|
||||
|
10-23-2015, 08:56 PM | #41 | |||||
Hall Of Famer
Join Date: Jul 2004
Location: The big smoke
Posts: 15,628
|
Quote:
Quote:
Quote:
Quote:
Quote:
I'm not sure what type of leagues you run but I've had significant success manipulating LTM even within a season with fine adjustments to stats like ground ball outs to get a very realistic statistical output, even in the minor leagues. Maybe we should exchange league files to see why our impressions are so different.
__________________
Cheers RichW If you’re looking for a good cause to donate money to please consider a Donation to Parkinson’s Canada. It may help me have a better future and if not me, someone else. Thanks. “Conservatism consists of exactly one proposition …There must be in-groups whom the law protects but does not bind, alongside out-groups whom the law binds but does not protect.” Frank Wilhoit |
|||||
10-23-2015, 09:39 PM | #42 |
Hall Of Famer
Join Date: Apr 2003
Location: Minnesota
Posts: 2,722
|
Pitcher ERA is a little high right now. Not many sub 3.00 ERAs in seasons.
|
10-24-2015, 01:54 AM | #43 | |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
Quote:
i don't want an arbitrary change in an LTM to dictate anything about statistical output and fluctuation - unless it is tied to a tangible change in my league, or concept in my head. if you want to manually pick when various types of eras begin or end, that's a good reason for you to fiddle with them, but the function is arbitrary in nature. about shorter sims - the average you get from them is loaded with percent error. they are not precise. this isn't about opinion, this is about sample size and level of confidence in calculated averages. it is necessary if you want an environment that doesn't require the use of autocalc. autocalc will cause fluctuation but for reasons other than player talent, coaching, development etc... changing LTM has nothing to do with the talent in your league. you or autocalc is deciding that 90 power should equate to fewer or more homeruns in a particular year based on a small sample size. changing LTM is changing the scale of various player ratings (simplified). increase HR LTM means more HR per power rating (simplified). if you don't put in the time and effort to flesh out the averages and LTMs that create an equal environment regardless of era, but you want something like that, then autocalc is for you. with a little effort you can center the natural fluctuation around a point of your choosing. if you want to say the mound has been raised again and institute a pitcher's era - cool beans. you have a cause to use it. something tangible... about adding .001: okay, this is all hypothetical: let's say K's for a 28t/156g league over 100 years averages to ~32,750. adding .001 to that LTM will result in 33,600 average over 100 years in second sim, or an 850 increase. if it were at 1.000 you would expect .001 of 33,000 (pretend it's the league total in settings), which is +33 SO's not +850 SO. it's so drastically different there is no explanation. you just have to roll with it. this does happen at certain points and not only with SO's. you can't recognize this happening in 1 or even 20 years of data. 50 will give a good idea, but it still is off by +/-X% that's significant enough to cause problems in guesstimating the real resulting average. other varibles influence at what LTM value this occurs. it isn't a static point. this actual average is not something you can see in 1 year of data. in fact, you can't see anything from one year with any confidence. to show you what i mean, i attached a min/max/average/range (in that order and in order of LTMs, sorry no headers in that area) of various stats from a 100year sim with static LTMs - this was 28teams and 156games. the min/max will show that any year can be drastically different. so, even if you have a high level of talent and there is an off year due to bad luck, autocalc will inflate the related LTM causing an artificial increase in those stats not based on something tangible. when this situation arises, players will get a benefit in those years for poor reasons. sure it can happen the other way too, and my argument still stands. without somethign tangible(in the game/in a person's head) tied to the ltm change, it doesn't make much sense to do without cause. now, if in anyone uses this, justified or not, that's fine. it won't ruin anything. regardless, it is still an arbitrary re-scaling of ratings and whatever else it influences to cause change. even the title is scary relative to a sim.. "for statistical accuracy." it will cover up leagues with fundamental flaws that result in whacky statistical output without using it. this type of thing is not natural for a simulation. it's taking the results and saying "i don't like this, so i'll arbitrarily inflate this over here and maybe change my mind a few years later." (that's the game speaking to itself) Last edited by NoOne; 10-24-2015 at 02:46 AM. |
|
10-24-2015, 02:30 AM | #44 | |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
Quote:
due to formatting in the spreadsheet, please have headers inserted and use comma as a delimiter (settings control these things in game). depending on how many years of data you have, it may or may not fit in the PMs here. i can break it into zip chunks but that's annoying, lol. you should try to play around with it yourself, too. i tried to write up some verbose instructions. once you get it down, it literally takes seconds to use. the moment you open it, it asks to update links to the csv files, and then it's completely updated, and results are calculated. that's it. it is ready to read. is this a fictional league? the first 15-30 years of data will be bunk. this is one of the rare times you will hear me say... this time period is a good place to use autocalculate for statistical accuracy, LoL, !@!$%#. but! it's not for good reasons, it's to cover up a shortcoming of the game. you will have to wait for created players to filter though your league. the minor leaguers you get to start a fictional league are substandard unless you got really lucky. they are not created in the same way as amateurs for your draft and international FA kids. i sim out 35 years. overkill, but better safe than sorry. i figure a 20 year career and 6 years of minors (26 years) will cover 99.9% of players. 35 years assures it beyond any doubt. real-world leagues you have to sim out until all the real players are replaced... same idea but slightly different reasons. the stats will be all whacky during this transition period, without using autocalc. there's no feasible way to calculate averages when the talent distribution is undergoing a significant and permanent shift. Last edited by NoOne; 10-24-2015 at 02:47 AM. |
|
10-24-2015, 07:27 AM | #45 | |
Hall Of Famer
Join Date: Apr 2003
Location: Minnesota
Posts: 2,722
|
Quote:
I know I probably shouldn't have, but I was adjusting League Total Modifiers as I simmed. I would only change them every 5-10 years. So numbers might be a little off. league_history_batting_stats.csv league_history_fielding_stats.csv league_history_pitching_stats.csv |
|
10-24-2015, 09:20 AM | #46 | |
Hall Of Famer
Join Date: Jul 2004
Location: The big smoke
Posts: 15,628
|
Quote:
__________________
Cheers RichW If you’re looking for a good cause to donate money to please consider a Donation to Parkinson’s Canada. It may help me have a better future and if not me, someone else. Thanks. “Conservatism consists of exactly one proposition …There must be in-groups whom the law protects but does not bind, alongside out-groups whom the law binds but does not protect.” Frank Wilhoit |
|
10-24-2015, 03:10 PM | #47 | |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
Quote:
the bottom line is that on its own, this function literally reacts to randmness of results from a small sample size and alters the scale of various ratings, or whatever else, in the game for the following year. i think i read it only uses 3yrs of quickly calculated data to come up with new LTM? the percent error associated with the sample size for all the various stats is atrocious. changing LTMs without good reasons is like conducting a poll and when you don't like the results you decide to change something to coax a more preferred answer. (assumes the original method was logical and rational) once i get the LTMs to center around a baseline that i feel is accurate to modern MLB, i leave them alone. very simple, indeed - not even a thought is applied to them after that. |
|
10-24-2015, 03:26 PM | #48 | |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
Quote:
Edit: i'm working on it as i type, so i'll post a reply soon. okay, some issues i see ... 1) one this spreadsheet is setup for a static number of teams... it would take some persnalized edits to account for the change in # of teams each time it occurs. # of games and number of teams will need to be accounted for. 2) if you changed LTMs throughout, there is little information to glean from this spreadsheet. it's purpose is to make a change after enough years of data are collected. *there is nothing wrong with adjusting LTM, don't get me wrong. if you haven't centered your stats around a baseline of your choosing, there's nothing else you can do but alter them on occasion or use autocalc. this is the path of least resistance that you may want to take. it's easier. i'll send back 2 spreadsheets, nonetheless. just don't put too much weight into the suggestions, because of the 2 reasons i listed above. the data is tainted, so take it as a grain of salt. Last edited by NoOne; 10-24-2015 at 04:35 PM. |
|
10-24-2015, 04:08 PM | #49 |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
ugh, they have to be zip files, LoL... why?!?
i removed any yellow background on cells you do not need to use. All yellow background cells on the Input worksheet must be filled in. (actually in your case the current LTMs arent necessary to fill in due to what i explained in the previous post, but you can put them in if you wish). Do not touch any other cell's contents. the only other thing you will ever adjust outside that input worksheet is the table size on the "Work" worksheet to accomodate the correct amount of years. the "adjusted" section of the Work worksheet is scaling the LTMs that also have totals listed for them in the league settings. this is so that you can compare apples to apples. e.g. 4500hr with 150,000ab is not the same as 4500hr with 148,000ab. this adjusts them to an even scale. the raw data is at the far end of the table. if you didn't make majore adjustments to your LTMS (guessing <.010 changes???), you might get some half-way useful info from the suggested new mods. however, apply what you changed... e.g. if you incresed HR, take a little off the new suggestion for HR LTM. okay, both are labeled with the years. ***1900-1949*** and 1950-1953 everything is set, if it asks to update links, say no. only update them if there is new data to use. (use the 1900-1949*** file for that) the 1950-53 i did sloppy and fast, don't allow it to update the links when you open it up in openoffice calc. (if you use excel, hopefully it works, lol, otherwise download apache openoffice). it will be tougher to re-use this file, because of how i adjusted it. keep the 1900 one for future use. put it and the 3 csv files in the same directory and each time you export new stats, copy them over those 3 files. once you open the spreadsheet it should ask if you want to update links... do so and you only have to adjust the table size on the worksheet to account for the correct amount of years. since you sent 50 years, it's set for 50 years at the moment. add or remove entire rows within the table to adjust its size (add/remove 2 rows per year). never remove the first or last row of the table. highlight 1 row of data and 1 empty line and "Fill" down/up. use help for info on Fill function in your spreadsheet program if you are not familiar. it's that easy to adjust the size of the table... starting at a different year than year 1 of data is a bit trickier. highlight and drag the new 'starting' year to the first row, then fill down as described above. do not copy/paste, it will not work right. to start at an earlier date, click and drag the first year down X number of years you want to start earlier and fill up from there. all formulas will adjust automatically if you do this process right. make a back up and play around with it until you are comfortable. there are instructions on how to use it as well as a general strategy to adjusting LTMs. they more LTMs you adjust at one time, the less likely you are to get the results you want. the order in which i suggest setting them is not perfect. there is some useful info (min/max/avg) from the Work worksheet that i should have on the input page (really should be "Input/Output"). if you want somem visual feedback on varous stats, make some line graphs. Last edited by NoOne; 10-24-2015 at 04:51 PM. |
10-24-2015, 05:21 PM | #50 |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
i attached an image of the results from a 100-year verification of the baseline i wanted. only 10 years of data is shown. this was my last run, so the suggested mods are meaningless. unless you do an infinite amount of years, there will always be a suggested change.
it took multiple long-term sims to hone down the LTMs i wanted. if you don't want to do this, just use autocalc. let it run overnight and make minor adjustments and continue until you get what you want. Last edited by NoOne; 10-24-2015 at 05:23 PM. |
10-24-2015, 09:09 PM | #51 |
Hall Of Famer
Join Date: Apr 2003
Location: Minnesota
Posts: 2,722
|
Thank you for this. I will take a look at it all, but might end up just adjusting as I proceed throughout the game.
|
10-24-2015, 11:03 PM | #52 | |
Hall Of Famer
Join Date: Jul 2004
Location: The big smoke
Posts: 15,628
|
Quote:
__________________
Cheers RichW If you’re looking for a good cause to donate money to please consider a Donation to Parkinson’s Canada. It may help me have a better future and if not me, someone else. Thanks. “Conservatism consists of exactly one proposition …There must be in-groups whom the law protects but does not bind, alongside out-groups whom the law binds but does not protect.” Frank Wilhoit |
|
10-26-2015, 01:31 PM | #53 | |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
Quote:
your numbers were reasonable. in case you are not happy with the changes you make, make a backup of the league, too. |
|
10-30-2015, 09:00 PM | #54 |
Hall Of Famer
Join Date: Apr 2003
Location: Minnesota
Posts: 2,722
|
If I'm just starting a new game, what should I do?
|
10-30-2015, 09:54 PM | #55 | |
Hall Of Famer
Join Date: Aug 2003
Posts: 9,502
|
Quote:
However, the result you're describing has a very clear and mathematical explanation. It is because "Totals" are not "Totals." "Totals are ratio-based modifiers that create stats based on the talent levels in the world you're playing with. You do not need to just roll with it. In order to know that a .001 tweak of a modifier to strikeouts will do, however, you need to know the base League Total for AB and K--which can well be different from one league to another. If you know these values, the values of all the other core League Totals, and the average ratings of all the hitters and pitchers who are going to the plate, you can very accurately predict the output of the league without running any sims. You're also correct that three seasons of sim isn't enough to guarantee perfection in doing what the feature says it's doing, but in the sloppy world of randomness, it generally serves the purpose of stabilizing statistical output so that all but the most mathematical purist are happy with the results. (In addition, I'm purely guessing that Markus and the gang do more than merely sim stats ... shrug). Many things can cause issues with stats output. Player talents can vary from era to era, for example. |
|
10-30-2015, 11:28 PM | #56 |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
i'd use autocalc the first 20-30 years or so. you can just leave it on and let it do it's own thing, or you can click it a few times and take an average of the results.
as time goes, this will be less important. the only reason i'd suggest it at the beginning of a league is because the minor leaguers are not of the same ilk as the players created for your draft. if you don't have a full minor league system, ignore this. either way, those players will ceases to exist once enough time passes. after that, maintain LTMs that make you happy. |
10-30-2015, 11:39 PM | #57 |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
i don't see that as an issue. i see that as the major reason why stat output should change over time, in addition to random luck. also, that's the reason i want a baseline statistical environment. i don't want those lesser players in down offensive eras putting up puff numbers.
Last edited by NoOne; 10-30-2015 at 11:40 PM. |
10-30-2015, 11:55 PM | #58 | |
Hall Of Famer
Join Date: Aug 2003
Posts: 9,502
|
Quote:
But there are many others who just want to see whatever they think of as "realistic" numbers. There are arguments that they may be "right" also ... especially if the player development engine gets a little wonky--as it most definitely been know to do on occasion. Hitting the system with autocalc may well be akin to kicking your television set to make it work, but sometimes a kick does the job well enough. One can make a very good argument that league totals (or their "modifiers") in real life have been steadily changing for a multitude of reasons both in control and outside the control of baseball leaders. I mean, it's quite likely that the differences you see in baseball output is not an effect of changing skill only. As I said, though, your basic view of the concept of modifiers is right. It's just that a lot of people can get as much enjoyment and "accuracy" out of using autocalc as you do adhering to strict environmental stabilism. That said, (and please read this in a non-targeted tone of voice) if you don't understand how changes in LTMs result in disproportionate changes in stats, you have more to dig into regarding how League Totals actually work. You really, really do not need to do mega-sims to get a good idea of what the output of the stats engine will come up with. Last edited by RonCo; 10-30-2015 at 11:59 PM. |
|
10-31-2015, 08:00 AM | #59 |
Hall Of Famer
Join Date: Apr 2003
Location: Minnesota
Posts: 2,722
|
I'm now working on testing with some LTM I got via a spreadsheet (forget which one). I'm going to sim a few years and see how numbers look.
|
10-31-2015, 06:08 PM | #60 | |
Hall Of Famer
Join Date: Apr 2015
Posts: 7,167
|
Quote:
the development engine is likely just a scapegoat for those that made their league in a way that causes the oddity that they are experiencing. it's more than likely human error and not the development engine itself. if you have the actual math mapped out from the code, sure, you don't need long-term sims. however, a person is unlikely to have this information. understanding that it happens is very helpful to the context of what i was speaking about. i was trying to help someone recognize it when it happens. nothing more, nothing less i've seen an average change drastically from 1, 5, 10, 50, 100 years... the effects of sample size are hardly debateable. you do need longer sims to do what i was explaining. again, i think context of the original post is being ignored or something... what i was explaining requires more than just a good idea of expected results. __________________________ i appologize to the OP. i have a couple people who can't let go of a previous thread. i should have never replied to any of it, because it has very little to do with your thread and question. e.g. RichW's reply "What is the point in generating meaningless mods?" had he read the post he quoted it should have been obvious. instead, he is just taking one sentence out of context and using it as an argument (a logical fallacy). it is obviously personal for him at this point. ______________________ dumbed down version of logical fallacies: https://yourlogicalfallacyis.com/ Last edited by NoOne; 10-31-2015 at 06:09 PM. |
|
Bookmarks |
|
|