# Training Camp lvl 20 - Results

I’m a little concerned about sampling bias. Relatively few players participate on these forums, and perhaps those getting great TC20 results are over-represented because they want to show off their new toys, or those with bad results don’t want to fess up? Either is a bit of a stretch.

Assuming that the people on the forum are reporting all their results, sampling bias won’t matter that much.

It’s just getting enough data and 275 still has a significant error rate. I’m pretty sure the number will trend as expected to that over time, it just fits too neatly with other things in game (though perhaps we’re off by a percent) and I can tell you from tracking items there will be swings in the data during collection summarizing them each time but it will converge eventually.

ETA: just judging from the Internet and this forum too, the sampling bias would go the other way… people complain a lot more about negatives than crow about positives.

1 Like

The more data collected, the less the percentages should move. We were pretty locked in on 5% and to have it swing up to 7% is a pretty large deviation. The reason we’re taking our own out of band samples is because of the ad-hoc way the data is being collected here. We capture every single draw from the people participating (everyone here has been invited but there have been no takers), including which hero was pulled. No our sample data isn’t as large but it’s complete and unbiased. It’s still small at 275 but again both communities were locked in at 5% so it’s odd to me that one is swinging up and the other down entire percentage points. I guess we just don’t have enough data yet, a few extra 5s with only 275 results can swing things a whole point. With the numbers here it would take 6 or 7 though and it’s swung 2 points so that’s an abnormally large number of extra 5s

Doing the math on this last result from talisax

3* 551. 75%
4* 136 19%
5* 48. 7%

5* is 6.1%, that’s much less worrisome than 7%

48/735 = 6.5% I round up

3* 565 75%
4* 138 18%
5* 49. 7%

1 Like

I got 2 3* today. I am back to 1 lvl20 training for a couple of days.

I’m starting to train TC20 again. Add 3 x 3* to my personal results, for totals of:

53 x 3* = 68%
18 x 4* = 23%
7 x 5* = 9%

For the record, here are the 5* heroes I have pulled:

Marjana (x 2) (fire)
Isarnia (ice)
Sartana (dark)
Domitia (dark)
Horghall (nature)

4 Likes

And another 3* today.

3* 576 75%
4* 140 18%
5* 50 7%

New results since previous post. I have now two TC20
3 Valen
3 Valen
4 Boldtusk
5 Elena

3 Prisca
4 Sabina

On month ago I got 0 5* then teltoc reborn event starts. I got marjana in TC20 and same day owl gard with 300 gems. One week later vivica with a token. Yesterday Elena with TC20 and today a natural chest brings me two heros tokens and a troup token. Natalia and a 4* red troup I was missing were my New Year reward. Patience finally pays What is disappointing here is that I got 5 5* in one month but from two colours. And between Vivica and Owl guard looks like Vivica is the best option even if from what I saw in 8 months it’s a poor option. And for red I was going to ascend Marjana and forget about Elena. But know I’m not really sure Marjana is the best option as Natalia is a HoT heroes.

3 Likes

My first ever 4 TC 20 results were…

4 3*…boooo.

At least I can contribute now!

Another 3* today. I have my second tc running again, so will have 2 results in 2 days.

Perhaps it would simplify the data collection efforts if people post results weekly, not daily.

1 Like

Sure, but since I wasn’t going to keep track myself, I asked Talisax if he still wanted my results if I posted them every 2 days, so there you go…

I’m good with however often you want to post resultts. I sometimes update it multiple times in a day

1 Like

Traning TC20:

3*: 31

4*: 14

5*: 2

Pulled my second 5* yesterday (Richard)! Here are my results since my last post:

3* : 8
4* : 2
5* : 1