed techno than this old fella to collate such info, I can barely turn the pc on.

With what is coming down the pike even the techies are going to have a time of it.

[them_apples]

If a man were to collect boxing data, which data might provide meaningful insights? Taking all suggestions.

We could break down some points of collection: with enough questions even a yes or no tally at the end might be able to give us some insight.

I would stay as far away from speed and power in the beginning, since both of those can be completely nullified based on other factors.

1) the experience the fighter has in big fights
2) the performance they give in big fights
3) The fighters performance in the later rounds
4) The fighters performance in early rounds
5) The fighters chin (a good chin can change everything when assessing a fight)
6) Is the fighter a brawler or boxer (this could be broken down into many parts)
7) Which styles did that said boxer perform best against
8) The betting odds going into the fight (psychological temperature of the day)
9) The weight class

This is why machine learning in the future will help so much, for example through rapid testing it could come to the conclusion than a hard punch going up against a chinny opponent gives the hard puncher a great chance over 12 rounds, even in the event that the chinny boxer has every advantage over the puncher in other categories. You could do this by testing tons of fight outcomes and adjusting the weights of each question until the machine gets its guesses right. obv easier said than done we would need tons of data some labelled by humans others from box recs general statistics.

For example, if we were to bet on who wins in the Patterson vs Liston matchup, lets say the fight hasn't happened. Well, Pattersons general fear of Liston actually overpowers all the other technicalities you might consider (Patterson is faster, fought in big events, relatively good defense and power for his small size). Marciano being the same size, all things considered wouldn't be scared of Liston based on the data we have of him (wether he wins or not, this question would hold more weight when determining an outcome.)

[Bundana]

If a man were to collect boxing data, which data might provide meaningful insights? Taking all suggestions.

From time to time you hear the argument, that the oldtimers had to be better conditioned - since they fought 10 and 15 rounders (and sometimes even 20 rounds) all the time!

Wouldn't it be interesting to test this theory? We could count all the fights the 1930 and 2000 groups took part in - and see how many fights scheduled for 4, 6, 8, 10, 12, 15, 20 rounds, can be found for each group.

However, this would be extremely time-consuming... as it involves more than 15.000 fights, that need to be counted and categorized. So probably not something anyone would be eager to take on?

[Willie Pep 229]

From time to time you hear the argument, that the oldtimers had to be better conditioned - since they fought 10 and 15 rounders (and sometimes even 20 rounds) all the time!

Wouldn't it be interesting to test this theory? We could count all the fights the 1930 and 2000 groups took part in - and see how many fights scheduled for 4, 6, 8, 10, 12, 15, 20 rounds, can be found for each group.

However, this would be extremely time-consuming... as it involves more than 15.000 fights, that need to be counted and categorized. So probably not something anyone would be eager to take on?

Just for thought sake . . .

Would we have use ratios not sums as it would seem, on any given night there will be more fight cards in 1930 then 2000.

Yes? NO?

I guess one could actually put those reasonable numbers together - what would it tell us?

Were there more fights in 1930 than 2000, World wide? In the USA?