# May 2020 probabilistic senate prediction competition **URL:** \ **Category:** Politics & Elections\ **Created:** [May 15, 2020, 10:54pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889 "2020-05-15T22:54:08Z")\ **Posts on this page:** 20\ **Page:** 2
**Author:** ![septimus](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/septimus/32/410_2.png) [@septimus](https://boards.straightdope.com/u/septimus)\ **Post date:** [May 17, 2020, 5:41pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/21 "2020-05-17T17:41:21Z")
To check my understanding of the scoring, I compared **Calavera** ’s score to mine in three cases. Case a) **Calavera** ’s predictions are all correct Case b) **Calavera** ’s except Collins wins in Maine Case c) **Calavera** ’s except Collins wins and Tillis is defeated **Calavera** ’s score a) 0 (perfect) b) -1 c) -2 **Septimus** ’ score a) -.8275 b) -.9275 c) -1.0275 Arithmetic to compute **Septimus** ’ score a) 0 -0 -.04 -.01 -.04 -.09 -.04 -.04 -0 -.2025 -.04 -.1225 -.2025 b) 0 -0 -.04 -.01 -.04 -.09 -.04 -.04 -0 -.3025 -.04 -.1225 -.2025 c) 0 -0 -.04 -.01 -.04 -.09 -.04 -.04 -0 -.3025 -.04 -.1225 -.3025 If several people enter the contest, presumably some will take the opportunity to play for perfect scores in cases b and/or c. I’m not sure that splitting the prize money among all those tied for best score is the right approach here. @ **Lance** – Does this look correct? ---
**Author:** ![Lance\_Turbo](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/lance_turbo/32/6156_2.png) [@Lance\_Turbo](https://boards.straightdope.com/u/Lance_Turbo)\ **Post date:** [May 17, 2020, 6:42pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/22 "2020-05-17T18:42:50Z")
Your calculations are not correct. It’s mean squared error. ```auto Scenario Lance Turbo septimus Calavera a 0.0889 0.0637 **0.0000** b 0.1232 **0.0713** 0.0769 c 0.0933 **0.0790** 0.1538 ``` ---
**Author:** ![Bijou\_Drains](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/bijou_drains/32/2814_2.png) [@Bijou\_Drains](https://boards.straightdope.com/u/Bijou_Drains)\ **Post date:** [May 17, 2020, 6:55pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/23 "2020-05-17T18:55:00Z")
> [@Lance\_Turbo](#): > > You can keep on talking about how much smarter you are than prediction markets or you can do something about it. This thread is the perfect opportunity for you to show your stuff. How did prediction markets do for Clinton vs. Trump? I know most “experts” predicted Clinton to win easily. Including many UK bookies. Nate Silver was a rare case where he gave Trump about 1/3 chance of winning. I’ll bet you $100 to charity of your choice if Tillis loses. ---
**Author:** ![Lance\_Turbo](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/lance_turbo/32/6156_2.png) [@Lance\_Turbo](https://boards.straightdope.com/u/Lance_Turbo)\ **Post date:** [May 17, 2020, 7:23pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/24 "2020-05-17T19:23:55Z")
I won’t bet you for two reasons… > [@New Rules for Politics and Elections – January 2020](https://boards.straightdope.com/t/new-rules-for-politics-and-elections-january-2020/845727/1): > > **No betting.** Gambling between posters in GD and Politics and Elections over debate outcomes or real world events is forbidden. In the past we’ve had posters use this as a rhetorical device to abuse other posters and accuse them of lacking the courage of their convictions. It’s a jerk move we have decided to no longer allow. > [@U.S. Senate races 2020](https://boards.straightdope.com/t/u-s-senate-races-2020/852876/33): > > I don’t bet on political races. If you don’t wish to participate in this competition, just don’t participate. ---
**Author:** ![septimus](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/septimus/32/410_2.png) [@septimus](https://boards.straightdope.com/u/septimus)\ **Post date:** [May 17, 2020, 7:26pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/25 "2020-05-17T19:26:30Z")
> [@Lance\_Turbo](#): > > Your calculations are not correct. It’s mean squared error. I think your numbers become almost[sup]\*[/sup] _identical_ to mine if you multiply by (-13) but you’ve come to the right message-board to practice pedantry! 🙂 [sup]\*[/sup] - Only _almost_ identical because my numbers are exact; yours are rounded. ---
**Author:** ![Jonathan\_Chance](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/jonathan_chance/32/701_2.png) [@Jonathan\_Chance](https://boards.straightdope.com/u/Jonathan_Chance)\ **Post date:** [May 17, 2020, 7:27pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/26 "2020-05-17T19:27:29Z")
> [@Bijou\_Drains](#): > > I’ll bet you $100 to charity of your choice if Tillis loses. No, you won’t. It’s specifically against the rules as of Jan 1 for posters to bet. Please don’t do so again. No warning issued. ---
**Author:** ![Bijou\_Drains](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/bijou_drains/32/2814_2.png) [@Bijou\_Drains](https://boards.straightdope.com/u/Bijou_Drains)\ **Post date:** [May 17, 2020, 7:34pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/27 "2020-05-17T19:34:33Z")
Still waiting to hear how betting markets did on Trump v Clinton in 2016? ---
**Author:** ![Lance\_Turbo](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/lance_turbo/32/6156_2.png) [@Lance\_Turbo](https://boards.straightdope.com/u/Lance_Turbo)\ **Post date:** [May 17, 2020, 8:11pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/28 "2020-05-17T20:11:35Z")
> [@Bijou\_Drains](#): > > Still waiting to hear how betting markets did on Trump v Clinton in 2016? Why? This thread is about probabilistic forecasts. ---
**Author:** ![Bijou\_Drains](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/bijou_drains/32/2814_2.png) [@Bijou\_Drains](https://boards.straightdope.com/u/Bijou_Drains)\ **Post date:** [May 17, 2020, 8:15pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/29 "2020-05-17T20:15:58Z")
> [@Lance\_Turbo](#): > > Why? > > This thread is about probabilistic forecasts. In other words, they did badly. ---
**Author:** ![Lance\_Turbo](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/lance_turbo/32/6156_2.png) [@Lance\_Turbo](https://boards.straightdope.com/u/Lance_Turbo)\ **Post date:** [May 17, 2020, 8:20pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/30 "2020-05-17T20:20:50Z")
> [@Bijou\_Drains](#): > > In other words, they did badly. No. If you’d like to discuss the effectiveness of betting markets at predicting political outcomes, start a thread. I think you’ll be surprised at what you learn. If you’d like to participate in the competition in this thread, submit an entry. ---
**Author:** ![DrDeth](https://avatars.discourse-cdn.com/v4/letter/d/b487fb/32.png) [@DrDeth](https://boards.straightdope.com/u/DrDeth)\ **Post date:** [May 17, 2020, 8:24pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/31 "2020-05-17T20:24:11Z")
Hey, maybe you guys that want to argue about how good or how useful polls are- can start another thread, and leave us to discuss the Op and senate predictions? Please? We know polls arent always accurate. _Thankyouverymuch._ ---
**Author:** ![septimus](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/septimus/32/410_2.png) [@septimus](https://boards.straightdope.com/u/septimus)\ **Post date:** [May 17, 2020, 9:01pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/32 "2020-05-17T21:01:45Z")
> [@Bijou\_Drains](#): > > Still waiting to hear how betting markets did on Trump v Clinton in 2016? IIRC they did MUCH better than most pundits. Pundits — and Dopers — thought Hillary was a shoo-in. Prediction markets, however, were giving Trump a big chance (though not more than 50% IIRC). ---
**Author:** ![Bijou\_Drains](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/bijou_drains/32/2814_2.png) [@Bijou\_Drains](https://boards.straightdope.com/u/Bijou_Drains)\ **Post date:** [May 18, 2020, 1:31pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/33 "2020-05-18T13:31:54Z")
I can predict races in October. For now I predict most incumbents will win with a few possible exceptions such as Colorado and Maine and Alabama. ---
**Author:** ![Lance\_Turbo](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/lance_turbo/32/6156_2.png) [@Lance\_Turbo](https://boards.straightdope.com/u/Lance_Turbo)\ **Post date:** [May 18, 2020, 1:52pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/34 "2020-05-18T13:52:23Z")
> [@Bijou\_Drains](#): > > I can predict races in October. For now I predict most incumbents will win with a few possible exceptions such as Colorado and Maine and Alabama. Great. I’ll probably do another competition or two as the election gets closer. You’re welcome to participate in those. In the meantime, this thread is not for you. ---
**Author:** ![Jonathan\_Chance](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/jonathan_chance/32/701_2.png) [@Jonathan\_Chance](https://boards.straightdope.com/u/Jonathan_Chance)\ **Post date:** [May 18, 2020, 3:54pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/35 "2020-05-18T15:54:50Z")
Let’s dial it back a bit, people. We’re all friends, here. ---
**Author:** ![Elendil\_s\_Heir](https://avatars.discourse-cdn.com/v4/letter/e/7cd45c/32.png) [@Elendil\_s\_Heir](https://boards.straightdope.com/u/Elendil_s_Heir)\ **Post date:** [May 18, 2020, 6:51pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/36 "2020-05-18T18:51:02Z")
And see this earlier thread on U.S. Senate races nationwide in 2020: [https://boards.straightdope.com/sdmb/showthread.php?t=895277](https://boards.straightdope.com/sdmb/showthread.php?t=895277) ---
**Author:** ![Lance\_Turbo](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/lance_turbo/32/6156_2.png) [@Lance\_Turbo](https://boards.straightdope.com/u/Lance_Turbo)\ **Post date:** [May 25, 2020, 7:28pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/37 "2020-05-25T19:28:13Z")
There’s about a week left and there are only three entries thus far, although a couple of other posters dropped in to concede defeat by stating that they could not possibly enter such a contest until October. Here’s where we stand now including some minor updates to my entry. Others are free to update their entries as well. ```auto Race Lance Turbo septimus Calavera Alabama 2.7% 0.0% 0.0% Alaska 7.3% 0.0% 0.0% Arizona (special) 89.6% 80.0% 100.0% Colorado 96.2% 90.0% 100.0% Georgia 26.0% 20.0% 0.0% Georgia (special) 30.9% 30.0% 0.0% Iowa 25.2% 20.0% 0.0% Kansas 19.3% 20.0% 0.0% Kentucky 9.0% 0.0% 0.0% Maine 71.0% 55.0% 100.0% Michigan 87.2% 80.0% 100.0% Montana 44.6% 35.0% 0.0% North Carolina 67.1% 45.0% 0.0% ``` So we have prediction market, pessimistic prediction market, and pessimistic prediction market rounded to the nearest integer ;). It looks like some form of prediction market derived entry is going to win this thing unless we get some poll or pundit based entries. Deadline is May 31 11:59:59 PM CDT. ---
**Author:** ![septimus](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/septimus/32/410_2.png) [@septimus](https://boards.straightdope.com/u/septimus)\ **Post date:** [May 25, 2020, 9:52pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/38 "2020-05-25T21:52:14Z")
If we get a surge of new entries before month’s end, I will probably move to an all-integer submission, like **Calavera**. I’ll want whichever such submission seems most likely among the all-integer cases which have _not_ been submitted by anyone else. @ **Lance** — I’m not sure what your approach is here, but even supposing prediction perfection — that your submission minimizes expected total squared error — that is NOT the solution contestants should seek, _given the Winner-Take-All scoring._ ---
**Author:** ![Oredigger77](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/oredigger77/32/3181_2.png) [@Oredigger77](https://boards.straightdope.com/u/Oredigger77)\ **Post date:** [May 25, 2020, 11:39pm UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/39 "2020-05-25T23:39:11Z")
Alabama 1 Alaska 0 Arizona (special) 1 Colorado 1 Georgia 0 Georgia (special) 0 Iowa 0 Kansas 0 Kentucky 1 Maine 1 Michigan 1 Montana 1 North Carolina 1 I’m taking a differently course since you guys are all clustered. I’ll finish either first or last. Still the Dem odds though. ---
**Author:** ![Lance\_Turbo](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/lance_turbo/32/6156_2.png) [@Lance\_Turbo](https://boards.straightdope.com/u/Lance_Turbo)\ **Post date:** [May 26, 2020, 1:53am UTC](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889/40 "2020-05-26T01:53:14Z")
> [@septimus](#): > > […]_given the Winner-Take-All scoring._ There’s nothing to take (per board rules) so there’s no reason to treat this contest as winner take all. Integer solutions aren’t really in the spirit of the contest, but I’m not going to disallow them. However, we don’t really learn anything from integer solutions. Sure guessing each race correctly would result in a Brier score of zero, but is that a reproducible method? Is picking the integer solution that you find most likely and hasn’t been picked by anyone else something that something that can be applied to future elections if it wins this contest? My goal for this thread (and future threads like it) is to determine which probabilistic forecasts produced by the SDMB were most accurate 5 months out, 4 months out, etc. But it’s beginning to look like my PredictIt derived forecast will win that distinction by being the only probabilistic forecast produced. There’s nothing wrong with a contest to see who can pick the most winners, but that wasn’t the intent of this thread. If everyone but me picks distinct integer solutions, I’ll probably finish either first or second. That’s a perfectly fine result for my method as far as I’m concerned. If it’s only competition is guessing everything correctly, I’ll be happy. [Previous page](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889.md?page=1) [Next page](https://boards.straightdope.com/t/may-2020-probabilistic-senate-prediction-competition/853889.md?page=3)