BuzzerBeater Forums

Yep, naturally it is 10% of the training, which diminishes with age and is dependent on the type of training itself.

If true however, that would mean ~25 TSP difference in the final builds of max potential players, or ~15-20 TSP at age 24-25 where potential caps don't contaminate the data, between players trained with a lvl 3 gym and no gym. As NT coach, closely monitoring training for many players - I have not seen such a huge difference.

Before the gym and youth coach we were used to getting 15-16 pops in the first season of 1v1. Now the meta is ~17-19 pops. This is around 15-20% increase due to youth coach and gym combined. If we attribute 5-7% of this to the youth coach (it is a popular belief that youth coach effectiveness is 1% per level), then again the gym increase would be 10-12% at best, not 30% for sure.

I think part of the reason you won't see it via TSP is that sometimes the training can go to FT, ST (things that don't attribute to TSP growth). They can also go into skills you have no intention of training in the long run and you may never see a pop there. I think about 1/6th of the difference can be seen there.

At a certain age too, the training will go into skills that you no longer want to pop (ones that are already at 20), or they will be diminished by elastics in some way. Some explanation might lie there. Maybe there are additional penalties to these 10%'s that we don't know about under the hood. Maybe they are counted at 10% of the raw training value before your trainer and youth trainer increase them.

I also believed gym constituted about 8-10TSP difference at level 3 but not 25 TSP. I think about 10 of the 25TSP can be explained by us simply looking at the distribution of skills, FT/ST, hitting skill caps or elastics. The other 5 TSP may be something about the exact details of the implementation that we do not know about. Or that is my assumption when I got this knowledge.

I think part of the reason you won't see it via TSP is that sometimes the training can go to FT, ST (things that don't attribute to TSP growth). They can also go into skills you have no intention of training in the long run and you may never see a pop there. I think about 1/6th of the difference can be seen there.

That part I agree on. I'd even say 1/4th to 1/5th since most training types improve at least three different skills and ST/FT are like 2/9 left. Then again the same percentage of the no-gym cross-training goes to ST/FT too, so it's more like 3-4 TSP lost there, not 10.

There are many other aspects (like whether elastics matters for cross-training, does cross-training tend to go into lower skills etc) that would make this way to over-complicated.

One thing though, and it's a huge conundrum for me because in general I am against BBs sharing this much detail with the community, but since we are already at this point - it would be cool if Justin could throw us a bone here.

Hahaha yes, I sure now!

Go ahead, please.

Yeah, that's pretty much given.

How about you just reveal how many slots there are in the base (no gym) cross-training algorithm and we leave it at that?

Justin: I went ahead and edited it out. Thank you for checking the implementation of that again.

mink0ff: In the future I think we'll be getting something similar to x% rather than exact number to avoid these conversations and less potential for incorrect or confusing information. I think broad ideas about how the engine, training, etc... functions without exact numbers is probably the best way for us to receive new information. Thank you for double checking the information on here and helping improve the accuracy!

In the future I think we'll be getting something similar to x% rather than exact number to avoid these conversations and less potential for incorrect or confusing information. I think broad ideas about how the engine, training, etc... functions without exact numbers is probably the best way for us to receive new information.

Yep, I too feel this would be the right approach going forward.

You're most welcome! I'm a huge fan of community-driven knowledge deduction and data gathering, so once again - this is awesome!

You can also find it in this thread (312347.1).