Scorinators (OOC Discussion)

[Andso Mangis]

xscorinate doesn't work for me are their other ways to scorinate game

[Tumbra]

xscorinate doesn't work for me are their other ways to scorinate game

Hi! You're going to have to go into more detail as to how it doesn't work for you. Could you do that for us so we can help diagnose your issue? Alternatively, try following the instructions in this post, which I've quoted below. You can click on the green "wrote" text next to Graintfjall's name to access the original post.

1. When you open it you should have a page like this:

(Image)

2. Click on New Event on the left and choose your sport. For now I'll use Association Football, SQIS Formula. Choose the options you want such as home advantage on or off, and whether to use additive or multiplicative modifiers. (Additive formulas should always be used for international tournaments; multiplicative are OK for domestic ones.) Continue.

(Image)

3. Set the Max skill (this matters more when using NSFS over SQIS) which is usually the max rank + the max RP bonus. For now let's use 50 for ease. Enter Participants with their rank (I've used my puppets with their approx KPB rank) and style modifiers. (You can ignore Teams; this is relevant only for multiple individual athletes from the same nation e.g. Bill, Bob, and Ben, who all represent the same nation in Figure Skating. For team sports, all you need is Participants.) Continue.

(Image)

4. Under competition type, choose Individual Matches if you want single games e.g. friendlies, knockout rounds. Toggle Allow Draws accordingly. (Usually, friendlies allow draws, knockouts don't.) But instead, let's choose Round Robin. Choose the number of legs: 1 means they play each other once e.g. the group stage of most international football tournaments, 2 means they play each twice, etc. I've chosen 2 legs, so double round robin. Usually for group stages, Allow Draws is kept checked, but toggle accordingly. You don't need to check Results Grid. Choose your tiebreakers: you can change around the order, delete those you don't want, or add new ones. I've decided to remove Goals For, and move the Head to Head tiebreakers higher. If needed, you can also change the points, although 3/1/0 is the standard for soccer. Continue.

(Image)

5. Set up your groups. You add a group using the big Plus button. You add an individual team to that group using the next button to its right.

(Image)

Alternatively, you can add all participants using the middle button. The next button removes whichever participant you've selected (or, if you select a group, deletes that entire group); be careful as there's no Undo! Finally, you can randomize the order of teams in a group using the Dice button. Make sure the teams are put in the right order. For now I've put all teams in a single group and used the Dice button. Continue.

(Image)

6. You now can start scorinating matches. By default Matchday 1 is selected, so click the Dice button and you'll get your results and a table.

(Image)

You can scroll down through the Matchdays. If you want to scorinate all days in one go, you can go all the way down to Matchday 6 and scorinate that (e.g. if you're scorinating your domestic season, you might do it all in one day). If you do this, and then you rescorinate Matchday 1, you will lose the results for Matchdays 2 - 6 and need to rescorinate them. You can check results from past Matchdays from this menu.

(Image)

7. Good results! I'd like to keep them. So let's give this competition a name by clicking on the pale bar on the left hand side.

(Image)

I've called this the Awesome Cup.

8. To save the file, use the Save As button (Floppy disk + Pencil). Give it a name and it will be saved as a .xml file in the default folder such as Documents. To save the file again after you've modified it, use the Save button (Floppy disk).

9. To access a file you've previously used, use the Open button (Folder + sheet of paper). This should bring up the folder where you save your file, and you can begin editing it again.

[Andso Mangis]

xscorinate doesn't work for me are their other ways to scorinate game

Hi! You're going to have to go into more detail as to how it doesn't work for you. Could you do that for us so we can help diagnose your issue? Alternatively, try following the instructions in this post, which I've quoted below. You can click on the green "wrote" text next to Graintfjall's name to access the original post.

1. When you open it you should have a page like this:

(Image)

2. Click on New Event on the left and choose your sport. For now I'll use Association Football, SQIS Formula. Choose the options you want such as home advantage on or off, and whether to use additive or multiplicative modifiers. (Additive formulas should always be used for international tournaments; multiplicative are OK for domestic ones.) Continue.

(Image)

3. Set the Max skill (this matters more when using NSFS over SQIS) which is usually the max rank + the max RP bonus. For now let's use 50 for ease. Enter Participants with their rank (I've used my puppets with their approx KPB rank) and style modifiers. (You can ignore Teams; this is relevant only for multiple individual athletes from the same nation e.g. Bill, Bob, and Ben, who all represent the same nation in Figure Skating. For team sports, all you need is Participants.) Continue.

(Image)

4. Under competition type, choose Individual Matches if you want single games e.g. friendlies, knockout rounds. Toggle Allow Draws accordingly. (Usually, friendlies allow draws, knockouts don't.) But instead, let's choose Round Robin. Choose the number of legs: 1 means they play each other once e.g. the group stage of most international football tournaments, 2 means they play each twice, etc. I've chosen 2 legs, so double round robin. Usually for group stages, Allow Draws is kept checked, but toggle accordingly. You don't need to check Results Grid. Choose your tiebreakers: you can change around the order, delete those you don't want, or add new ones. I've decided to remove Goals For, and move the Head to Head tiebreakers higher. If needed, you can also change the points, although 3/1/0 is the standard for soccer. Continue.

(Image)

5. Set up your groups. You add a group using the big Plus button. You add an individual team to that group using the next button to its right.

(Image)

Alternatively, you can add all participants using the middle button. The next button removes whichever participant you've selected (or, if you select a group, deletes that entire group); be careful as there's no Undo! Finally, you can randomize the order of teams in a group using the Dice button. Make sure the teams are put in the right order. For now I've put all teams in a single group and used the Dice button. Continue.

(Image)

6. You now can start scorinating matches. By default Matchday 1 is selected, so click the Dice button and you'll get your results and a table.

(Image)

You can scroll down through the Matchdays. If you want to scorinate all days in one go, you can go all the way down to Matchday 6 and scorinate that (e.g. if you're scorinating your domestic season, you might do it all in one day). If you do this, and then you rescorinate Matchday 1, you will lose the results for Matchdays 2 - 6 and need to rescorinate them. You can check results from past Matchdays from this menu.

(Image)

7. Good results! I'd like to keep them. So let's give this competition a name by clicking on the pale bar on the left hand side.

(Image)

I've called this the Awesome Cup.

8. To save the file, use the Save As button (Floppy disk + Pencil). Give it a name and it will be saved as a .xml file in the default folder such as Documents. To save the file again after you've modified it, use the Save button (Floppy disk).

9. To access a file you've previously used, use the Open button (Folder + sheet of paper). This should bring up the folder where you save your file, and you can begin editing it again.

I download it however I cant acces the files for some reason is there a app I have to download

[Tinhampton]

xkoranate works on all versions of Windows. It does not work on newer Macs.

[Andso Mangis]

xkoranate works on all versions of Windows. It does not work on newer Macs.

So I have a newer MAC so what should I do

[New Eastern Canada]

Hey everyone!

I recently installed NSFS for my Regional League, and was making teams when it said "Team Style". What does that mean?

[Ko-oren]

Good question, and something we totally didn't forget to put in the Guide until literally 72 seconds ago.

[Andso Mangis]

Are their any scorinating methods you can use without downloading it

[Sel Appa]

Did we figure out a linux solution yet?

Im using a windows for the turtle cup, but thats less than ideal

[Andso Mangis]

New scorinator dropped for lacrosse https://docs.google.com/spreadsheets/d/ ... id=0#gid=0

[The Kytler Peninsulae]

As I have previously mentioned, I have been working on a prototype new scorination formula (originally designed for association football, but the mechanics could be adapted to at least some other sports - basketball will be the easiest, while white-ball cricket is another if more complex possibility).

LISA (Logic Inversion Scorination Algorithm) is that prototype, and it is now in a release-candidate status. While I did work on an entire chonky Google Doc to explain everything about it, much of it is a mess that's mostly serving me as a reference guide rather than being fit for public consumption. So here's the (slightly) distilled explanation of LISA.

TLDR: the process

• Each team's rank is converted into Elo Above Replacement (EAR) - i.e. their advantage in Elo over a 0-rank team (for whom EAR is 0) - with a formula that has adjustable constants to determine the extent to which rank affects EAR and by implication win probabilities.
• The difference in EAR (and hence Elo) between the two teams - after accounting for home advantage if relevant - determines the share of win probability for each team using the standard Elo formula, and the winner is selected based on this (with an additional calculation to determine the probability of a draw).
• If the result is not a draw, the next step is to determine the winning margin, based on the win probability of the actual winning team; if it is a draw, the margin is 0, and there is a separate calculation for extra time if needed.
• Finally, the number of goals for the losing team is determined, based on the style modifiers of the two teams; the winning team score is the losing team score plus the margin (in the event of a draw, the margin is 0 and hence the "losing team score" and "winning team score" are the same).
• A modified version of the same process is applied to extra time if needed; penalty shootouts are rankless coinflips

The nerdy details, feat. Oops! All Formulae

The EAR formula
The foundation of LISA is the conversion of rank to implied Elo - or more specifically to Elo relative to a 0-rank team, Elo Above Replacement (EAR) - using this formula:

EAR = s(ln(r+1)^x)

In this formula, r is the rank as inputted (i.e. incorporating RP bonus/no-roster penalty). As the natural logarithm of 1 is 0 and the natural logarithm of >1 is >0, EAR=0 for r=0 and EAR>0 for r>0, thus meeting the definition of EAR as the (implied Elo) value above a zero-rank team for any non-zero rank. Note that the formula can work for all positive levels of r and hence there is no maxrank; instead, there is a reference rank, provisionally 10.93 not only for the in-joke but because the geometric mean of the top 125 teams in KPB+ typically approximates to this, but can be set to an alternative level y for rank distributions that differ from KPB+ - in which case all ranks are multiplied by y/10.93.

x is an adjustable constant referred to as the power scalar, that determines the extent to which EAR continues to rise with increasing rank. s is an adjustable constant referred to as the multiplier; for ease of use, it is set indirectly by setting a given level of EAR for the reference rank (REAR, for Reference EAR); specifically,

s = REAR/ln(11.93)^x

Consequences of this approach:

• The strength of a team with a given rank is not affected by the strength of their opponent (unlike SQIS, which is based on the ratio of the team ranks and hence means that, for example, all non-zero ranks are equal against a zero-rank team), nor by the presence or absence of a dominant #1 team (unlike NSFS, which uses maxrank).
• LISA does not have a hardcoded level of upset propensity, because the multiplier and power scalar can be changed (and the resultant probabilities can be readily illustrated in a spreadsheet, so it will be possible for the impact of these changes to be easily quantified and visible not only to prospective hosts but to those voting on their bids).

Converting EAR to results
The established Elo formula is used to determine the xW (expected win percentage) for each team:
xW = 1/(10^(-g/400)+1)
where g is the team's EAR minus their opponent's EAR (the EAR gap!); if home advantage is applicable, g is increased by 100 for the home team (i.e. they hold 100 Elo points of home advantage). (This could be defined on a team-by-team level, but this would have a metagaming effect if it was a nation-defined variable like style modifier. However, this is not an obstacle for domestic leagues, two of which (NSL, Q-League) used a beta version of LISA incorporating team-specific home advantage in the last IFCF cycle.)

This formula does not determine the probability of a draw, so a further step is required to do this. Evaluation of a sample of both xkoranate results and RL match probability estimates from Opta Analyst (which uses a rescaled Elo rating for its Power Rankings) led to the following process:

• Let u be defined as the lower xW of the two teams, i.e. the underdog xW.
• Let d be the share of u from draws, defined as d=0.5-0.5u.
• The probability of a draw is then defined as 2du (as a draw counts as 0.5 wins), the probability of an underdog win as u-du (i.e. that part of u not covered by the draw), and the probability of a favourite win is 1 minus the sum of these two values (1-u-du).

As d is a function of u, these can also be expressed solely in terms of u:

• Draw probability: 2du = 2u(0.5-0.5u) = u-u²;
• Underdog win probability: u-du = u-0.5(u-u²) = 0.5u²+0.5u; and
• Favourite win probability: 1-u-du = 0.5u²-1.5u+1

The sum of these three values is always 1.

Determining winning margin
With the winner (or lack thereof) determined, the next step in the LISA process is determining the winning margin.

Naturally, if the result is a draw, that winning margin is (by definition) zero; if the result is not a draw, a further calculation step is required. In NSFS and SQIS, winning margin is a function of the broader calculation of score - each team has a certain number of “attacks” with a rank- and style-dependent probability of being converted to one goal. LISA, however, does the opposite - the scoreline follows the establishment of the winner.

Intuitively, favourites can be expected to win by more goals than underdogs on the occasions they do win, and this is inherent to the attack-based processes of NSFS and SQIS. LISA has no such equivalent, and hence the margin calculation needs modelling - and tables of RL result distribution do not tell us anything in themselves about if victories have come from underdogs or favourites. Betting odds on a sample of various high-level RL matches were used as a proxy for evaluating this, and the patterns from these demonstrated that the winning margin could reasonably be modelled by a zero-truncated Poisson distribution.

The probability mass function (the formula defining the probability of the random variable exactly equalling a given value) for this distribution is:
P(X=k|k>0) = λ^k/((e^λ-1)k!)
where k is a given value - in this instance, the winning margin - and λ is a constant, higher values of which are associated with higher expected values of k.

For this purpose, λ = 1.093 exp (0.002g) (recall that g is the EAR gap). This calculation is made specifically for the winning team, so g will be negative in the event of an underdog win.

Illustrative examples:

• In an entirely even match, g = 0, λ = 1.093, and the mean winning margin is 1.644 goals with a 55.1% probability of a one-goal margin.
• In a much less even match won by the favourite where g = 350, λ = 2.201, and the mean winning margin is 2.475 goals with a 27.4% probability of a one-goal margin. (For reference, this level of gap is approximately equal to a match between a ranked team and an unranked team in SQIS without home advantage, and equates to an 83% chance of the favourite winning and a 6.6% chance of the underdog winning.)
• If the underdog won that same match, g = -350, λ = 0.753, and the mean winning margin is 1.296 goals with a 75.3% probability of a one-goal margin.

Determining the actual score
The above steps establish who (if anyone) wins a match, and the margin of victory if there is any. However, this does not cover an exact score, whereas existing attack-based scorination methods such as SQIS and NSFS inherently do.

Notice what hasn't been covered yet. In LISA, style modifiers do not play any part in determining who wins, nor by how many - they literally are not incorporated into those parts of the process. They are only applied at this point, determining the total number of goals.

This is determined by calculating the number of goals scored by the losing team, to which the winning margin is added to determine the winning team's score. (In a draw, this margin is 0 and the "losing team score" and "winning team score" are the same.) The losing team score is modelled by a regular Poisson distribution:
P(X=k) = λ^ke^-k/k!
In this distribution, λ is both the mean and the variance, and for this purpose, λ = 0.75+0.064s, where s is the sum of both team's style modifiers. Based on the assumption of a style modifier range of -5 post-football to +5 Funball, this means λ = 0.11 for Audioslavia v Ko-oren-5 v -5 and 1.39 for +5 v +5. Again, none of this affects the probability of which team wins, or even by how many - it simply determines the number of goals.

At λ = 0.75, 47.237% of matches have a losing team score of 0 (i.e. at least one team kept a clean sheet, or both if it was a draw); at λ = 0.11 this is 89.583%, and at λ = 1.39 it is 24.908%.

The expected number of goals in a given match is equal to the mean projected winning margin plus 2λ from this formula (as λ is the mean number of goals for the losing team and the winning team will score that many plus the margin). In a match between two equal teams with neutral style modifiers, there will be an average of approximately 2.7 goals per game, very similar to the current levels in SQIS and NSFS.

What if there's extra time?
The formula as discussed only covers 90-minute matches. In the event of extra time:

• Let t be the probability of a decisive result in extra time (i.e. the match does not go to penalties) and g be the absolute EAR gap (i.e. always positive), t = 0.3109g^0.1093+(1/10930) iff g > 10; otherwise, t = 0.4
• Then, iff there is a decisive result, the probability w that the winner is the favourite (i.e. the team with the higher EAR) = w = 1.093t^0.837 iff t > 0.4; if t = 0.4, w = 0.5
• The calculations for losing team score and (if needed) the winning margin are identical to those made for normal time, but with λ divided by 3 in each instance (as the duration of extra time is one-third of normal time)

These formulae are derived from analysis of betting market odds ahead of a number of finals over the 24-26 May 2025 weekend with various levels of projected imbalance between the two teams.

Penalty shootouts use the Qusma shootout scorination algorithm with no regard for rank. This algorithm assigns P (converted penalty) at 3/4 for the first three takers for each team, 2/3 for the next seven, and 1/2 for the eleventh.

Let's go back to those adjustable constants...
Much community discussion has arisen over how rank-deterministic the World Cup and its associated tournaments "should" be - a question that is subjective with no inherently correct answer, but one complicated by the inherent added randomness of a 48-team World Cup (because 16 fewer teams are eliminated in the large-sample-size qualifiers and 16 more teams advance to a single-elimination knockout bracket) and a format debut in World Cup 97 which was notably random even after accounting for that. LISA was designed (in part after its creator was on the positive side of that randomness at WC97 and felt awkward about it...) explicitly to be able to provide multiple answers to that question, rather than just the one that is coded into NSFS (SQIS has an adjustable constant (constantB) that modifies the effect of rank difference, but it is rarely used).

In summary, the two adjustable constants work as follows:

• The multiplier, for which REAR is used as the user-adjustable proxy, determines the general steepness of the distribution. A higher REAR (and hence multiplier) has an equal multiplicative effect on the entire distribution.
• The power scalar determines the shape of the distribution; a lower scalar flattens more quickly above the reference rank but increases faster below it.

This is illustrated in the image below, with REAR set to 250 with a power scalar of 2.5, to 300 with a power scalar of 1.8, and to both 250 and 300 with a power scalar of 1.984:

Notice that the green line (2.5 power, 250 EAR) increases much more slowly than the others at first, then crosses over the yellow (1.984 power, 250 EAR) line at 100% of reference rank (because they share the same EAR) and then increases much faster thereafter, crossing the blue line (1.8 power, 300 EAR) at around 220% of reference rank and the red line (1.984 power, 300 EAR) at around 300%.

The provisional "default" setting for LISA is 1.984 power/300 REAR. Characteristics of this setting:

• Matches between highly-ranked and lowly-ranked teams are much less likely to produce upsets than (unadjusted) SQIS. With constantB at the default 0.07, any zero-rank team has a 5.51% chance of a draw and a 1.66% chance of a win away to any non-zero-rank team, even the very highest; these probabilities both more than double without home advantage. These probabilities equate to an Elo gap of around 525 points where the ranked team is at home, 390 points on neutral ground - though samples from actual xkoranate testing have produced lower estimates of about 350 for this - and 345 points where the ranked team is away. Top-ranked teams will typically have around 250-300% of reference rank prior to any RP bonus, which equates to an EAR of 550-600, or an Elo gap of 650-700 points when playing an unranked team at home and 450-500 away; these numbers will increase further, albeit slowly, with additional bonus.
• New teams can rise relatively quickly given sufficient bonus; the step up from 10% to 50% of reference rank is worth almost 150 EAR.
• There is, however, an initial slow increase at the very lowest rank. Whilst inadvertent, this does serve to keep teams with very low non-zero rank very close to actual zero ranks, unlike SQIS; as this will primarily affect non-RPing nations, I see this as a feature rather than a bug, but it can affect non-BoF debutants early in qualifying in cumulative-bonus tournaments. (The use of a weighted-average bonus avoids this issue.)

But this is very much not set in stone - nor do I expect it to stay set in stone even across tournaments it is used in! And because of how the win probabilities are literally a function of EAR, you can easily see them change with the constants for any given match. Here is a link to a sheet you can make a local copy of for just that. (It also serves as a single-match result generator, but I haven't added ET to that yet.)

Anyway, this is what I have. Feel free to tweak it to destruction.

[Northwest Kalactin]

TLDR: the process

• Each team's rank is converted into Elo Above Replacement (EAR) - i.e. their advantage in Elo over a 0-rank team (for whom EAR is 0) - with a formula that has adjustable constants to determine the extent to which rank affects EAR and by implication win probabilities.
• The difference in EAR (and hence Elo) between the two teams - after accounting for home advantage if relevant - determines the share of win probability for each team using the standard Elo formula, and the winner is selected based on this (with an additional calculation to determine the probability of a draw).
• If the result is not a draw, the next step is to determine the winning margin, based on the win probability of the actual winning team; if it is a draw, the margin is 0, and there is a separate calculation for extra time if needed.
• Finally, the number of goals for the losing team is determined, based on the style modifiers of the two teams; the winning team score is the losing team score plus the margin (in the event of a draw, the margin is 0 and hence the "losing team score" and "winning team score" are the same).
• A modified version of the same process is applied to extra time if needed; penalty shootouts are rankless coinflips

Unsure as to how much my opinion actually matters to this community on the issue of scorinators anymore, but add my name to what I'm sure is already a long list of endorsements for this formula.

I think the idea of using the EAR as a mathematical benchmark for scorination is brilliant, and in my opinion, is superior to both the NSFS and SQIS formulas.

NSFS compares the ranks of each team to a max rank, which is set by the host on xkoranate (typically the highest rank + RP bonus). This works well when you have a team that is highly ranked, and therefore close numerically to the max rank; however, it induces randomness when you have two lower or mid-ranked teams, as differences in rank become negligible when compared to a team with a top-5 KPB earning max RP bonus each match day.

SQIS, on the other hand, compares the ratios of the two teams. This allows each team to be compared directly to their opponent, reducing arbitrary randomness; however, it theoretically would play a 50-25 KPB difference the same as a 2-1 KPB difference. While I personally do not believe this is a pressing issue (and the 'randomness' of SQIS has been debunked numerous times), it does have more randomness than NSFS for higher-ranked teams.

The LISA formula seems not to suffer from either of these issues. While, admittedly, I do not fully understand all the math (though KP seems entirely competent on this issue, and I have no doubts he rigorously tested this aspect of the scorinator well), the formula looks very strong, and I am excited to see what it potentially turns into in the future.

[The Kytler Peninsulae]

Unsure as to how much my opinion actually matters to this community on the issue of scorinators anymore, but add my name to what I'm sure is already a long list of endorsements for this formula.

There do seem to have been some positive comments on the NSS Discord. (Along with plenty of "IDK, I'm not a mathematician" which I understand, I had to look up half of this stuff myself and quickly found at university that my command of calculus quickly reached... a limit...)

I think the idea of using the EAR as a mathematical benchmark for scorination is brilliant, and in my opinion, is superior to both the NSFS and SQIS formulas.

NSFS compares the ranks of each team to a max rank, which is set by the host on xkoranate (typically the highest rank + RP bonus). This works well when you have a team that is highly ranked, and therefore close numerically to the max rank; however, it induces randomness when you have two lower or mid-ranked teams, as differences in rank become negligible when compared to a team with a top-5 KPB earning max RP bonus each match day.

SQIS, on the other hand, compares the ratios of the two teams. This allows each team to be compared directly to their opponent, reducing arbitrary randomness; however, it theoretically would play a 50-25 KPB difference the same as a 2-1 KPB difference. While I personally do not believe this is a pressing issue (and the 'randomness' of SQIS has been debunked numerous times), it does have more randomness than NSFS for higher-ranked teams.

The LISA formula seems not to suffer from either of these issues. While, admittedly, I do not fully understand all the math (though KP seems entirely competent on this issue, and I have no doubts he rigorously tested this aspect of the scorinator well), the formula looks very strong, and I am excited to see what it potentially turns into in the future.

I really appreciated this feedback!

From the testing I did with both NSFS and SQIS in the process of developing LISA, the thing with SQIS is that it's only "more random" in terms of top teams being a little more vulnerable to upsets in general and upsets against low-rank teams in particular, as (before accounting for any constant modifiers that can be done but involve delving into the xkoranate code) the hard cap on rank separation is lower in SQIS. (NSFS gives maxrank teams ~0.98xW against zeroes, SQIS only ~0.88xW.) However, in most scenarios, SQIS has at least comparable rank separation from NSFS, and sometimes more - and those times where it has more are the kind that often come up in the Proper! I've just done a test of 1,000 simulated matches between a 90%-max and 60%-max team in each formula, and in SQIS the higher-ranked team went 533-250-217 (0.658xW) compared to 528-229-243 (0.6425xW) in NSFS - a small gap, but one in the opposite direction to the one frequently perceived to exist. (And the Paradigm Comparison calculator actually estimates more separation, in that same direction - NSFS 0.6092xW, SQIS 0.6595xW.)

So, in other words, the "SQIS in qualifiers because home advantage works in it, NSFS in the Proper because it's less random" turns out to specifically be untrue in the kind of matches most seen in the Proper, and the SQIS-is-too-random trope specifically applies to matches not seen in the Proper!

Now, whether LISA will be the right level of random or too little or too much... well, that depends on the parameters, and that's the bit I'm most deeply unsure of, in particular the tradeoff between "faster progression for low-ranked teams" and "have RP bonus mean something for high-ranked teams" that I've been agonising over. It's entirely possible that, despite the provisional 10.93 reference rank designed into the formula specifically for WCC use, the different shape of distribution created with a change of this value may be seen as preferable. But that's the thing with LISA - it's inherently flexible as well as internally consistent in how rank converts to skill, and not just in terms of "one number change makes it more/less random" but in terms of the shape of the distribution. I can envision a scenario in which LISA becomes established as the standard WCC formula but with half the WCDT being filled with debates over where the adjustable constants should be!

[Quebec and Shingoryeo]

As someone who's trialled LISA's domestic BETA file for the last IFCF season, and has also run several third-party scorinations on his behalf, I can safely say that KP has done an excellent job.

A particular commendation goes to its ability to adjust home advantage where necessary. IRL home advantage varies dramatically between teams (Man Utd's horrendous home form last season proves this) and leagues (see: Nigerian Premier League). KP has indeed taken right steps to make sure the home advantage does factor well into the real-life and NSS-specific situations. Needless to say that it has helped providing a more realistic set of results which work well in a country like Q&S where hostile climate, long travel and the RQFA's decades-long ban on away support exist. I hope people will be able to enjoy the domestic football version of LISA scorinator in due time.

For similar reasons, I would also be more than happy to try out a prospective basketball version of this scorinator in coming months. I will be looking forward to it.

[Epic bannana]

are there any scorinators that you can do online like league simulator?

[Andso Mangis]

are there any scorinators that you can do online like league simulator?

There’s some spreadsheets but those you have to go through a link but even those aren’t online so no

[Ko-oren]

I'm trying to get a feel for how LISA works. Having read the documentation earlier in this thread, what would advisable settings be for the power scalar, reference rank, REAR - and what ranges they should be ideally be in?

[The Kytler Peninsulae]

Good question! Especially because it's still quite up in the air to be honest, because the power scalar changes the shape of the distribution and that has implications for the impact of a given level of EAR / reference rank.

What I can say is that the should-have-been-linked-here-before RC2 version of LISA (which realistically is the final release at this point, pending further updates anyway) has a feature that shows the EAR for each team at a given setting. An EAR of 550-ish is broadly in lj e with NSFS maxrank, an EAR of 350-ish with a SQIS ranked v unranked match, and the latter is also about in line with top-to-bottom in most of the top RL men's domestic leagues.

There is also an xPts tab that shows the expected points for every team in a double round-robin league of up to 24 teams, which can also be used for balance checking. Based on Opta Analyst's modelling of them, the typical top RL men's domestic leagues have xPts for the top teams around the 2PPG mark and the bottom teams fall around 1PPG, for reference.

The reference rank of 10.93 was picked because the geometric mean of the top 125 in KPB+ was reliably close to it, but making it higher even within that context has some value, and of course in domestics you have a free hand here. Broadly speaking, with a power scalar around 2 (which is what I tentatively lean towards) and REAR at 250-300, reference rank of about 30% of where you'd put maxrank in NSFS would be a reasonable rule of thumb. Higher power scalars might require lowering REAR and/or raising reference rank to create the same balance. (I believe that a power scalar of e creates a roughly linear relationship between rank and EAR, but haven't properly checked this. Lower scalars mean that returns on incremental rank increases are higher further down the scale, which is an intended feature for international use to give new RPers a stronger "return on investment" if you like.)

But TLDR; for domestics at least, I'd change the parameters to fit the distribution you want, with a rough rule of thumb being about 300 EAR between the top and bottom of a division. There'll be more detailed advice on LISA for domestics in future documentation

[Jiangbei]

I would like to announce that I have managed to edit the source code, compile this scorinator on my laptop, open the app, and run xkoranate successfully (Apple M chip on MacOS Sequoia 15)

[Jiangbei]

Download link for macOS (both Apple Silicon and Intel) is here. This should be the exact same as version 0.3.3

The current main repository of xkoranate is at https://github.com/NS-Sports/Xkoranate-CE, the version I made is at https://github.com/bohaska/Xkoranate-CE

[Sel Appa]

I would like to announce that I have managed to edit the source code, compile this scorinator on my laptop, open the app, and run xkoranate successfully (Apple M chip on MacOS Sequoia 15)

([url=https://i.ibb.co/k2z61cRP/xkoranator.png]Image)[/url]

Any idea if that would work in Linux??

[Borschtslatvia]

Download link for macOS (both Apple Silicon and Intel) is here. This should be the exact same as version 0.3.3

The current main repository of xkoranate is at https://github.com/NS-Sports/Xkoranate-CE, the version I made is at https://github.com/bohaska/Xkoranate-CE

Yay!

Now if I can only find a tutorial on how to use it, I can host a Pig Racing event someday.

[Cobrio]

Download link for macOS (both Apple Silicon and Intel) is here. This should be the exact same as version 0.3.3

The current main repository of xkoranate is at https://github.com/NS-Sports/Xkoranate-CE, the version I made is at https://github.com/bohaska/Xkoranate-CE

The pre-compiled macOS version is now also available in the releases section of the github repo for xkoranate-CE, just in case anything happens to this link

I would like to announce that I have managed to edit the source code, compile this scorinator on my laptop, open the app, and run xkoranate successfully (Apple M chip on MacOS Sequoia 15)

([url=https://i.ibb.co/k2z61cRP/xkoranator.png]Image)[/url]

Any idea if that would work in Linux??

Possibly, if you compile the source code for either Jiangbei's fork or mine, since both have had updates to remove deprecated functions etc and bring them up to date with newer versions of C++ and Qt

Next on the list of tasks is to get a working version for the three main OSes in use, though. So if you're not in a position to try it, someone (probably me or Legalese I imagine) will at some point.

Download link for macOS (both Apple Silicon and Intel) is here. This should be the exact same as version 0.3.3

The current main repository of xkoranate is at https://github.com/NS-Sports/Xkoranate-CE, the version I made is at https://github.com/bohaska/Xkoranate-CE

Yay!

Now if I can only find a tutorial on how to use it, I can host a Pig Racing event someday.

There's xkoranate for Dummies
There is a short video guide here as well.

[Anjan Kloss]

I would like to announce that I have managed to edit the source code, compile this scorinator on my laptop, open the app, and run xkoranate successfully (Apple M chip on MacOS Sequoia 15)

([url=https://i.ibb.co/k2z61cRP/xkoranator.png]Image)[/url]

As a Mac user, I really appreciate you making this! The only slight problem I have with it, is that I can't seem to find basketball on it lol? Is it under a different name or something? Or is it a me issue?

[Free Republics]

I would like to announce that I have managed to edit the source code, compile this scorinator on my laptop, open the app, and run xkoranate successfully (Apple M chip on MacOS Sequoia 15)

([url=https://i.ibb.co/k2z61cRP/xkoranator.png]Image)[/url]

As a Mac user, I really appreciate you making this! The only slight problem I have with it, is that I can't seem to find basketball on it lol? Is it under a different name or something? Or is it a me issue?

Basketball should be right after baseball. You have to hit the green plus sign first to create an "event" so that you can choose a sport. If it isn't showing up, something is likely wrong with your sports folder which would likely imply a bug with the Mac xkoranate build (I haven't tested it myself yet as I mainly use a Windows laptop these days not my old Intel MacBook Pro from 2018/19). Then put the teams in as "participants" not "teams". The "team" field is used for NS Olympic style trigrams. Meaning if you enter Michael Jordan and Scotty Pippen as "participants" with Chicago Bulls as the "team" for both, it will treat Michael Jordan and Scotty Pippen as if they were basketball teams from a nation called Chicago Bulls.