Small or Far Away – Father Dougal on Cows
This is an exciting week, where we have two weeks of data and can bring in cows we may have missed in our initial teams. If you do, be sure the upside of the new cow is enough better to be worth using a trade. Judlow’s rookie review will help you with who is and is not worth bringing in. But as far as cows go….
A big thing, at least for me, is getting to look at the breakevens and projections within the game. With two weeks of scores, projected scores, projected prices changes based on those scores, and Breakevens, along with no prior adjustments in place, we have the data we need to analyze the Supercoach price model. So, I did. Starting with a few cows, and not liking the results, I moved on to include a wider variety of different price points and breakevens. I tried a bunch of things, thought a lot, and feel like I eventually figured a bit out, which is below.
So, we know the base formula used for price changes. [ (Current-Price * .75) + (Average-of-the-last-three-scores * .25 * Magic-Number) ] with either the magic number changing over time or a multiplier being used. If we plug a player’s breakeven into the above calculation along with his first two rounds scores, we should get a price change of close to zero, that being the idea of a break even. That does not happen. I can’t say I am surprised that there is some variance from that base formula. But, after some fiddling and looking at more players, I can see seven of nine (my favorite ratio for obvious reasons) are all off by about the same %, and the ones with a starting price over $200,000 are all within about 0.06% of each other.
I tested both ways that the base formula could be altered to see if I could determine which one was being used. Short answer, I could not get the numbers to work by changing the magic number around. But. by adding a * 0.985 to the formula, I was able to get pretty close to the actual price change estimates using the published numbers. The first 4 columns in the table are all right from the SC site. “$ Diff at SC BE” is the difference between what I got running the base formula and the R3 projections, and the next column is what % of the projected price that difference is. The last column shows how far off [ (Current-Price * .75) + (Average-of-the-last-three-scores * .25 * 5396) * 0.9875 ] is from the SC predicted price changes. As you can see, most are very close. And I mean *very* close. Papley, who is by far the most wrong, is 0.004% off. Hall is 0.00025% off.
Thinking about it, I can see why there is a multiplier. The idea is to keep the total cost of all players the same. So, they calculate all the new prices, then they scale them by dividing the current total by the new total, and then multiplying the result by each player’s price. For example, with made up numbers:
Total price of all players is $100,000,000. After prices changes the new total price of all players is $110,000,000. Dividing $110,000,000 into $100,000,000 gives us 0.9091. So, if Libba should have a value of $421,900 we multiply that by 0.9091 to get his new price of $383,549.29 which is rounded to $383,500.
Now I have no idea why they have a small price drop before R3, but clearly there is one. I plan to run numbers for a lot more players and see what that shows. I also want to see if, after finding a formula that works for most players, I can apply it to the outliers and see if the variance is due to rounding errors or something else like that. It is also possible that they do something at the extremes, but I suspect it is rounding issues which are getting in the way of finding the right formula, and it will all make sense one I have it. The outlier so far being the very low priced Papley leads me to suspect that. But we’ll see.
I’m looking forward to the most obvious proof of the official BEs being off, when a player scores exactly his BE but still has his price change. Pretty sure I saw that last season but extra want an example from R3 this season. If you find one after the prices update please let me know!
Since cows seem extra subject to variance I’m going to track a set of more expensive players as well, in the hopes of getting better data as I did above.
Another thing I am curious about is if they keep a not rounded price internally and just round what is displayed, or if they store and use the rounded amounts for future calculations. I’m going to try and figure that out as well.
Next week I plan to talk more about this after we have seen the first prices changes. After that I’ll take a shot at how to tell when a cow is no longer worth fattening. And if anyone has a cow related request, please post it in the comments!
Leave a comment / Scroll to bottom