+- International Simulation Football League (https://forums.sim-football.com)
+-- Forum: Community (https://forums.sim-football.com/forumdisplay.php?fid=5)
+--- Forum: Media (https://forums.sim-football.com/forumdisplay.php?fid=37)
+---- Forum: Graded Articles (https://forums.sim-football.com/forumdisplay.php?fid=38)
+---- Thread: *NFL injury rate vs. ISFL teams (/showthread.php?tid=47091)
*NFL injury rate vs. ISFL teams - Troen - 08-20-2023
Seeing as it's now the NFL preseason and injuries have started taking the annual toll on the league, I thought it could be interesting to consider what injuries might look like in the ISFL if they happened in the current environment at a rate similar to the NFL.
(As an aside, I previous thought about if the ISFL should have injuries and concluded that we probably should not for user experience reasons; this is looking at what might happen if we did anyway)
For my source on injury data, I used this football outsiders article as my starting point (they explain the source data in this article - one note is that it's only looking at weeks of regular reason lost). That was analyzing injury data over 2000-2014 and I derived the following per-week rate for a team's injuries that result in losing playing time based on the top injury per position charts (rounded to 4 decimal points):
QB: 0.0774
RB/FB: 0.2078
OL: 0.3382
WR/TE: 0.4892
DL: 0.3531
LB: 0.3426
DB: 0.4776
K/P: no chart so 0?
Giant disclaimer: I'm reading the bar chart to get total and average weeks lost and then deriving count as total / average. Since this is "reading" the chart and not using the underlying numbers, the actual rates are going to be very imprecise - eg for the OL other knee injury, being off a tenth of a week lost on average changes the derived injury count by a couple dozen. I think it still should be close enough that I can produce approximately plausible stats to do the analysis on, though if I were writing a science paper and not a media article I'd expect complaints about using such an imprecise methodology. But what's the alternative, actually paying to get the data or something? Clearly ridiculous. Additionally, the source data is not purely injuries that happen during play - it includes any reports from training during the week, for one, and so it is not a model of how many injuries happen during the actual game.
With that aside, I'll now try to apply those rates to the ISFL. The first tricky part in doing so is that an ISFL team roster is set up very differently than an NFL team, with much less in the way of backups, rotational, and devoted special teams players. Based on the budget sheet as of August 19th, the average team has:
QB: 1.07
RB: 1.86
OL: 5.07
WR/TE: 4.29
DL: 3.5
LB: 3.5
DB: 5.36
KP: 1
For the purposes of looking at injuries, I'm going to completely ignore the depth issue to start - since teams still need to field 11 players per play, I'll basically assume that applying the injury rate works to continue on through any backups the team would have in place. That's not a perfect match to the ISFL roster construction and additionally not fully representative of how a game plan would change if there were injuries (though I don't know if the sim can actually support that kind of dynamic update anyway), but that at least gives a starting point.
Taking the per-week injury rate vs. the average roster, the expected injuries each team would see per position for a week/season:
QB: 0.083/week | 1.328/season 0.040/week | 0.643/season
RB: 0.386/week | 6.175/season 0.187/week | 2.988/season
OL: 1.715/week | 27.439/season 0.830/game | 13.276/season
WR/TE: 2.097/week | 33.545/season 1.014/game | 16.230/season
DL: 1.236/week | 19.775/season 0.598/game | 9.568/season
LB: 1.199/week | 19.184/season 0.580/game | 9.282/season
DB: 2.559/week | 40.937/season 1.238/game | 19.807/season
KP: no data so 0 I guess?
Total: 9.274/week | 148.383/season 4.487/game | 71.792/season
- Edit: after sleeping on it, I realize that I made a unit conversion mistake here - the injury rate is per position per week and when using the number of people per position in the ISFL I need to scale that vs. the average roster size in the NFL. I'm going to use 53 as the NFL roster size vs. ISFL average of ~25.65, which reduces the rate by just over half. I don't think the reduced numbers drastically change the conclusion though it could inform deciding how big a pool we'd need to be able to draw from to manage.
This is an estimate for "number of injuries that result in lost games" and so doesn't take into account the distribution of season ending vs. shorter 1-2 week types. It also includes injuries that happen between games for NFL teams - accidents in training and practice. But, I think it does show a little bit of the kind of depth we'd need to have for rosters, assuming the sim's injury rate is along the lines of the NFL rates - I'd guess at least 50% more people than the rosters we see now to be able to consistently have enough people, and that's just assuming that the team only has the average number of injuries. If one team gets unlucky, it could exhaust depth at a given position and then I actually don't know what the sim would do if it were during a game.
Moving on from the general impact, how could the league actually handle injuries? I think the first question is seeing if there would be some kind of additional sim between the games to be the time when non-game injuries can happen, or if they would instead all happen during the sims we have now. The decision is more of a process question - when and how will GMs be able to react to injuries and adjust the depth chart? The answer also would depend on what GMs can do to get replacements, but I'm going to talk more about that in a later section. Having some sort of additional simulation event would increase the load on the people actually running the sim and updating the index, changing from needing to do something 3 times a week to presumably 6. That would consequentially require GMs to react and potentially submit updates basically every day. Beyond that, rosters are generally fairly stable through the season now. Changing to a system which can have drastic changes in your best player (without the ability to plan in advance as in a trade) could create unexpected needs for GMs to test the team's strategies with the possible new player configurations. Plus, depending on how replacement players could be acquired, any negotiations could take time. The concerns about needing to do a lot of work unexpectedly remain if the injury simulation happen at the same time as the existing sim, but it does potentially keep the number of times GMs and the sim team have to do work down.
For options on what the GMs can do to get replacements when there are injuries, I went over those more in my previous injury article. Fundamentally, though, teams will need to have more players on the rosters if injuries happen during games, and there will need to be pools of players available to add to rosters as replacements for injuries. If those replacement players have agency in terms of negotiation or preference, that will add a lot of midseason complexity, especially if multiple teams are trying to get a given player. The main way I can think of to avoid this is to have some sort of automatic bot system so that teams are have sufficient rosters.
Taking these considerations for level of work on the GMs, I don't think that injuries happening by chance and require GM work to replace could ever happen for purely organizational/level of effort considerations (without swapping the sim or how GMs manage teams drastically). Random injuries that result in the team getting automatic bot replacements might be a plausibly small amount of GM effort, but if your 1500 TPE QB gets injured and the replacement is significantly worse I imagine that will still result in needing to change strategies a lot. Are there other ways injuries could exist without ruinous amounts of unpredictable GM work? Making the GM work be small or predictable qualifies by definition, but what are are some ways to do that? And how, getting back to the big reason we don't have them now, might those alternatives not alienate users?
Making GM work small: If injuries were more like small debuffs rather than complete removal of the player from being able to play, it might reduce the GM work needed. At one extreme, if injuries accumulated within the game but all went away afterwards, all GM management of injuries would be encapsulated within normal game planning (though obviously that would require sim support and thus not work now). Injuries could also be along the lines of the player losing 5% of TPE in a given stat for the duration of the injury, where by "5%" I mean "some amount small enough to not require completely resetting the game plans, but ideally large enough to actually matter". I feel like this ultimately ends up being either too small to feel like an injury - basically being like a sack or turnover - or ends up failing at reducing the GM work. And no matter which approach is involved I imagine injuries to your player will still just feel bad.
Making GM work predictable: What if instead of relying on the sim's randomness, injuries were known in advance? If GMs knew at the start of the season that certain players weren't going to be available and could plan around it up front, that might get the work required down to a reasonable amount by making it able to all be done in the offseason rather than as an interruption during the regular season. If the injuries were still determined randomly per team it could still end up being a huge amount of work - if you end up unlucky and get 3 season-ending QB injuries, it's going to be a really big task to try and find 4 QBs good enough to help you win (see also: 49ers). That could be avoided by doing other ways of assigning injuries - a point buy system (especially with some ability to roll points over some) could let teams be strategic though it would probably need a lot of tuning, or some sort of draft where teams pick from a pool of injuries (though again being stuck with QB season-ending injury will probably not feel good). I do think giving teams control over a selection of injuries in some way actually could work with the goal of user engagement (or at least, not be as negative as random injuries). Things like trying to have a bunch of injuries on the week you're playing a team you're hoping will be weak could add a new strategic dimension, or letting a rookie you didn't really need yet be injured for the season could let a player who wouldn't have gotten many stats as a 4th WR still help out by clearing the way for the team to do better overall.
Even with a few mitigations or upsides where injuries don't feel super random or a an element of a team's strategy, would it actually make the league better? I feel like it's interesting in increasing the verisimilitude of the sim matching things we see in the NFL itself, but I don't think it's going to feel good getting pressured to let your rookie player have an injury when you want to have them play, or feeling like trying to help your team wins is best done by reducing the chances of hitting some player stat goal that you had. Now don't get me wrong, for those who are really into the game theory and strategy elements of the league then having another dimension to interact with could be interesting and get them more involved, possibly moving from just being a player to instead helping with team planning. My intuition is that the number of those people who aren't already working on the team strategy is low, and since making a change in this direction and just hoping it will get people more involved is risky I'd say it's probably not worth doing. However, if we for some reason are definitely 100% going to add injuries but don't want pure random, I could see an approach like this being one of the more fair ways to do it.
As a smaller note, injuries (and resulting playtime lost) would also be another factor to change comparisons of players from different eras, though the sim switch, growth in league size, and various schedule changes complicate any 1:1 comparisons already anyway. I don't think that's a consideration large enough to keep at the core of a discussion around adding injuries, though there are certainly engaged flamewars debates for eg how good Griffey Jr. would have been without his injury troubles that keep coming up in sports discussion.
So my conclusion, for anyone who skipped here, is that trying to have league injuries anywhere around the NFL rate would drastically complicate the lives of GMs and that I don't see anything that has changed with regards to the part where injuries happening to your player feels bad and likely hurts the league's health in the long term. Allowing GMs to strategically select injuries rather than having them be randomly could be interesting, but I don't think it's interesting enough to overcome the feels bad personally part, but I'm definitely open to discuss/investigate other ideas people might have.