BuzzerBeater Forums

I want to comment on what someone else said, that we were one result away. Technically yes, but beating spain by 46 on the road is no way to get there. If you need to win all of your games, your best approach is to try to keep the odds of winning each game similar to eachother if it's a linear relationship. So better to have a 50-50 chance in each game than 10-90 in one and 90-10 in the other (the odds of winning both games in the first scenario is 25%, the 2nd scenario it's only 9%). I feel Juice violated this rule, among countless other mistakes.

... snip ...

Again this is relative. How much of a gain would it have been to play Peru vs Brasil in odds of winning? you can set up a fairly simple decision tree, and usually it will yield the best answer.

Can you walk us through the decision trees as you saw them at the time (or as you would see them "at that time" now)? What sort of "level of trying to win" would you have put into the various games in our final Worlds group last time through? If you knew ahead of time that Brazil would CT us, would you have counter-CT'd and actively tried to win the game? Would you have intentionally TIEd and thrown it (again, knowing ahead of time they're going to CT)?

Granted, it's Monday Morning Quarterbacking, but I'm interested to hear your opinion a posteri, since you seem pretty adamantly opposed to the methods that were used at the time.

So lets take a look back at the start of season 10.

ok, so lets mmqb, which is unfair to coco because predicting a brazil crunch time is unreaspnable, but for academic purposes lets pretend instead of brazil CT'ing its italy or spain or something.

Argentina had initially beaten us by 14, with what i am going to say was possibly better enthusiasm the first time and putting more into the game.

In the semi finals, argentina clearly had a rather sizable enhtusiasm edge, only peru in their three games leading to the semifinals might have been a normal.

so they are still probably favorites because we know they took it easy last game and likely have an enthusiasm edge.

so lets put our odds at beating argentina at 30% if we normal the game against brazil, 50% if we TIE, and 5% if we CT and lose



we also know that peru had taken it easy the game before, but were a beatable, albeit good team. lets give ourselves a

70% chance if we tie the game before and face them, 50% chance if we normal, and 20% chance if we ct the game before and lose


so now all we need are the odds of beating brazil, knowing they are CT'ing. as it stands we only lost by 7. this seems about right. we beat them earlier that season.

lets put the odds at 70% of winning if we ct, 40% if we normal, 10% if we tie



so now.


if we ct, there's a 70% chance we face peru with wasted enthusiasm, and 30% we face argentina

i count our odds of winning the semis on aggregate at 15.5%


if we normal, there's a 40% chance of peru, and a 60% chance of argentina

our aggregate odds are now 38% of winning the semis game

if we tie we face argentina 90% of the time and peru 10% of the time

i put our odds that way at 52% of advancing


the #'s could be slightly off, so there may be a situation where normaling was the right call or close to it, but generally speaking, ct would be a fiasco, and TIE is probably best against brazil if we know they are shooting their load

and as promised, i am not talking about juice's failures anymore this campaign, so i will discuss those after the election, preferably over email, but you can see my general thinking in the u21 discussion. all told coco didnt know brazil would CT, so i find his decision to normal the game defensible

At the risk of sounding impertinent... what do you base your percentages on? Can you at least somewhat justify the 50/30/5% chances against Argentina? To me, your post pretty much looks like randomly chosen numbers... I'd like to have an idea of the basis from which you're generating them.

At the risk of sounding impertinent... what do you base your percentages on? Can you at least somewhat justify the 50/30/5% chances against Argentina? To me, your post pretty much looks like randomly chosen numbers... I'd like to have an idea of the basis from which you're generating them.

Despite being beaten down constantly URP is consistently showing in depth his thought process (rightly or wrongly)... this debate seems to be stacked against him (maybe a lot of his own doing initially) but if you look beyond this his tactical arguments seem far more convincing than others who have opted to perhaps rely on the fact they've devoted more man hours to the cause previously.

I'm curious to see how many votes URP takes down in the end... I feel he is fighting a losing battle unless interestingly the votes for yourself and wozzvt come out pretty even.

As a neutral I would be coming around more to this campaign and feel that this guy has to be worth a shot. You can almost forgive his worryingly poor club record as with NT you just get handed some top players to go out and play with.

EDIT: fixed crap spelling



not random.

Warning: this is wonky, so if you're not at least competent with introductory level college statistics, this probably won't make sense.

-----------------------------

if you play the same tactics against the same team 10000 times, your expected score might not fall ina perfectly normal distribution, but it isnt super far off

more importantly, your points per possession number should be very close to normally distributed

your opponents points per possession number should also be normally distributed, though again their score is not fully independent of your score, but the two are less correlated tan you would think (the issues come at the end of the game when the computer coach starts "adjusting" with occasionally sub-optimal strategies)

All of that is fairly obtuse anyway, so let's oversimplify because the oversimplification is more or less true enough to apply here.

With normal paces and adjusting for talent of opposition, the standard deviation for most decent teams is +/-10 points. (10 isnt the exact number but it's close enough that this explanation should make sense, without being so close that I would feel uncomfortable about some of the work i've done (brute forcing alot of games to find standard deviations no less))

This matters. With this information, and a normal distribution, and a similar deviation from the opponent, we know that, using basic math, that the standard deviations in score between the two games is:

10 squared + 10 squared = 200, but take the square root for lets call it 14.


So if you are a 14 point favorite, you are going to win that game 84% of the time. That doesnt sound like much and sounds like you are very prone to upsets but thats reallly a dominant number, and even a slight underestimation.

So if you think argentina on a neutral court was 7 points better than us in a normal, and the decision to ct vs tie moves the needle say 5 points, you can plot where you are on the normal curve, calculate the number of sigmas away you are and calculate your odds of victory.

you can also facto in enthusiasm for future games, and how many points that impacts things. these are things you can estimate pretty damn good through observation.

so while i didnt do the math to the n;th degree, i have a feel for whre things might be on the curve and adjust.

some of this is intuitive. If you were a 20 point underdog and a ct saves you 7 points, then ct'ing moves yor odds of victory from like 5% to 20%, in other words its still a huge freaking reach. but if you were in a cointoss if you normal, that ct is going to be pretty freaking sweet in tiltin the odds in your favor.

again, i personally might have missed slightly on the curve and estimated to the nearest 10% to keep my math easy, but the math my thinking is based in is rock solid.

Last edited by TheUnrepentantGunner at 12/17/2009 9:27:34 PM

Wozz has mentioned for example that he has done studied the efficacy of various tactics. I have seen his results and use them to guide many tactical choices. I don't know if everyone can see it but the database website (which is entirely wozz's creation) also has a scouting tool that breaks down opposing teams and reveals for each player the average PP100 allowed (unfortunately it doesn't work with NTs, but you can look up individual players). This is info that's available nowhere else.

Yeah, there's actually 2 tools there, a team scout tool and a matchup evaluator. Obviously, they only work for club teams since we can't get NT info through the BBAPI. They were both intended to be released to either the BB or BB-usa community (for now I think it's usa-restricted), but I'm still adding some features I want to be included before making them public. But, if anyone would like to try them and help me beta-test, shoot me a BB-mail.

I outlined my plan for a player development department here: (122310.130).

Obviously, the U21 coach would need to have a high degree of input into the player development process, possibly even having more direct say than I would into the training plans of 18/19/20yo prospects. Having worked on Coco's scouting team over the past 2 seasons, I'm pretty confident that I can work well with Coco in NT-U21 coordination.

To all candidates

How would you work with the U21 coach to ensure the pipeline of players being developed fits with the National Team?

Things are different now. What i mean by that is 3-4 seasons ago people capped at all star level were full national teamers, and the best u21 candidates were also instantly likely to soon have full senior team caps soon after. Now I am not so sure that's the case.

Let me explain:

The classic u-21 player probably has at least high 3 ball potential out of the gate, with 6k++ salary, and very favorable heights (ie 7'4 center or 5'10 pg). Player gets powertrained in the most important skills for 2+ seasons, gets called up, contributes to the u21's.

In the old days there were discussions over the top 21 year olds who were also really senior team material (or near it)

Now we don't have that problem.

THe most important thing the senior coach can do, is help the u21 coach long term with identification of prospects, especially the 2nd tier, guys with MVP+ potential, that might have 4.5k salary at age 18. They might not quite have the ideal height (6'11 bigman or 6'2 guard), and won't likely get u21 caps. But it is important to keep these owners focused on the fact that they have a stud, let them know that if they want caps for the player, here is what needs to be trained, and get diversity among these longer term players. To get a SF like wendon (Steve Smiley, currently for sale is i think our best answer to wendon) takes a LONG time to train, and we can help make sure the skills are brought up together as optimally as possible.

You will still have the base skill behemeoths, but even at age 21 these players are likely to be 2 seasons from the NT.

So thats a long answer, but hopefully it makes sense.

To get a SF like wendon (Steve Smiley, currently for sale is i think our best answer to wendon) takes a LONG time to train, and we can help make sure the skills are brought up together as optimally as possible.

Surely he cant be close to USA NT talent... cough* HARVEY SELF for USA cough*

Good luck to all candidates - I've enjoyed tuning into the debate......my vote wouldve gone to URP but this is based on what I've read and not what goes on behind the scenes.

Any predictions as to how many people have actually read this thread from the USA community (suggestion thread idea?) and more importantly how many people will actually vote?