If I understand correctly, pre or post decay would calculate to identical efficiency % for a single event, because, math. 50/100 wpprs is identical in percent to 25/50.
But wppr 6 uses your WPPRtunity which can be estimated with the efficiency % shortcut for active events (points on your card divided by eff%).
However, I think there must be an extra step somewhere that needs to use both decayed and not decayed values so the aggregate of all your efficiency is calculated properly when combined with new data, let’s do an experiment calculating two different ways… let’s say you have two events each where you got 50 points out of the possible 100 points, but one event decayed to 25 points.
a 50/100 event combined with a decayed 25/50 would weight out to an efficiency of 75/150.
Now if you get a new, third event of 20/200 your new eff would be 95/350 (27%).
Instead, what if we did the math like this:
A 50/100 event combined with a 50/100 event (we’re ignoring decay in this example) would be 100/200. Your third new event of 20/200 would turn your ratio into 120/400 (30%).
So depending how the decay “works” you can get two different numbers (27% or 30%). I assume it’s going to work like the former (decayed events have less of a say on your eff), but yeah I’m not sure if it currently works like that ?
I had thought it was always post-decay but sounds like I was wrong. The current v6.0 proposal doesn’t change this as it’s not something that’s currently functional in our system. It’s up to @Shep if he wants to modify the code to adjust things.
I’m curious as to how big of a difference this makes. I did Elwin’s calculation both ways and the difference was super negligible.
Looking at @jjszpgh and the analysis he did on himself, those results were insanely close as well.
Eff% always using 100% values: 22.4535%
Eff% using decayed values: 22.5797%
The main difference you’ll see is how quickly you can effect your efficiency with a big move, such as in my example.
If you are using decayed values, it’s easier to raise / lower your efficiency. If not, it’s much stickier, and harder to effect and actually does make me a bit concerned as taking a dump in a tournament may not effect your efficiency now too much, but it’ll stick with you for 3 whole years
I was also under this impression based on how the conversations have gone over the RayDay discord.
Since actual points are subject to decay, I would think it would be the most fair and most accurate to also calculate efficiently on a decay to reflect a purposeful recency bias.
I know you pointed to @Shep for the decision here, but I’m sure there’s enough programmer geeks who would be willing to pitch in on the back-end to help code up any changes.
Everything you have here looks right to me, @raydaypinball.
It sounds like things could go one of three ways depending on how @Shep wants to implement it:
Calculate WPPRtunity based on current points and adjusted tournament value with no changes to the Eff percent calculation
This incorporates decay into the calculation, but it’s hard for a player to precisely calculate their own WPPRtunity value (but their normal Eff percent will get them pretty close)
Calculate WPPRtunity value based on Eff percent with no changes to the Eff percent calculation
This does not incorporate decay into the calculation, and it’s easy for a player to calculate their own exact WPPRtunity value
Change the Eff percent formula to use current points and adjusted tournament value (rather than original points and original tournament value) and calculate WPPRtunity value based on the new Eff percent
Incorporates decay into the calculation and it’s easy for a player to calculate their own WPPRtunity value
I do think it should be attempted to be fixed though… bombing at a big tournament will sting a lot more if it follows you for the full amount for three years. Isn’t this why you solicit feedback ahead of time, to find issues like this so we can fix them?
Your analysis was based on a using 300 WPPR’S of total play.
The denominator will never be that small when it comes to the actual results we’ll be looking at.
We’re talking about thousands of actual WPPR’s attempted making the impact of any one event quite small in reality. I haven’t had time to deep dive into actual examples yet but will at some point.
The feedback is appreciated and has already impacted things on our side (for example the dynamic distribution part of our formula is likely changing a bit). We’ll get to those details when we make our official announcement on our site later.
Ultimately it’s on Brian to code, but I’m certainly not going to make that mandatory for us to move forward on this thing. If he has the time to get it in, great, if he doesn’t, the plan is still a good one IMO.
Coincidentally, in 12 years the kids will be old enough for dad to return to full-time competitive play.
Anyway, I’m pleasantly surprised that this change might disincentivize local killers from showing up to small, high-variance events for fear of blemishing their efficiency percentage.
So I have a horrible efficiency at ~14% as I never looked at it and was playing often just for the fun.
I will now drop 240 places. If in a future tournament the WPPR points won’t do any good for me,
I could still drop out before the last game and the tournament wouldn’t count at all, right?
For group match play qualifying, a player is excluded from the tournament results if they drop before playing half of the qualifying rounds.
For knockout tournaments, a player is excluded from the tournament results if they drop before the round where the first player is eliminated by strikes.
Those rules are already in place regardless of v6.0. I think they will do a good job of discouraging the “I’m doing bad so I’ll drop” mindset because for that to “work” you need to have played in less than half the tournament.
Except one of the posts in this thread mentioned the rule will probably be changed from the current rule of “anyone in top 250 will count if they play a single game in any format” to “anyone in top 1000”
