|
Post by frunobulax on Nov 18, 2017 10:48:11 GMT -5
I still don't get how you calculate negative prices, please explain it to a stupid old man like me It's pure math. Located 4 distinct offers that combined the same 4 different items (Bulgarsi+Tempest, Inquisitor+Shocktrain, Bulgarsi+Shocktrain, Inquisitor+Tempest). That gives you 4 equations (the offers with their number of components and their total price) with 4 variables (the conversion rate). That allows to calculate all the variables (ie. conversion rate). After that I could use the now known values in other offers to calculate the rest of items. Ah But that assumes that the prices are fixed in some way, and I'm pretty sure that if you use something like 6 equation with 4 variables, using todays offers, you'll come to an insolvable equation
|
|
|
Post by hon_shu on Nov 19, 2017 1:42:02 GMT -5
It's pure math. Located 4 distinct offers that combined the same 4 different items (Bulgarsi+Tempest, Inquisitor+Shocktrain, Bulgarsi+Shocktrain, Inquisitor+Tempest). That gives you 4 equations (the offers with their number of components and their total price) with 4 variables (the conversion rate). That allows to calculate all the variables (ie. conversion rate). After that I could use the now known values in other offers to calculate the rest of items. Ah But that assumes that the prices are fixed in some way, and I'm pretty sure that if you use something like 6 equation with 4 variables, using todays offers, you'll come to an insolvable equation You are correct. But the problem is mostly rounding errors. We are asked for example 500 Au where the "correct" price could be 498.123 or 501.333 or 521.1 Au. That's why I used a least squares solver (see my previous post). I'll wait another day or two to get more data before I post the conclusion, but so far the WSP estimates work out nicely with <2% error and an estimated cost per Dash bot of approx. 80000 WSP (without the rebate). An Inquisitor would cost you 115000 WSP. I don't have much confidence in the Au estimates yet. Maybe the Au prizes are not fixed, then the result would give us only an average price. Which is still usefull. Or Pixonic is occasionally changing the prize for a single offer to throw a monkey wrench into my calculations
|
|
|
Post by maverick on Nov 19, 2017 5:31:18 GMT -5
I think it all depends. In the past during the BM I opened only bronze. Best thing I got was a Jesse. Which I didn't have so not bad. During the event this time since I was getting so many keys I decided to only do silver once I got to 1000 keys. I got: Stalker (already have one) Gepard (never had one and one day may use it) 1000 keys (opened 10 more silver chests) 700 Tempest components 700 Tempest components 300 Ember components 100 Haechi components Later in the event I opened some bronze and got a Zeus. My third total so that saves me gold. Also someone in my clan got a Butch from opening a bronze and a Zeus from opening a silver. I think the chances of getting something good from a gold are better. But you have alot less chances. Someone at Pixo likes you.
|
|
|
Post by frunobulax on Nov 19, 2017 14:35:03 GMT -5
Ah But that assumes that the prices are fixed in some way, and I'm pretty sure that if you use something like 6 equation with 4 variables, using todays offers, you'll come to an insolvable equation You are correct. But the problem is mostly rounding errors. No. There are differently structures offers, but you can get essentially the same offer on one day for 500 gold, one day later for 400 gold and the day after that for 700 gold. Any linear equation approach is doomed here.
|
|
|
Post by hon_shu on Nov 20, 2017 12:57:54 GMT -5
You are correct. But the problem is mostly rounding errors. No. There are differently structures offers, but you can get essentially the same offer on one day for 500 gold, one day later for 400 gold and the day after that for 700 gold. Any linear equation approach is doomed here. Well, if the price for the same offer changes from day to day then any approach is doomed really. But I'm not convinced it's the case and i haven't observed that either. Can you give a specific example where the exactly the same offer was available at two different price points?
|
|
|
Post by frunobulax on Nov 21, 2017 3:21:47 GMT -5
I don't track the offers, but maybe I can give you an example later.
|
|
|
Post by frunobulax on Nov 21, 2017 5:35:30 GMT -5
@hon_shu, look at this post. All WSP offers. (1) 180 Ku + 240 Em = 2000 (2) 180 Ku + 120 In = 2900 (3) 480 Te + 240 Em = 1900 (4) 480 Te + 120 In = 1900 From (3) and (4) we get 240 Em = 120 In. But from this, (1) and (2) should have the same WSP price, but they haven't. There is no underlying structure. Offers are generated somewhat randomly, with certain premium components (Haechi, Bulgasari) coming only at 120 to 360 components while more "common" components (Scourge, Tempest) come in offers of up to 1430.
|
|