Thoughts on Group Knockout/Strikes format w/ 3 strikes per group?


Swiss should be fairly predictable (but with some variance). I haven’t run any models though – I’m generally not smart enough for that kind of work :slight_smile:

I would apply the “Sharpe Rule” from above: If there are four or fewer players left and they all have the same amount of strikes, the next round will be the final round.

Balanced don’t have to be longer tournaments. We ran SFPD B & C division finals with 20 and 16 players respectively and balanced pairings (10 strikes). Both took 11 rounds.

The next week a monthly tournament was held with 22 players, swiss pairings and only 9 strikes. It also went 11 rounds. :man_shrugging:


I did some manual simulations on # of rounds with varying player field sizes, and varying # of strikes. My aim here was a relatively quick evening tourney to last between 4-5 hours.

I tried to attach the spreadsheet, but was reminded that I can’t attach .xls files here. Whoops. If anyone wants the file, please message me with your email address, and I’ll send to you.

To get a tourney of roughly 8-10 rounds… My conclusions:

<= # Players Strikes
64 6
48 6
36 6
28 7
20 8


So, is this how long it would take with all final rounds played? Or modified as people were mentioning?


Just as an additional data point, I ran a 9 strike progressive knockout last night with 16 players. There were 10 rounds with average times around 20 minutes.

It came down to 3 players all with 8 strikes, so no need for manual strikes given. Total running time was 3 hours.


I ran a 22 person 10 strike a couple weeks ago. Average round time was 15-20 mins. Final three all had 9 strikes so no adjusting needed. Right at 3 hours.


Hmm, maybe I could expose round data more generally for knockout tournaments…


My analysis didn’t use any accelerated strikes for the final two players. Granted, I didn’t have a sophisticated program built to run bunches of iterations of each. It was a lot of randomly assigning finishing positions within each group, re-sorting in Excel based on new cumulative strikes while excluding those that had reached the strike threshold, Swiss match-ups, repeat.


I just wanted to share that we ran a progressive strikes event last night, and it seems that everyone had a lot of fun with it.

We had 21 people and we did 8 strikes.

21 seemed to be an ideal number for a few reasons. We had three 4-Player groups and three 3-Player groups for the first four rounds before anyone got eliminated. The fact that Progressive only gives out two strikes in a three-player group, coupled with the fact that we had multiple 3-Player groups, means that people who played poorly were in 3-Player groups, and we didn’t eliminate anyone until Round 4. After that, the bottom fell out and we started losing lots of people very quickly:

Round 1 - 21 Players
Round 2 - 21 Players
Round 3 - 21 Players
Round 4 - 21 Players
Round 5 - 20 Players (-1)
Round 6 - 14 Players (-6)
Round 7 - 8 Players (-6)
Round 8 - 4 Players (-4)
Round 9 - 2 Players (-2)

We didn’t need to “move up” finals because the two finalists had 7 Strikes and 4 Strikes.

It took us 2 1/2 hours to get through, and that was with attempting to limit to faster-playing games.

For a one-off I personally like this format a LOT more than head-to-head strikes, and I like it marginally more than group knockout with strikes going to the bottom two. It keeps people involved in the event for longer but the gauntlet seems to fall very quickly in the later stages.

Yes! I would love to see this.


Completely agree. I tend to favor formats that use more data, and progressive does that better than standard knockouts. The fact that there are 4 outcomes in a match rather than just 2 makes it a better measure of performance, plus it’s still simple to run and still whittles down the field in a reasonable amount time. Hoping this becomes the new normal where knockouts are held!


I am definitely going to try this at our next tournament. So in a 3 player group does 2nd place get 1 strike and 3rd get 2?


Exactly. Just think of it as a strike for every player you lose to in a match.


I’ve played a lot in these and everyone loves it more than the usual bottom two get strikes - it really balances out the 3-person groups…


@coreyhulse @ChubbyGoomba I threw up some raw numbers at

Lots of scrolling involved, but you can see the average, median, max, min number of rounds for each type of knockout tournament. It’s only looking at completed, non-test tournaments, but there will still be some test tournament data in there because not all people classify their test tournaments as such.

I also filtered out tournament that only have a single round because that’s bound to be a test or a tournament where something went wrong. But still pay attention to the Min column. The Tournaments column shows how many tournaments were used to calculate the other numbers. For some combinations there’s only a single tournament.

There is not much data on progressive strikes since there haven’t been many tournaments using that format.

Let me know if you’d like the player count breakdown done differently. Only limitation is that the breakdown has to be identical for all the tournament types since this is all generated programmatically.


…and if you reload Median is calculated correctly (don’t attempt math before breakfast)


My reaction reading the post:

This is super exciting stuff, and the type of thing that can help give a guide to directors for number of strikes to use, especially if you have a view on how long an average round takes.

I’ll have a think over the weekend if other views would be useful. Off-handedly (I know this is a large ask), but for each type of tournament, a Histogram with # of Rounds on X, Frequency on Y, giving a view of how many rounds a particular format went (with mean, median, min, max as supporting numbers in the columns as they currently exist).

Thanks for putting this together!


@coreyhulse Head to to see some really ugly auto-generated histograms. They should give you what you want though :slight_smile:


This is great! Thanks for putting this together.

For the “Group Knockout” category, is this just “bottom two get strikes” or a combination of the various options? I know I had a Group Knockout Event with 12 people where only the bottom person received a strike. It went 20 rounds.


The group knockout are bundling all the different strike options together. It explains the wide range of round counts there. I don’t really have time to split them out :frowning:


So I was thinking of running a progressive strikes tournament, and Luke M suggested keefer scoring (02, 012, 0112)
I thought it was great, and contacted Andreas about a month ago to see if if he would support it. Unfortunately he said it was too hard to add and not as simple as plugging in some numbers.

The tournament is next week so last night I ran a simulation on matchplay and manually edited every ones scores from the normal 0123. Well I broke Matchplay math and the result of this is that Andreas has added the 0112 scoring to matchplay!!! How good is that!
He has called it Aussie strikes because I never explained where the idea came from, but I have asked him to rename it to ‘keefer strikes’.

I ran some simulations for the tournament with 5 strikes. I have no idea how many people will come

40 people v1

round people
1 40
2 40
3 40
4 30
5 24
6 14
7 4

40 people v2

round people
1 40
2 40
3 40
4 37
5 30
6 13
7 4

20 people

round people
1 20
2 20
3 20
4 20
5 18
6 6
7 3

60 people

round people
1 60
2 60
3 60
4 60
5 55
6 17
7 5
8 2

These are just random button clicks, but you can see the the results are pretty consistent. The tournament won’t really have the ability to drag on and round 5 and 6 are like the bloodbath rounds where all the action happens. More people in for longer means more fun for more people!

Thank you Keith
Thank you Luke
Thank you Andreas for all of your hard work for the pinball community

I will report back next week with how it all went!


Awesome! I will definitely give this a trial run for my next strikes format tournament in a couple weeks.