Posted by bagchucker on 12/29/2020 5:45:00 PM (view original):
why

how

you don know who's on that team till the season opens

That's pretty obvious, isn't it?
Cookies are cookies bc they appear on most teams at a particular cap level. If you play 1 OL, you get a good idea of what you're going to be up against in every OL. Even if the players aren't the exact same, the type of player is similar... Everyone used to use fast guys who are good on D & deadball pitchers... If you know that, you can build a team to beat it. If everyone is using high avg guys, you can build a team to combat that. If everyone is using HR hitters, you can build to combat that. If deaedballers are all the rage, you can build a team to combat that. If everyone is using 2020 pitchers, you can build a team to combat that... once the cookies (or the cookie strategies) are established, you dont need to see the rosters before the season starts to know what the majority of them will look like. Cookies and cookie strategies exist at all cap levels. Yes, you do have to play in 1 or 3 leagues at each cap level first to figure out who the cookies are and what the main strategies are before you can build to beat them, but IF everyone is doing something similar- it makes it easy to do (relatively). If everyone is doing something different, it makes it harder to figure out the ultimate strategy to win.

The total salary base should remain constant as salaries are adjusted. If you're attempting to identify value for each player by referencing market demand then you need a fixed money supply. A fixed money supply will best communicate market demand in the same way that a fixed language in a marketplace will best communicate market demand. As important as words are in communicating demand and determining value, money is even more important.

The only flaw in the original DP model that I can identify was that it failed to view different salary caps as different markets. Owners with 180mil at their disposal will have different spending habits than owners with 80mil. Have separate pricing and waiver wires for different salary ranges. Perhaps something like 1) 0-79mil 2) 80-119mil 3)120-159mil 4)160-199mil 5) 200-255mil.

Also, someone mentioned that the counter for uses wasn't reset after each salary adjustment. It should be.

For those against dynamic pricing whatsoever, it's worth pointing out that having a handful of overpriced players that are unusable is many times better than having a handful of underpriced players that make the rest of the database unusable.

I don’t have time to write an essay, but I’m definitely on the

-Don’t reset salaries

and

-total salary shouldn’t stay the same

hype trains

also think the cutoff for counting players is leagues above $120 mil

As many have pointed out, keeping the total salary the same is pointless. A small percentage of players will get a huge increase in salary and the rest of the database will get a tiny & meaningless decrease in salary. It just doesn't work in practice.

Many have pointed out the other side as well, locked in total salary. While the locked in total salary contingent voiced here may not outweigh those against 10:to:1, many stepping up here for non-set total salary have W/L success benefitted in the past and many for a long time under the current system. Hmmm . . .

Fairy safe assumption dynamic pricing is coming back. Yay. For those against set total salary, an honest question, neverminding for now whether a rolling ever changing total salary is practical or even reasonably possible fron a programing rewrite position (while it may be possible, and while I am impressed with what WIS can and has done with technology, this is fantasy baseball, not Wall Street or building a skyscraper) : to what degree do you propose its fluid movement ? If PlayerA and many like him who have long needed an increase adjustment goes up by DollarX, but the total salary also goes up by the same DollarX, has there really been an adjustment ? Or does 90% of DollarX get filtered elsewhere ? 50% ? 10% ? Who's to say, while also acknowledging there'll still extreme outliers on both ends of the DollarX distribution still not happy.

But taking $10,000 away from 1000 guys that are literally never used , and adding 10mil to 95 Maddux, isn’t really balancing things out.

I have been reading the posts and wondering how did these players become cookies and what makes it so obvious to some owners that if you draft these players and put them in a certain park that you will win 90 to 95 games in a open league. I am not a math genius but I have played over 70 times and my record is a .503 percentage. What am I missing that seems so clear to all in this forum.

I've read through the recent comments and can't help but notice that the opinions seem to largely be from the exact guys who created this mess by identifying and leveraging "cookies" in OL (and porting that insight into tournaments and theme leagues).

The easiest way to "fix" the dynamic pricing model is to STOP basing it on salary (and especially locking total salary), and start basing it on the thing that all of us (even grinders like me) must pay attention to... $/PA and $/IP. Break up the tiers into usable chunks, say, 5% of the available players sorted by $/PA, and monitor the usage within that tier for caps between $80M-120M (which seems to be the consensus range). Top 5% of players used have their $/PA go up 10%, next 10% of players usage have their $/PA go up 5%. Same for the bottom 5% and next 10%. Leave the middle 70% alone.

Obviously each cycle with see some players change "tiers" and some players around the edges will bounce back and forth with every update. Some (like catchers with A+ arms but batting .190) will fall for multiple cycles until they reach their balance point.

The problem was using total salary as the limiting factor for changes when the initial formula was used to determine $/PA and $/IP

Not only do I doubt only 1 or even a few players will bear the salary increase brunt for many more but smaller salary decreases, but there's a cap of 10% adjustment increase or decrease per player per adjustment period. This was noted in the How It Will Work main section of dynamic pricing talking points, link at the top of this thread. Also in that top priority section as a 'key principle' was maintaining the total salary pool across each group as equal; if increases are $X, then decreases are $X.

Nowhere in those talking points, not even in the down ballot Considerations section, is there any mention of non-set total salary. Where did this even come from ? Some biz wiz out of the blue that others are now running with to suit their previous success purposes ?

* Yes to dynamic pricing done better this time.
* Yes to maintaining set total salary.
* Yes and thanx for the ease in easier to transition digest 10% cap per player per period adjustment amount.

As I demonstrated in an earlier post:
If salaries must remain the same, then one of 2 things will happen,
Either A: thousands of players will see negligible decreases while 2000 Pedro and 21 Ruth go up by 10+M

or B: The players no one uses now become more interesting, but now 95 Maddux costs 50-60M

The ratio of never used players to bargains is not 1:1, not even close. Once that ratio gets very high, the system breaks because you can no longer lower the prices of the never used players by any meaningful amount without making the more popular players salary's completely ridiculous.

I've said from the start, why not just recognize who the most used players are and either A: manually increase THOSE prices or B: adjust the formula to account for that and let it happen naturally. Also note: The fact that 1921 Ruth is on literally every single team with a $200M+ cap should never be factored in since at that level all the teams are going to pretty much look the same.

The other only solution is to allow the total salaries to decrease over time.

Does fixing the problem with dead ball pitchers shutting off power help with this issue somewhat? 1902 Bill Bernhard is one of the most popular cookies in the sim. He gave up 0.16 HR/9 IRL with 122 OAV+, 166 ERA+, and 150 HR/9+ and costs $35,456 per IP. 2018 Aaron Nola has been used 61 times in the sim. He gave up 0.72 HR/9 IRL and has 125 OAV+, 170 ERA+, and 152 HR/9+ and costs $34,825 per IP. He's the closest comparable to Bernhard in each of these three + categories with at least 200 IP/162.

Bernhard's average performance in the sim is .roughly 254 OAVG, 3.20 ERA, 1.23 WHIP and 0.28 HR/9 in around 1,100 seasons. Nola's average performance is .236 OAVG, 3.44 ERA, 1.24 WHIP, and 0.72 HR/9.

The HR/9 for these two pitchers should be far closer. Nola isn't used more because of the far higher HR/9 figure. If Bernhard and Nola's sim HR/9 were more equalized, which is appropriate given their comparable HR/9+ and similar cost per IP (Bernhard about 2% higher), wouldn't that help with the cookie issue and spread price changes over a larger pool of players because more players would be used?

Warren Spahn, Lefty Grove, Jim Palmer, and Tom Seaver are four all-time great pitchers who are seldom used in competitive theme leagues because their nominal HR/9 and OAVG are too high. These and other great modern pitchers are outperformed by mediocre dead ball pitchers. I have to believe that having more options to build a successful team would not only make the sim more interesting but would reduce cookies as well because more players would be used.

Deadball pitchers only have good value if you build the right team, in the wrong hands they are disastrous.

Also factor in, many "great" pitchers are considered great because they were very good for 10+ seasons. Many of those pitchers never had that one incredible season, or some kind of weird statistical anomaly to make them into a WiS super star.

WiS doesn't care that someone won 200 games, especially if they are spread out over 15 pretty good seasons, but if an overall mediocre or even poor pitcher has one anomalous season (43 Niggeling) you end up with someone more popular than lots of Hall of Famers.

This is one reason I liked the idea of having some kind of committee make a decision on why a player is so popular, is it because they are seriously under priced, or are they just a popular player because of the name on the jersey? This would result in a much more accurate pricing scheme.

For what it's worth, Bernhard did record 116 wins in 200 starts in his career, so he wasn't exactly chopped liver back then :)

Posted by PennQuaker on 1/2/2021 1:57:00 PM (view original):
Does fixing the problem with dead ball pitchers shutting off power help with this issue somewhat? 1902 Bill Bernhard is one of the most popular cookies in the sim. He gave up 0.16 HR/9 IRL with 122 OAV+, 166 ERA+, and 150 HR/9+ and costs $35,456 per IP. 2018 Aaron Nola has been used 61 times in the sim. He gave up 0.72 HR/9 IRL and has 125 OAV+, 170 ERA+, and 152 HR/9+ and costs $34,825 per IP. He's the closest comparable to Bernhard in each of these three + categories with at least 200 IP/162.

Bernhard's average performance in the sim is .roughly 254 OAVG, 3.20 ERA, 1.23 WHIP and 0.28 HR/9 in around 1,100 seasons. Nola's average performance is .236 OAVG, 3.44 ERA, 1.24 WHIP, and 0.72 HR/9.

The HR/9 for these two pitchers should be far closer. Nola isn't used more because of the far higher HR/9 figure. If Bernhard and Nola's sim HR/9 were more equalized, which is appropriate given their comparable HR/9+ and similar cost per IP (Bernhard about 2% higher), wouldn't that help with the cookie issue and spread price changes over a larger pool of players because more players would be used?

Warren Spahn, Lefty Grove, Jim Palmer, and Tom Seaver are four all-time great pitchers who are seldom used in competitive theme leagues because their nominal HR/9 and OAVG are too high. These and other great modern pitchers are outperformed by mediocre dead ball pitchers. I have to believe that having more options to build a successful team would not only make the sim more interesting but would reduce cookies as well because more players would be used.

I think a move towards modern pitchers and away from deadball pitchers is already happening based on the combination of the prior dynamic pricing and the amount of stud pitching seasons from 2016 on that didn't get hit with the dynamic pricing.

Bernhard's price would be a lot higher if WIS hadn't messed up the dynamic pricing and treated his two 1902 combined seasons as separate for counting his usage. Plus he has a partial 1902 season that's almost identical to the combined.seasons.