My last 7 rolls in Crum the underdog has won grant it, I was the underdog on 4 of those, but I’m really inclined to believe the percentages are just window dressing like Ht and Wt, and it’s truly a coin flip.

I have thee new rolls added to my report all to the underdog. But I was the udnerdog in two of them. New Strategy going forward. especially for low end D! teams.... which is me because I don't have any elites haha

If the devs have run through the calculations .. and if it is indeed a roll and if the lookup table for the roll is correct. Then all the rest of this is irrelevant.

If the RNG roll is a number between 0 to 1 and really random , and if a 0.75 to 0.25 (75% to 25%) final chance the compare. As long as calculation works correctly and it gives the proper result, none of this in any way matters.

0-0.25 .. Underdog Wins.
0.26-1 .. Favored Wins

This is only an issue if the RNG is broken or the math is programmed wrong.

Posted by hughesjr on 12/6/2022 2:32:00 PM (view original):
If the devs have run through the calculations .. and if it is indeed a roll and if the lookup table for the roll is correct. Then all the rest of this is irrelevant.

If the RNG roll is a number between 0 to 1 and really random , and if a 0.75 to 0.25 (75% to 25%) final chance the compare. As long as calculation works correctly and it gives the proper result, none of this in any way matters.

0-0.25 .. Underdog Wins.
0.26-1 .. Favored Wins

This is only an issue if the RNG is broken or the math is programmed wrong.

The thing I think about, since some users seem to be having this kind of experience for extended periods of time (I think most of us have it for a few seasons here and there) is that at some point there was another external factor added that is complicating rolls; and that it gets "stuck" for some users/programs. So it's (hypothetically) not straight random, but there's another potential modifier in there affecting luck on anything from recruiting rolls, to injuries/grades, to game results (maybe separately, maybe not).

The other thing I think about is "feedback". We know the game design uses that concept to manipulate simulations under certain circumstances, when the results are outside certain parameters. Like team A is really overachieving at the half, and the game engine takes that into account and cools it moving forward. In another long and tedious thread from years ago, someone used the example of flipping over cards on a table. And instead of those cards being put back and shuffled once flipped, they're just removed. So a 70% FT shooter, when under conditions triggering feedback, every made FT is going to lower the odds of making the next - because it's removing a card off the table, essentially. *IF* the game uses that concept world-wide in recruiting - and maybe it does? - that could explain a bit why we sometimes see runs of underdogs winning. If it's looking for a certain % of underdog wins, especially if it is counting 100-0 battles where teams offer but don't fight hard enough to get in range, that could really start to skew results. I don't really think it could be *all* of those types of results, I think it'd be very obvious to most users if it was that big of a tilt. But maybe some subset of them is in there screwing with them?

But also, again, it also could be - probably is - small sample size mixed with confirmation bias and the long-lasting memory of the bad beat.

"But also, again, it also could be - probably is - small sample size mixed with confirmation bias and the long-lasting memory of the bad beat."

Which is why I am am tracking all my rolls. There is one roll that I won that I don't have tracked since I started that I am not sure of.

I also started tracking because I found that I was winning more rolls as a dog than I would have expected to and believe this might be an ideal way to lower mid level D1 teams to recruit. I am now just trying to get and stay high on as many recruits as I can that I think fit my team. I end up getting to Very High on some but the goal is to get High and Gamble. Which Ironically is also a life goal.

Posted by Fregoe on 12/7/2022 7:38:00 AM (view original):
"But also, again, it also could be - probably is - small sample size mixed with confirmation bias and the long-lasting memory of the bad beat."

Which is why I am am tracking all my rolls. There is one roll that I won that I don't have tracked since I started that I am not sure of.

I also started tracking because I found that I was winning more rolls as a dog than I would have expected to and believe this might be an ideal way to lower mid level D1 teams to recruit. I am now just trying to get and stay high on as many recruits as I can that I think fit my team. I end up getting to Very High on some but the goal is to get High and Gamble. Which Ironically is also a life goal.

wonderful

I kept track of the top 100 recruits for a few worlds awhile back. Here are some numbers

VH vs H
329 battles, High won 29%

VH vs VH vs H
68 battles, high won 34%

VH vs H vs H
49 battles, high won 35%

Maybe 68 battles isn't the biggest sample size in the world but the High team winning 34% of the time is way too high.

Posted by Fregoe on 12/7/2022 7:38:00 AM (view original):
"But also, again, it also could be - probably is - small sample size mixed with confirmation bias and the long-lasting memory of the bad beat."

Which is why I am am tracking all my rolls. There is one roll that I won that I don't have tracked since I started that I am not sure of.

I also started tracking because I found that I was winning more rolls as a dog than I would have expected to and believe this might be an ideal way to lower mid level D1 teams to recruit. I am now just trying to get and stay high on as many recruits as I can that I think fit my team. I end up getting to Very High on some but the goal is to get High and Gamble. Which Ironically is also a life goal.

One user, or a group of a few users (who were likely drawn to participate because of negative experiences) can show certain “flaws” in the system with enough data - like the kinds of bias I talked about above. But they’re not going to show anything wrong with the core math, as in high and very high are universally flipped or meaningless or whatnot; for that, you’d need a much bigger pool of coaches meticulously tracking for many seasons.

I’ve tried out the get-to-high-on-many approach (instead of all-in VH on 3, for example), and it isn’t particularly good or bad. It’s just different. You’ll pick off a few recruits, yes. But you’ll have less choice over who signs with you, so your team needs to stay flexible. Works best with zone teams in that way, so you can carry multiple walkons and not think about it. I did not find that I was winning more battles than expected from high, though.

For what it’s worth, the 2 battles I currently have eligible to report from 5 teams, I’ve won both as Very High against High opponents. One was a 68-32 battle (Fresno), and one was a 57-23-20 battle (UConn).

Posted by Benis on 12/7/2022 9:34:00 AM (view original):
I kept track of the top 100 recruits for a few worlds awhile back. Here are some numbers

VH vs H
329 battles, High won 29%

VH vs VH vs H
68 battles, high won 34%

VH vs H vs H
49 battles, high won 35%

Maybe 68 battles isn't the biggest sample size in the world but the High team winning 34% of the time is way too high.

good on you, 329 battles is a lot. 29% is probably right on point? the 3 ways look off but still?

Here's my stats over the last 4 seasons for 4 teams (15 total seasons). I've had 26 rolls and this is how they've shaken out...

2-Team Rolls
<30% Odds: 0 win - 1 loss
31-40% Odds: 1 win - 0 loss
41-50% Odds: 1 win - 2 loss
51-60% Odds: 5 win - 2 loss
61-70% Odds: 4 win - 4 loss
>70% Odds: 2 win - 1 loss

3-Team Rolls
<20% Odds: None
21-30% Odds: None
31-40% Odds: 1 win - 1 loss
41-50% Odds: None
>50% Odds: 1 win - 0 loss


Obviously not a huge sample size, but nothing out of the ordinary so far. But I will also say that I started tracking these because I went on a year-long unlucky streak, basically losing most of my rolls despite usually being the leader.

Posted by Fregoe on 12/7/2022 7:38:00 AM (view original):
"But also, again, it also could be - probably is - small sample size mixed with confirmation bias and the long-lasting memory of the bad beat."

Which is why I am am tracking all my rolls. There is one roll that I won that I don't have tracked since I started that I am not sure of.

I also started tracking because I found that I was winning more rolls as a dog than I would have expected to and believe this might be an ideal way to lower mid level D1 teams to recruit. I am now just trying to get and stay high on as many recruits as I can that I think fit my team. I end up getting to Very High on some but the goal is to get High and Gamble. Which Ironically is also a life goal.

This guy is a wordsmith.

Team	Year	Player	Fav	Dog	Winner	winning team
Coppin St	188	Glenn Miligan	62	38	Underdog	Coppin
Drake	168	Jan Threlkeld	57	43	undergod	Missouri
Wyoming	169	Thomas Mathis	74	26	Underdog	Air Force
Miss V St	134	Jason Morejon	63	37	undergod	Miss V St
Coppin St	189	Shawn Smith	49	30	underdog	Morgan St
Coppin St	189	Jay Barrett	58	42	underdog	Coppin St
Miss V St	135	Jason Morejon	63	37	Underdog	Miss St
Clemson	132	Richard Lebel	74	26	Fav	Indiana
Boise St	129	Douglas Vogel	69	31	Fav	Gonzaga
Wyoming	170	Tim Ward	75	25	Fav	Gonzaga
Wyoming	170	JR Nagata	69	31	Fav	Iowa St
Lamar	135	Fred McCullough	76	24	Fav	Lamar
Drake	170	Ahmad Whitehill	71	29	Fav	Drake
Drake	170	Zach Parks	55	23	Fav	Illinois
Drake	170	Roy Jones	48	17	Underdog	ND
Clemson		Craig Bryant	51	49	Fav	Vandy
Drake	170	Ruben Mullins	38	12	Fav	Missouri	5 WAY ROLL
Miss V St	135	Daniel Puckett	66	34	Fav	Alcorn st
Miss V St	135	eric Hemphill	72	27	fav	Drake
Coppin St	191	j Cambell	57	23	fav	Pitt	3 Way
Miss Valley St	135	Brandon Randolph	53	47	Undedog	G Tech
Miss Valley St	135	Bryant Wetzstein	74	26	underdog	Tenn Martin
Miss Valley St	135	John Park	59	20	Fav	Aub
Prarie View	134	James Phillips	53	47	Underdog	Florida St
Coppin St	192	Michael Larose	68	32	Underdog	Coppin St
Coppin St	192	Thad Evans	72	28	Underdog	Clev St
Drake	172	William Chilton	53	18	Fav	Hou	3 way
Clemson	134	Bryan Lacy	57	43	Underdog	Clemson
Clemson	134	Michael Williams	51	19	undergod	Clemson	3 way
Prarie View	135	Hults	72	28	underdog	Houston
Miss V State	136	Tyler Smith	55	45	underdog	Ark Pine B
Prarie View	135	Kerry Phillippi	68	32	Fav	Prarie View
Lamar	137	James Mingus	59	41	underdog	PrarieView
Coppin St	193	Elmer Redden	61	39	underdog	wagner
Coppin St	193	Micahel Dillard	54	46	unerdog	Coppin St
West Virginia	137	Mathew Gilchrist	52	48	underdog	West Virginia
Texas SA	173	Roy Samons	57	43	Underdog	Oklahoma
Texas SA	173	John Baehr	48	20	Fav	Oklahoma	3 way
Texas SA	173	James Vaetch	51	49	Fav	New Mexico
Coppin St	193	James Knight	50	20	Fav	Maryland	3 way

I have been tracking my battles for a quite some time. If you are trying to figure out whether you are running above or below expectation. Sum up all the % you had in your flips. Subtract 1.00 for each person you Won. Formula might look like this. SUM( % Considered ) - COUNTIF("W"). Frogoe can't really do that analysis on what you posted since I can't tell whether you where the favorite or not, but would be interested to see how you are running.

My main goal is to just see it makes sense to get as many "better" recruits to High.

Posted by Fregoe on 12/12/2022 12:49:00 PM (view original):
My main goal is to just see it makes sense to get as many "better" recruits to High.

I really do think the numbers, except for maybe 3-way VH-VH-H battles seem to work out as expected.

If you get 10 guys to 67-33 ahead .. you should win 6.7 players. You would need to get 20 players to 33% to mathematically win the same number of players.

So, the question becomes, is it cheaper to get 20 players to high or 10 players to very high.