6 Minutes vs 5 Minutes (Data about Smash Attack 3 and Smash Attack 5)

[Jehtt]

I have been collecting data over the past few days about the average match time in Smash Attack 5 and Smash Attack 3. Each one was very similar in terms of rulesets and players, with the difference being the timer. Smash Attack 5 used 5 minutes, while Smash Attack 3 used 6.

Here is a link to the raw spreadsheet.
https://docs.google.com/spreadsheets/d/19HLdtDPYj7A94tIsX73YPimDcRpegGfEt5E0ZFR6-M4/edit?usp=sharing

Now, this is by no means a perfect experiment. There are variables I could not control, specifically the players present. Match times are obviously very different depending on the characters.

Also, I used Smash Attack 5 instead of 4 because I was misled into believing Smash Attack 4 had 6, when in reality it was 5. The first part of Smash Attack 4's stream used 5 minutes, but the ruleset was later corrected. I would take more data, but I wanted to get this out before EVO.

I am not particularly good at extrapolating from data (nor am I well versed in the way of excel), but here are some interesting things:

• Average time in 5 MINUTES: 3:02
• Average time in 6 MINUTES: 2:51
• Number of timeouts in 5 MINUTES: 3
• Number of timeouts in 6 MINUTES: 0
• For about 80% of matches, the player who took the first stock also won the match.
• Sheik is the most common character, appearing 52 times across 18 sets.
• Smashville is the most common stage, appearing 46 times.

EDIT: MORE DATA

Winter Brawl 9 (Atlantic North)
2 stocks, 6 minutes
47 games logged
Average time: 2:36
Standard deviation: 0:56

https://docs.google.com/spreadsheets/d/1cjIGgR0G7YXonFN1910LpYmEA0FzCElxlOjvSEMooPg/edit?usp=sharing

Though the average times may not be too different, the number of timeouts is considerable. I feel it is ideal for 2 stock 6 minutes to be the norm, unless 3 stock can take the spotlight. For a little Anecdote, I was playing friendlies with a Sonic today. Mega Man vs Sonic is one of those matchups like Kirby vs Sheik, Sheik vs Villager, etc... that just needs that extra 1 minute to come to a proper conclusion. I took stock 2 at 5:04.

 

[Fig Newton]

Clearly there is going to be a few things that can't be controlled or missed when conducting an experiment, however I do believe this is enough to prove to Mr. Wizard that by adding just an extra minute to the EVO ruleset, we can have more if not all games to not go over time. The deadline for the finalization of the ruleset is on the 27th of this month so this must be done rather quickly.

 

[Zurreal]

What I'd really like to see is the standard deviation of each set. I believe most matches will take the same time except for in cases where one player is deliberately attempting to stall out the clock, creating a few outliers. If this were the case the medians of the two sets would be the same.

 

[thehard]

Nice work. Smash Attack 4 was a 6 minute timer though.

 

[TSM ZeRo]

This has to be spread around. I don't think 5 minutes is the way to go with a 2 stocks format.

 

[Jehtt]

Nice work. Smash Attack 4 was a 6 minute timer though.

That is why I used Smash Attack 3, not 4.

 

[thehard]

That is why I used Smash Attack 3, not 4.

But 3 and 4 were the same time?

 

[Jehtt]

But 3 and 4 were the same time?

No they weren't. Smash attack 4 had 5 minutes, at least it seems so. I used Smash Attack 5 (which definitely had a 5 minute time limit) and Smash Attack 3 (6 minute time limit).

EDIT: There seems to be an interesting issue with Smash Attack 4. The first part of the tournament on stream seemed to be using 5 minutes, when it later turned to 6. This would have made for some confusion with the data.

 

[thehard]

No they weren't. Smash attack 4 had 5 minutes, at least it seems so. I used Smash Attack 5 (which definitely had a 5 minute time limit) and Smash Attack 3 (6 minute time limit).

EDIT: There seems to be an interesting issue with Smash Attack 4. The first part of the tournament on stream seemed to be using 5 minutes, when it later turned to 6. This would have made for some confusion with the data.

Interesting. I have my own (imperfect) data comparisons between Customs SA4 and SA5 (6 minute timer [this was the 2nd part of the tournament] and 5 minute)
http://smashboards.com/threads/competitive-smash-ruleset-discussion.368605/page-29#post-18774017
They seem to mirror your own. SA5 had higher avg. match time and more timeouts.

 

[Shadó Chimera / シャドーキメラ]

Just wondering, how many matches on the 6 minute timer lasted 5+ minutes?

 

[Jehtt]

Just wondering, how many matches on the 6 minute timer lasted 5+ minutes?

3. Two of those games involved Sheik, who also happened to be involved with a timeout under the 5 minute ruleset.

This is purely anecdotal, but I did notice Sheik's matches seemed to go on for a while. I imagine it's her lack of kill moves.

 

[Ansou]

This is some interesting data. I guess this shows how the timer affects players' mindsets.

 

[Piford]

The most interesting match I see is that 3 matches went over 5 minutes in Smash attack 3. They were Sheik vs Sheik, Sheik vs Diddy Kong, and Mario vs Kirby. What's interesting to me is that all of those matchups are usually offensively oriented and not focused on camping out. While I didn't see the clock, I'd make the assumption that those matches needed the extra minute to to prevent timeouts, but to actually finish the match.

On the flip side, the timeouts at Smash Attack 3 were Villager vs Sonic, Villager vs Sheik, and Kirby vs Sheik. Now Villager vs Sonic is bound to be a timeout and was only made easier because of the timer. Villager vs Sheik is a match I feel would be less likely to go to time, or above 5 minutes, with a longer timer. Kirby vs Sheik is again a very offensive matchup and I feel that's one that probably wasn't really focus on camping out and likely just didn't really have sufficient time to play out the match.

So the longer matches at SA3 seemed to be because they needed the time to have a match going, while the timeouts at SA5 seemed to be a mix of people actually trying to timeout and not being able to finish a match.

 

[Ansou]

The most interesting match I see is that 3 matches went over 5 minutes in Smash attack 3. They were Sheik vs Sheik, Sheik vs Diddy Kong, and Mario vs Kirby. What's interesting to me is that all of those matchups are usually offensively oriented and not focused on camping out. While I didn't see the clock, I'd make the assumption that those matches needed the extra minute to to prevent timeouts, but to actually finish the match.

On the flip side, the timeouts at Smash Attack 3 were Villager vs Sonic, Villager vs Sheik, and Kirby vs Sheik. Now Villager vs Sonic is bound to be a timeout and was only made easier because of the timer. Villager vs Sheik is a match I feel would be less likely to go to time, or above 5 minutes, with a longer timer. Kirby vs Sheik is again a very offensive matchup and I feel that's one that probably wasn't really focus on camping out and likely just didn't really have sufficient time to play out the match.

So the longer matches at SA3 seemed to be because they needed the time to have a match going, while the timeouts at SA5 seemed to be a mix of people actually trying to timeout and not being able to finish a match.

I might not have watched the matches, but my experience in the Kirby vs. Sheik matchup is that both characters have a hard time approaching eachother and especially getting the KO, which kind of leaves the situation in a stalemate.

 

[Overswarm]

IF ONLY SOMEONE COULD HAVE PREDICTED.

https://www.youtube.com/watch?v=whAmI2HdBqw


Not 6, 8. For 2 or 3 stock, 8 minutes.

THE ACTUAL NUMBER ON THE TIMER IS IRRELEVANT. THE ONLY THING THAT MATTERS IS IF IT CAN GET TO THE LAST 60 SECONDS ON THE CLOCK.

3 stock, 8 minutes. If you don't have enough time to run an event, don't hamstring the event. Cap entrants.

3 stock 8 minutes takes about an hour more total for a 60 man tournament with 8 setups. We've run 3 stock with no issues in the midwest, europe has run 3 stock with no issues. It is a cakewalk.

 

[BestTeaMaker]

@ Overswarm your profile title is quite apt for this thread.

 

[RobinOnDrugs]

This should be shown to EvilMrWizard.

 

[TaFoKiNtS]

Hmm, interesting statistics, but there were 3 matches that lasted above 5 minutes with a 6 minute time limit. These would have been considered "timeout" scenarios in 5 minute limit matches, but aren't in 6 minutes. It'd be interesting to get some data on time with respect to characters and stages, but that would take a significant amount of data points.

 

[thehard]

Mr. Wizard's biggest fear is that every 6 minute match will run the full 6 minutes. If we can somehow convince him there is no chance of this happening...

Also, what Overswarm said (says).

 

[Overswarm]

The biggest timesink in any tournament is out-of-game time. Scheduling matches, getting people to stations, etc., etc.

If you want to reduce tournament time then that's where you do it.

If you're worried about in-game time, I can almost guarantee you that you'll have at least one set per tournament that nearly goes to time for every match in the set. But overall, it's a matter of "should I camp the timer". If you have 5 minutes left one the clock and both have one stock, your 13% lead doesn't mean that much. If there's 30 secnods left on the clock, you've basically one the game.

Increase timer, get shorter games.

 

[Piford]

Hmm, interesting statistics, but there were 3 matches that lasted above 5 minutes with a 6 minute time limit. These would have been considered "timeout" scenarios in 5 minute limit matches, but aren't in 6 minutes. It'd be interesting to get some data on time with respect to characters and stages, but that would take a significant amount of data points.

The data is in the spreadsheet, but it'd take some work doing the math.

 

[TaFoKiNtS]

Mr. Wizard's biggest fear is that every 6 minute match will run the full 6 minutes. If we can somehow convince him there is no chance of this happening...

Also, what Overswarm said (says).

The thing is Mr. Wizard looks at this data and sees match times at 5:23 or 5:44 or w/e and it doesn't really help the cause of 6 minute timers. He's looking at worst case scenarios for Evo (which every good TO should) and he sees the potential of 18:00 min gameplay for sets as opposed to 15:00 max gameplay sets for BO3. Even if 6 min timers has a lower avg time, the fact that you could potentially increase sets upwards of 3 minutes for a tourney that has 2000 sets is risky.

I also agree with Overswarm on reducing out-of-game time, but that isn't really the most enforceable rule.

 

[王女 クマトラ]

"For about 80% of matches, the player who took the first stock also won the match."

This is why I really dislike 2 stock in general. Although, if I have to deal with it, I definitely prefer 2 stock 6 minute to 2 stock 5.

 

[M@v]

For about 80% of matches, the player who took the first stock also won the match.

And this is very likely due to the rage mechanic. Whoever takes the first stock will more likely than not have a lot of rage from being at high percent (unless there is an early first kill) . That gives them a big advantage once the opponents invincibility wears off on their second stock. It can force the person behind to play different, whether its "Crap, I need to kill him NOW before i die early!" or "I need to be SUPER careful and play passive to avoid taking too much damage!" In either thought process, the person ahead can (and will) abuse that to get even farther ahead.

Besides Rage, 2 stock just doesn't give that much wiggle room for a comeback. By the time you've learned the opponent that game, you are at 80% on your last stock, and basically have to wait for the next match (if you even get one) to utilize what you've learned.

 

[RedNova]

Mr. Wizard's biggest fear is that every 6 minute match will run the full 6 minutes. If we can somehow convince him there is no chance of this happening...

Also, what Overswarm said (says).

Maybe this is an unpopular opinion, and of course it could not be implemented at evo, but if we could figure a way that, if the match goes to timeout, there would be a 50/50 chance of either side to win.

After thinking around a bit, it does make a little sense. With no guaranteed chance of winner after a timeout, players will make the most out of the timer, and discourage stalling. Maybe it could be anticompetitive, but with the extended timer and the possiblity to come back in losers, maybe it would be worth a shot

(Pls don't hate me)

 

[M@v]

Maybe this is an unpopular opinion, and of course it could not be implemented at evo, but if we could figure a way that, if the match goes to timeout, there would be a 50/50 chance of either side to win.



After thinking around a bit, it does make a little sense. With no guaranteed chance of winner after a timeout, players will make the most out of the timer, and discourage stalling. Maybe it could be anticompetitive, but with the extended timer and the possiblity to come back in losers, maybe it would be worth a shot

(Pls don't hate me)

Its not bad to think outside the box, but unfortunately there's a fatal flaw with that idea . Say I'm behind and I'm clearly going to lose the match, I'd be encouraged to be the one to run away and try to force the match to time. Why? If I know I'm not killing him or getting the percent lead, I have nothing to lose by forcing the match to time and potentially getting a win by some tie break mechanic. Its the same reason we don't play out a tie breaker unless it is a perfect tie, and even then we make the players do a 1 stock rematch.

 

[RedNova]

Its not bad to think outside the box, but unfortunately there's a fatal flaw with that idea . Say I'm behind and I'm clearly going to lose the match, I'd be encouraged to be the one to run away and try to force the match to time. Why? If I know I'm not killing him or getting the percent lead, I have nothing to lose by forcing the match to time and potentially getting a win by some tie break mechanic. Its the same reason we don't play outa tie breaker unless it is a perfect tie, and even then we make the players do a 1 stock rematch.

Yeah, that was one of the problems I still can't fix. Maybe the extra time will help a bit, tho. If the first stock is taken at around 2 min, I don't think one could camp for the full 4 mins left. Also, the tournament where I saw that rule listed the RNG tie breaker in place of the percent lead (the stock lead was still there).

I hope someone else can come with a solution, because with no stalling/camping tactics, maybe we could see a bigger stage list.

 

[Overswarm]

There's 3 ways to fix the timer.

1) Increase the timer to a high enough level to where it is unlikely to get to a point where a timeout is a viable chance at winning.
2) Turn off the timer
3) Arbitrary rulings

The third is crazy and requires a ref (see: shot clock), the second is the best except a tournament just might not end because of one guy with an awesome sense of humor, and the first is easily done by saying "8 minute timer".

 

[FreshYarn43]

Yup. This demonstrates what I've been saying for the longest time, 5 minutes isn't enough in some match ups. We all saw the Pac-man vs. rosaluma set at apex, that was horrid to watch. EVO needs to be the one to pilot new smash rules, this game is so young and we can't predict what future rule sets will be.

 

[popsofctown]

80% of first stocks resulting in a win makes me kinda worried about the warping effect rage has. Given that seeding puts a lot of really good players against really bad players it's probably just people getting totally bodied, so I should probably worry less. It's hard to say for sure how much rage snowballing messes with things or not.

I kinda wish we had 1 stock Bo7 to disable rage's snowball effects entirely, but no one will go for that, sadly. It also mitigates SD's and spikes' disproportionate impact and allows safer experimentation with character swapping. In my dreams only...

 

[M@v]

There's 3 ways to fix the timer.

1) Increase the timer to a high enough level to where it is unlikely to get to a point where a timeout is a viable chance at winning.
2) Turn off the timer
3) Arbitrary rulings

The third is crazy and requires a ref (see: shot clock), the second is the best except a tournament just might not end because of one guy with an awesome sense of humor, and the first is easily done by saying "8 minute timer".

6 mins is fine tbh. I think i've only seen one maybe 2 6 minute matches go to time so far since wii u came out.

 

[thehard]

I don't think Rage has a whole lot to do with it, I think that losing one (1) stock meaning you are now at a 50% disadvantage has more to do with the "winner takes first stock" trend. Even if you even things out, your opponent has likely tacked on extra credit. Compare this to a 3 stock, 33.3% disadvantage.

 

[Overswarm]

6 mins is fine tbh. I think i've only seen one maybe 2 6 minute matches go to time so far since wii u came out.

If you aren't playing 3 stock you aren't playing smash 4

 

[thehard]

This is kinda ridiculous now

No response from Mr. Wizard after weeks of every major Smash player/community leader tweeting him out BEGGING for a 6 minute timer. ZeRo, Keitaro, Max Ketchum, several dozen more...

Who is the tournament even for at this point? There's even the possibility of no Miis, just because.

I hope this doesn't come off as entitled but I'm mildly frustrated

 

[Jehtt]

If anyone here wants to tweet this thread to Mr. Wizard, I urge you to do so. Even if it doesn't affect him, there is no point in not trying to get him to see it.

 

[RedNova]

If anyone here wants to tweet this thread to Mr. Wizard, I urge you to do so. Even if it doesn't affect him, there is no point in not trying to get him to see it.

If we are gonna do that, maybe you should update the main post with the possible modifications to the ruleset

 

[[Deuce]]

I am a proponent of 6 minutes, or rather, 3 stock 8 minute matches, but I just wanted to tell you there's not enough data presented to fully determine whether the timer difference is statistically significant or not.

Sample size (matches), number of entrants, standard deviation (for BOTH tournies) is missing. 10 seconds difference with a minimal sample is only enough to justify an argument for the results being inconclusive

 

[thehard]

I am a proponent of 6 minutes, or rather, 3 stock 8 minute matches, but I just wanted to tell you there's not enough data presented to fully determine whether the timer difference is statistically significant or not.

Sample size (matches), number of entrants, standard deviation (for BOTH tournies) is missing. 10 seconds difference with a minimal sample is only enough to justify an argument for the results being inconclusive

I did my own imperfect number crunching for different SAs and came up with the same trend: 6 minute timers cause matches to be shorter on average and lead to less timeouts. Dabuz says 5 minutes does not allow for some matchups to naturally play out. This isn't a one-off. Many players attest to the same logic.

 

[Jehtt]

I added data from another tournament to the main post, courtesy of @ Lanzoma . He has the spreadsheet with the data on it, which he will post soon. https://docs.google.com/spreadsheets/d/1cjIGgR0G7YXonFN1910LpYmEA0FzCElxlOjvSEMooPg/edit?usp=sharing

 

[Lanzoma]

Hi. I went ahead and calculated standard deviation from the Smash Attack tourneys, plus added my own data from Winter Brawl 9. You can see the raw data here: https://docs.google.com/spreadsheets/d/1cjIGgR0G7YXonFN1910LpYmEA0FzCElxlOjvSEMooPg/edit?usp=sharing

Smash Attack 5 (Chicago)
2 stocks, 5 minutes
52 games logged
Average time: 3:02
Standard deviation: 0:54

Winter Brawl 9 (Atlantic North)
2 stocks, 6 minutes
47 games logged
Average time: 2:36
Standard deviation: 0:56

Smash Attack 3 (Chicago)
2 stocks, 6 minutes
58 games logged
Average time: 2:51
Standard deviation: 0:58

The only tournament that had ANY TIMEOUTS was Smash Attack 5 with 5 minutes ruleset, which had 3 timeouts. There's very, very, VERY few matches that ever go above 4 minutes during a 6 minute ruleset.