 |
09-22-2007, 06:44 PM
|
#1 | |||
|
College Starter
Join Date: Oct 2002
|
FOF Mythbusters: What the heck do these combines mean, anyway?
In both multiplayer leagues I'm involved in (GEFL and IHOF), it appears there is a definite split in drafting strategy. Some guys draft by a multitude of combines, others draft by ratings on the player card screen (aka bars), and I'm sure there's some guys who draft by the adjusted bureau rating. Since the release of FOF 2007, it appears as if those who draft by the "combines-mentality" seem to have a better handle on the draft, and drafting guys with low combines seems to be a lot riskier. I believe part of this was the developer's mindset to attempt to get us away from looking at ratings and to guage statistics as to the true value of many of our players. Although things such as increased scouting error and "creepers" have only seemed to arrive since FOF2K7, there was some previous intent for combines to be of near supreme importance while drafting. According to Quiksand's FOF2K4 Diary, the original plan for when combines were implemented were for there to be no bars or ratings seen. (http://www.operationsports.com/fofc/...991#post267991) It was only by request from beta-testers that bars were included in FOF2K4 and eventually in FOF2K7.
So, are bars completely useless? I think not. As will soon be seen, I think there are somethings that you need the bars to see. However, we have to understand the combines and what some of these numbers mean. With the new patch, I think it's finally time to pull the curtain a little bit and try to guage how and why combine drafting guys "seem" to have an edge. We're aided by Solecismic's decision to upgrade the draft screen a little bit, making it easier to see outstanding numbers and their relation to ratings. Also, I have a theory about the lessening of scouting error, but I'd need time to test it before I make any generalizations. As a starting point, I think some correlations from FOF2K4 still hold. Skydog initially tried to generate interest in this topic, and MIJB#19 was kind enough to share some of his correlations from FOF2K4 as a starting point. An early thanks to SD and MIJB for generating that material here. Today, I start by exploring quarterbacks, and I plan to reveal my findings when I start testing other positions. Of course, please feel free to do any of this testing for yourself. So, enough chatter, let's start the learning! HYPOTHESIS: Simply put, I believe that every combine skill has a stronger correlation with a player's true ability in that skill than what we see on the player screen. With the addition of two new combine scores in the FOF draft, I believe that the developer gave us the keys to more information about the "true" value of bars where scouting error minimizes true ratings found on the player screen. This is a perfect explanation for why we have "creepers". The bars indicated on the player card are our 'scout's' image of players whereas combines are the true ratings where these players will eventually creep to be. So, as time progresses, the scout sees more and more, what we should've initally seen from looking at the combines. TESTING METHOD: I created a 900-player draft file by making a comma-delimited spreadsheet and uploading it through the FOF2K4 draft file converter (which is still backwards compatible with FOF2K7). Started a regular career with X-factor turned off and simulated to Import Draft. With the comma-delimited file open, I'm able to make "extreme" players (guys who are 100 in short passes and zeroes for everything else). I uploaded the file, looked to see if there was a correlation with that skill, re-uploaded, rinsed and repeated. I did this 10 times for each skill, and then moved on to the next skill. As mentioned earlier, I wanted to test quarterbacks first, because this is the posiiton I felt I had the least handle on. ----01/17/07 EDIT - My new testing method allows me to do more trials in less time by comparing maxed out skills and relations to combine numbers globally. Below, you will see looking at "Wide Receivers", a player created with maxed out route running but zeroed out skills in anything else. As you can see, his 40-yard dash is not that impressive; but his SoleCombine numbers are. Each number results ten tests whittled down to an average and those averages rounded and averaged together (a total of fifty trials per combine score). Hopefully, this will result in more trials getting done faster and the completion of this table (finally!) IMPORTANT: Throughout this explanation, I give rough ideas of percentages of how I think these combine scores are correlated. Note that I did no testing to figure out the exact specific formula that goes with each combine score. These percentages given are a rough idea of how much I believe each combine score relates to skill based upon knowledge of what the player ratings actually look like and what combines are telling me. QUARTERBACKS: 40-yard Dash - This combine skill has a strong correlation with scrambling frequency and scrambling rating. However, some of you might not know that there are really "two" scrambling ratings given to each player. There is a "scrambling" rating and a "scrambling-frequency" rating. Most likely, scrambling frequency = how likely the QB is to scramble on a given play whereas scrambling = likeliehood of "quality" scrambling on the play. Keep in mind that ALL players have a rating in every skill, so it is unknown how much the QB's breakaway speed rating (which would be seen if you converted a QB to RB) has an effect on ability. That's another testing that would need to be done. However, players that had extremely high ratings in scrambling also had an excellent 40-yard dash time, those with extremely low ratings would always have extremely bad 40-yard times, and players with mediocre scrambling ratings = mediocre 40-yard times. Conclusion - 40-yard time = how likely the QB is able to scramble and scramble with success. Solecismic - When I originally tested the Solecismic test score, I made a small but significant miscalculation. While the Solecismic rating is still primarily determined by a quarterback's hidden avoid interception rating; it is also significantly related to a quarterbacks personality "intelligence" and read defense statistics. The goods news about this is that it is now easier to determine a quarterbacks avoid interceptions rating because the Solecismic test score only relates to avoid interceptions, intelligence, and read defense. However, the bad news is that there is now no combine correlation to short passing or two-minute drill. Conclusion - Solecismic score is a combination of avoid interceptions, intelligence, and read defense in that order. Think a player's Solecismic score is 50% avoid interceptions, 40% intelligence, and 10% read defense. A high Solecismic score likely indicates a high skill in each of the three above categories. A mediocre score can mean several things. This is what looking at the player card at bars is good for. For example, if a QB has a Solecismic score of 32, but his intelligence is in the upper 70's and his read defense bar "appears" high, then that means that his avoid interceptions is likely mediocre. Similarly, if a player has an extremely high Solecombine score, but his intelligence is low, it means that the quarterback is a major creeper or his avoid interceptions are god-like (either isn't a horrible thing,  )Bench Press - Thankfully, this combine score also scores high on the common sense score. A quarterback with high arm strength means that he has a strong arm and is more likely to be able to complete long and very long passes. The interesting part about this skill is how it's correlated. It appears that strength has a 50/50 correlation between long passes and very deep passes. When a QB had maxed long and deep pass skills, his strength rating was ALWAYS the highest and in red digits. However, when a QB was maxed out in long passes, but zeroed out in deep passes, he still had a high strength number (usually in the high black or low blue) and same when maxed out in deep passes but zeroed out in long passes. Conclusion - Bench press = Combination of long/deep passing. An extremely high strength number indicates a QB with excellent deep and very deep passing skills. A mediocre strength score likely indicates mediocre long arm numbers, however, it can also indicate a player who's very high in long passing, but very low in deep passing (and vice versa). A low strength score = the QB cannot consistently complete passes beyond 20 yards. Agility - Similar to FOF2K4, agility is still a very good barometer for sense rush. However, the agility rating can also be a barometer for screen passes. Keep in mind that screen pass ratings can be controlled by the draft class, but sense rush is randomly generated each time you load. Sense Rush is still the favored correlation here, as players with zero screen passes can still have blue agility numbers with a high sense rush rating. Thus, again, looking at the bars while keeping this in mind may give you an idea about his abilities. If a player has a mediocre agility score, but his sense rush seems high, the sense rush might be overrated but his screen pass rating might be solid. Similarly, if a player is tops in agility, it's safe to assume that his sense rush is very high. However, I can say that I did not see a player maxed out with screen passes and had zero sense rush that had a high agility score. Conclusion - Agility has about a 75/25 correlation with sense rush/screen passes. So, if a players sense rush bar appears very low, but his agility score is very high, it's very likely that the quarterback's sense rush is extremely high, but there is significant masking going on. Broad Jump - I did not expect to find this correlation at all, but that's why we do these things. When I started testing, I had no idea what would give a correlation to medium passes. However, when a player was completely zeroed out in other ratings but was maxed out in broad jump, he always had a moderate-to-high broad jump score. Similarly, third-down passing has a correlation to broad-jump, but it is lower than medium passing. Quarterbacks with a maxed out medium passing rating ALWAYS had a blue (and many times red) combine score. Quarterbacks with maxed out third-down passing, zeroed out everything else had a mediocre combine score (but usually never in the green). Quaterbacks with maxed out third-down passing and medium passing always had a RED combine score (literally 100% of the time). Conclusion - Broad jumping has a 66% medium passing - 33% third-down passing correlation. Meaning, if a QB has an extremely high broad jump score, he's most likely very talented at medium passing. Passing Drill - Much like long/deep passing; accuracy and timing are similarly related to position drill. Maxed out accuracy/timing = red combine score. Maxed out accuracy/zeroed out timing = blue or black combine score Maxed out timing/zeroed out accuracy = blue or black combine score Zeroed out accuracy/timing = green combine score Conclusion - Really no mystery with this combine score. If his combine score is green, he'll have no passing or timing, if red, he'll likely be maxed out in accuracy timing. If the score is blue/black, it could mean that he is actually a 75 rating in accuracy/25 in timing or vice versa, or that he is simply 50/50 in accuracy/timing. Here, again, the bars are a solid indicator of this rating. If a player has a blue position drill rating, and his bars indicate that he has better accuracy than timing, most likely, his accuracy rating will be higher than his timing rating. Once again, here is the short form of ratings and their relative combine indicators: Code:
RUNNING BACKS 40-Yard Dash - Seems to be a very strong indicator of breakaway speed (BRS), as in FOF 2004. However, there appears to be something more going on here. Another skill is at least partly attached to this combine score. With my test subject maxed out in BRS, and zeroed out in every other category, his 40-yard dash number would consistently have a blue color (tapping out around 4.5 secs) but would rarely be in the red. When the 40-yard time was under 4.5, it was usually b/c speed to the outside was filled in at least 25%. However, with BRS completely maxed out, after loading the draft file 25 times, I have yet to see an occasion where both speed to the outside + BRS are completely filled out. This seems to suggest that there is another skill for RB's which impacts both speed to the outside, and in turn, 40-yard dash time. RB's with zeroed out BRS but maxed out in every other category still have a surprisingly high 40-yard time and high speed to the outside skills. Perhaps not in the blue, but fairly high in the black (avg. combine score = 4.65). Code:
Conclusion - Breakaway speed is still the highest correlation. However, speed to the outside is still a significant impactor to 40-yard dash. Blitz frequency appears to have some impact, although very minor. However, it appears that an RB has to have a "trigger" of multiple solid skills before a high speed to the outside rating will be granted to a created RB. Figure a 80/20 split between BRS/STO/. Bench Press - again, very strong correlation to power inside. RB with power inside maxed out and everything else zeroed out; RB subject consistently had red combine score numbers (avg. 23). Code:
Conclusion - Bench press is a 100% correlation to power inside. Open and shut case. Agility - we've always interpreted agility to mean elusiveness, but from what I've seen, agility corresponds to three different ratings (elusiveness (E), third-down running (3R), getting downfield (GD)). and all are roughly 33% correlated to the agility score: Code:
Conclusion - Agility drill is an equal 33/33/33 ratio split amongst elusiveness, 3rd down rushing, and getting downfield ratings. Meaning, if an RB has an agility drill around 7.15, he could either be unbalanced w/ extremely high scores in two of the above three categories with a minimal skill in the other, or, he could simply be very good in all three categories. As always, the best bet is to look at the bars to interpret what this combine score means. Solecismic - In FOF2K4, this skill had the highest correlation with hole recognition. It appears to still be very significant, but there are other factors which have a small impact on the Solecismic score. Route running has a small effect as can be seen below. Code:
Blitz frequency is not directly related to Solecismic, HOWEVER, much like BS and speed to the outside, players with higher hole recognition appear more likely to have a high blitz protection rating. Yet, I'm not sure if blitz frequency itself has a correlation directly with Solecismic as one-dimensional players seem less likely to generate good to great ratings in speed to the outside and blitz frequency. In an additional test, I created an RB with high breakaway speed and hole recognition categories. When both of these ratings were combined, the running back was more likely to have better ratings in speed to the outside and blitz protection, and thus have a better 40-time and Solecismic score. What is making this the case, however, is a chicken-egg theory that I'm not really sure how to test. One last caveat, of any rating I've seen right in testing these combine scores so far, Solecisimic for RB's is the most volitile. I've seen more players that have very little related skill still score very high in Solecismic score. Further, I can confirm that the hidden avoid fumbles rating has no correlation with Solecismic score, not appears to have correation to any combine score that I can see as of yet. Conclusion - Solecisimic score is likely still closely tied to Hole Recognition, but also has a slight correlation to route running (probably 90(HR)/(10(RR). However, be very wary of selecting a running back with a high Solecismic score, but has other mysterious combine numbers lurking around. Broad Jump - For multiple positions, broad jump has been synonomous with endurance. So, it would be logical to assume a similar correlation with running backs. This is the case, however, broad jump also has a very strong correlation to speed to the outside. Code:
Conclusion - Really surprised to see that an RB with absolutely zero endurance can still put up a very high broad jump number if his speed to the outside is also very high. This seems to suggest that broad jump has a 50/50 correlation with endurance/speed to the outside. If an RB has a red combine score in broad jump, it's very likely the RB has very high speed to the outside and endurance (I did not see in any trial an RB with maxed out endurance that had a red combine number). Position Drill - This is the hardest skill to put to a correlation. However, I finally figured out this is because it's heavily tied to blitz pickup (which I have no control over). It also has a relatively minor correlation with route running/third down receiving. From what I've seen, the draft is very relunctant to assign blitz frequency (as well as speed to the outside ratings) to one-dimensional guys. In fact, the most common correlation I saw was that the top guys in the draft (regardless of individual strenghts), usually had a high position drill rating. So, perhaps the most important correlation here is that position drill might be the best indicator (especially if your player is a no-combine guy) that the running back is the real deal skill wise. Conclusion - Position drill likely has a 90/5/5 correlation to blitz pickup/route running/third down receiving. However, the most important correlation is that this might be a strong indicator of the overall ratings of the running back, most likely because the engine does not like to give a high blitz pickup rating to extremely poor guys. However, this assertion would need more testing using non-generated draft files. Comparing RB skills to combine correlations: Code:
WIDE RECEIVERS 40-Yard Dash Code:
Conclusion - A wide receiver's 40-yard dash is primarily related with his "big play receiving" rating. However, a receiver's getting downfield rating also has an appreciable impact on a receiver's 40-yard dash number. Thus, it will be very unlikely for a receiver to have a sub-40 yard dash number (and have a "red" score) unless both his big play receiving and getting downfield skills are high. However, a player with high big play receiving ratings but low getting downfield ratings will still have a 40-yard dash in the low 40's (4.40-4.45) and have a blue score. A player with a mediocre or worse 40-yard dash is likely mediocre (at best) in big play receiving. Thus, 40-yard dash = about 70% big play receiving and 30% getting downfield. Solecismic Score Code:
Conclusion - A receiver's solecismic score is a clear 50/50 ratio to his intelligence rating and route running ability. Thus, interview results become clear here. Where a receiver has a red combine number (high thirties/low forties), there is little doubt that he will have high route running (even with high intelligence). However, a combine number in the blue categories (low-mid thirties) means that interviewing the player becomes a necessity. If that player has low intelligence but a blue combine score; he likely will have high route running. In that vein, receivers with mediocre Solecismic test scores will likely have low route running unless they're dummer than a rock. Strength Code:
Conclusion - Pretty open and shut that a receiver's strength rating is 100% correlation to his courage ability. The more reps the receiver has, the more likely he is to make the big catch across the middle. No real debate looking at these numbers. Agility Drill Quote:
Conclusion - A receiver's agility drill score is 100% correlated to his getting downfield rating. A player with a red combine score (below 6.85) will likely have an 80+ rating in getting downfield. Similarly, a player with a blue combine score would likely have a getting downfield rating between 50-80; a player with a black combine score would likely have a getting downfield rating between 20-50, and a receiver with a zeroed out getting down field will be between 0-20. Position Drill Code:
Conclusion - A receiver's position drill is about 65% avoid drops and 35% adjust to ball. Because adjust to ball is one of those ratings that is randomly assigned when a draft class is loaded/created; it is hard to determine the exact amount related, bu tit most definitely is related. This is why there is the odd occasional outlier (as seen in the getting downfield trials) in the above chart. Sill, a player with low adjust to ball ratings will still have a position drill frequently above 50. Broad Jump Code:
Conclusion - As was previously predicted; a receiver's broad jump score is correlated to his kick/punt returns skills (50% punt returning/50% kick returning). A receiver blessed with high skills in both kick and punt returning will have a red combine number (interestingly; that red combine number is 11 feet). A receiver maxed out in punt returning, but not kick returning, will still have a blue combine number and be in the 10-foot range. Player's with black broad jump numbers would be mediocre in both categories. Comparing Receiver Skills to Combine Ratings Code:
OTHER DRAFT THOUGHTS - My hypothesis was largely proven correct. You can have a basic idea of what each quarterback will look like without even looking at his player card. However, it is still important to take certain bars into consideration when making a decision in who to draft. Hopefully, however, this will explain the creeper phonemonon and how to spot them and why they exist. If a player has an extremely high stength score, but his long/deep passing bars are in the 15-25% range, most likely, this is scouting error going on and his bars (when completely developed) will be in the 60-70% area. It's also important to know that, like in life, there are no sure things. FOF does a good job of draft generation that you might occasionally get a "workout wonder" in a specific category. However, just like in drafting, these findings are "more-than-likelys" of what combine skills relate to each position. Simply put, if you want to do well drafting in FOF, you must take everything into account. But, understand that combines are not only important, but essential to understand for success in drafting. It was a great learning experience to do this testing, and I'm happy to share it with you. I do plan on looking at other positions and will post my results over the next few weeks or so. Feel free to post comments or thoughts. SKYDOG EDIT PER RKG REQUEST...
Spoiler
RKG, feel free to fill this puppy in as the thread grows. Last edited by RedKingGold : 01-16-2008 at 11:02 PM. |
|||
|
|
|
09-22-2007, 06:53 PM
|
#2 |
|
Morgado's Favorite Forum Fascist
Join Date: Oct 2000
Location: Greensboro, NC
|
Good stuff, RKG. Very good stuff.
__________________
The media don't understand the kinds of problems and pressures 54 million come wit'! |
|
|
|
|
09-22-2007, 07:06 PM
|
#3 |
|
Morgado's Favorite Forum Fascist
Join Date: Oct 2000
Location: Greensboro, NC
|
Oh, and when RKG IMed me about this a while ago, I felt like a real dummy for not thinking of using the methodology of importing draft classes with extreme players when I worked on this a while ago.
{Tips cap to RKG}
__________________
The media don't understand the kinds of problems and pressures 54 million come wit'! |
|
|
|
|
09-22-2007, 09:00 PM
|
#4 | |
|
College Starter
Join Date: Oct 2000
Location: Berkeley
|
Quote:
|
|
|
|
|
|
09-22-2007, 09:10 PM
|
#5 |
|
College Starter
Join Date: Oct 2000
Location: Berkeley
|
DOLA, great work!
|
|
|
|
|
09-22-2007, 09:23 PM
|
#6 | |
|
College Starter
Join Date: Oct 2002
|
Quote:
Taken from the FOF2K4 Draft File Converter CSV Help File: Use 375 for players who really shouldn't have ratings at that position. For instance, defensive tackles should receive 375 for the Avoid Interception rating. All players are rated for all categories, though these ratings are ignored if not relevant to the player's position. |
|
|
|
|
|
09-22-2007, 09:28 PM
|
#7 |
|
lolzcat
Join Date: Oct 2000
Location: sans pants
|
Really good stuff here. A lot of it seems intuitive, but there are definitely skill translations that surprised me.
__________________
Superman was flying around and saw Wonder Woman getting a tan in the nude on her balcony. Superman said I going to hit that real fast. So he flys down toward Wonder Woman to hit it and their is a loud scream. The Invincible Man scream what just hit me in the ass!!!!! I do shit, I take pictures, I write about it: chrisshue.com |
|
|
|
|
09-22-2007, 10:09 PM
|
#8 | |
|
College Starter
Join Date: Oct 2000
Location: Berkeley
|
Quote:
|
|
|
|
|
|
09-22-2007, 10:56 PM
|
#9 | |
|
College Starter
Join Date: Oct 2002
|
Quote:
Sir Bustamove doesn't think so:   |
|
|
|
|
|
09-22-2007, 11:21 PM
|
#10 | |
|
Pro Rookie
Join Date: Oct 2000
|
Quote:
RKG, The ratings ranges start at 375 in the data files. So 375 actually equals 0. I would guess it is unlikely these ratings have changed since 2k4. So even though every player has a value in those slots, the ones that don;t apply to their position are always 0. |
|
|
|
|
|
09-22-2007, 11:23 PM
|
#11 |
|
Pro Rookie
Join Date: Oct 2000
|
Maybe I am not completeky familiar with how the position change works internally, but I wasn't under the assumption it only took the ratings the player already had in the other categories when you change positions. I guess maybe I am wrong though.
|
|
|
|
|
09-22-2007, 11:27 PM
|
#12 | |
|
College Starter
Join Date: Oct 2002
|
Quote:
Yes, I am aware of this. 375 = 0, 625 = 100 It's suggested so you don't have DT's running around with high-flying QB skills (not that those guys would convert over anyway). EDIT: What Daimyo's suggesting is that if I place a 625 in avoid interceptions for a DT, then that rating is ignored when the draft is imported. My example above gives evidence otherwise. Bustalot was an RB who I gave 375 (or zeroes) for every single running skill, but gave 625 (100 ratings) for every other skill. When he's converted to other positions, he may lose a percentage indicated in-game (aka "He will have 68% of his potential after the switch), and in this case, he would have 68% of a 100 (or, a skill of 68). Last edited by RedKingGold : 09-22-2007 at 11:37 PM. |
|
|
|
|
|
09-22-2007, 11:33 PM
|
#13 | |
|
n00b
Join Date: Aug 2007
|
Quote:
That is one of the worst examples I've ever seen.  I don't even understand what this could be designed to show.If you want to see if a offensive player is given values in defensive ratings, you have to change that player's position multiple times and compare the results. If he was given values in those ratings he would have consistently, either by 1-9 or by exact value, the same values. To see that this isn't the case all you have to do is take a cornerback and change him to receiver, reload and repeat a few times. The results will be radically different, indicating neither preference to specific ratings nor consistent values. Beyond this, a rookie cornerback changed immediately after being drafted will show not show a correlation to his combine scores. Perfect 40-yard dash and Agility scores as a cornerback won't make him any more likely to have Getting Downfield/Big-Play Receiving as a receiver. To see this even more blatantly, change that cornerback back to CB from being a receiver and see how his ratings compare to his original cornerback ratings. Certain ratings like interceptions for defenders and offensive line ratings for D-Linemen may be there, but to say that this holds true for all ratings is not remotely accurate. |
|
|
|
|
|
09-22-2007, 11:39 PM
|
#14 | |||||
|
College Starter
Join Date: Oct 2002
|
Quote:
As mentioned earlier, it is meant to show that (at least in a user generated draft file), you can attribute certain ratings (such as interceptions/pass blocking/etc.) which will carry over and be imported into the game. The example taken above in my response to Primelord's response is a better example of what I'm talking about. Quote:
I can and will do this tomorrow, but I have already tried similar tests and seen that (with some variance), these ratings do carry over. Quote:
There are some ratings which are completely game-generated that I have no control over (Bn'R, adjust to ball, play diagnosis, etc.). These will be completely random each time I upload a new draft file. However, certain ratings will stay consistent. Further, I never made any connection between draft scores and these hidden abilities. To my knowledge, there is no way to tell if a player from a game-generated rookie file has hidden ratings, other than to guess-and-tell based on bars pictured. Quote:
This would be comparing different things, however. Whenever you flipflop such a dramatic position (from CB-WR-CB again), that CB will most likely not have the original ratings he had before. Quote:
Care to produce direct evidence to prove me wrong? Last edited by RedKingGold : 09-22-2007 at 11:49 PM. |
|||||
|
|
|
|
09-23-2007, 12:01 AM
|
#15 |
|
n00b
Join Date: Aug 2007
|
CB1 changed to Flanker two seperate times
AD 00/43 00/69 GD 37/75 44/85 RR 00/43 00/55 3D 42/61 48/71 BP 60/60 100/100 Co 32/50 54/85 AB 00/66 00/62 PR 56/77 46/63 KR 43/77 35/64 En 50/77 50/77 ST 33/52 33/52 This is not the same result. The ratings he did have (in bold) stayed the same, the ratings he did not have were different. (the current ratings of 0 in AD, RR, AB are a result of 0% developed at the FL position). |
|
|
|
|
09-23-2007, 12:03 AM
|
#16 | |
|
College Starter
Join Date: Oct 2002
|
Quote:
What about scouting error? What was the % of his skill change per switch? Combines/year/etc.? Last edited by RedKingGold : 09-23-2007 at 12:04 AM. |
|
|
|
|
|
09-23-2007, 12:12 AM
|
#17 |
|
College Starter
Join Date: Oct 2002
|
Further, that test really proves nothing to the point that I was trying to get across. Once again, the point I was trying to get across was that, when a player is "imported" into FOF, he has ratings in other areas which do not translate to his natural position that are not zeroes (which is what Daimyo was asking).
That's all I was asserting, not that there are 335 LB DT's wandering around in FOF with hidden killer QB skills. Also, whether or not those skills actually translate into success in the game engine is anyone's guess, I haven't tested that because there's really no need too. |
|
|
|
|
09-23-2007, 12:53 AM
|
#18 | |||
|
n00b
Join Date: Aug 2007
|
Quote:
All that example showed was that when a player is moved from RB to SS he has defensive ratings afterwards. That didn't address in any way that they came from values listed in a draft file. It would have to, at minimum, show that those values were consistent (in terms of correlation between ratings as position switches have a degree of randomness) and most importantly that those correlations reflected what was in the draft file. Quote:
The comment about combine scores wasn't in direct response but instead to mention something I felt was closely tangental to what I was saying, hence why I prefaced that comment with "Beyond this,". [edit: the word choice was obviously poor as it coud just as easily mean "additionally"] Quote:
Of course not, but my point isn't that the ratings will be the same. My point was that there will be seemingly no preference between ratings. While a corner may drop in overall ratings, although they do occasionally improve over that transition, his ratings should still be similar as far as what his best ratings are and what his worst ratings are. This seems to show no preference between ratings, a corner with only great MtM could end up with no MtM and just run defense and zone for example. Last edited by jkat : 09-23-2007 at 12:57 AM. Reason: [*] |
|||
|
|
|
|
09-23-2007, 01:06 AM
|
#19 | |||
|
College Starter
Join Date: Oct 2002
|
Quote:
However, with scouting error present, there would be no way to prove that other than playing players out of those positions. Are you suggesting the game randomly draws those ratings when a player is transferred over? That may be, but you have about as much chance of showing me that as I do of proving that the ratings shown after position change are the same ones that were uploaded in the initial help file. Quote:
Well, I'd appreciate it if you would be sure to preface comments like that in the future as coming from your own opinion and not falsely attributed to what I'm trying to say with my comments. Quote:
Once again, I request of you to prove that assertion. With the challenges of scouting error, I believe you have as much chance of proving this as I do of proving exact ratings. In my tests with this phonomon, I''ve found that the rating shown on the player card is within 15-20 points of the actual rating(s) placed in the draft file. I believe that this fluctuation of points is due to the high level of scouting error found in this game. With pass coverages, you've seemed to figure out a way to answer a seemingly unanswerable problem. Perhaps, you can use some of this wisdom and testing to prove your assertions in this manner. However, I would prefer if you were to do so in another thread (for the purposes of this thread have gone completely off-kilter. Last edited by RedKingGold : 09-23-2007 at 01:07 AM. |
|||
|
|
|
|
09-23-2007, 03:17 AM
|
#20 | ||||
|
n00b
Join Date: Aug 2007
|
Quote:
I didn't suggest anything about how the values are transferred over from the file. I was addressing the example you had used as I didn't feel it showed anything as it consisted of only a picture of a RB and a picture of him changed to SS with no additional explanation. Quote:
I don't understand how I you say I attributed that comment to you. It had a similarity to what I was saying and it was relevant to the initial subject of the thread as far as what ratings affect combine scores; as it would seem to indicate that a cornerback's Getting Downfield/Big-Play Receiving (and other ratings) doesn't affect the cornerbacks 40-yard dash/Agility (and other combine scores) and conversely that a cornerback's combines don't affect their out of position stats. Though it wasn't in direct response to that specific post it was within the subject of the thread. Quote:
I stated, "Certain ratings like interceptions for defenders and offensive line ratings for D-Linemen may be there, but to say that this holds true for all ratings is not remotely accurate." In the example I posted above (from a specfic player file, not draft file) the cornerback had been at 0 in punt and kick returns before the change. Despite that, he had significantly more afterwards and changing every receiver and defensive back in that career, with very low return ratings, to the opposite side of the ball they all increased towards the 40-70 range. This seems to show something more than just using their specific ratings as listed, in regards to players given a specific rating in a player file (again not neccessarily to the draft files in question). The draft file may work differently than that and if the values are consistent (just over a generalized range, they don't need to be anywhere near exact considering the position switch) it would seem to indicate that it would be accurate to say 'it holds true for all ratings', at least with the exceptions of those ratings that you can't change in the file. It's quite possible the ratings (at least ignoring return ratings in a player-file) match up close enough after that first position change and may hold true for all the ratings from a draft-file. As far as proof, if you use that same player and save before the position swtch and then make the switch 3-5 times it should be adequate. If the overall shape of his bars, as far as whats high and whats low is close it would indicate that it is being based off his ratings. Even if the return bars aren't consistent (though they may well be from a draft-file) it would show a lot of support for your point. While it may not completely disprove "to say that this holds true for all ratings is not remotely accurate.", if it showed it held true for all ratings shown on that one player (excepting what can't be changed in the draft-file) it would be enough to have the weight of evidence in vast support of your argument. That case of course wouldn't have scout error as a factor as you are just using the one specific player. Unless you have the idea that the possibility of scout error is reapplied when a player changes position which, while it would seem peculiar to me, I suppose can't be discounted. As far as switching a player's position and then switching back getting a different overall blend of ratings for that player; I thought this was well known as a way to essentially re-roll a player and was one of the main reasons extensive house rules covering position changes were so widespread. That pretty much covers my interest in the matter, at least for this thread. I don't mean that in the sense of walking out of the discussion but to, as mentioned, avoid adding anymore significant clutter to the thread. |
||||
|
|
|
|
09-23-2007, 03:52 AM
|
#21 |
|
College Prospect
Join Date: Sep 2005
|
Great work RKG.
|
|
|
|
|
09-23-2007, 04:59 AM
|
#22 |
|
Pro Starter
Join Date: Sep 2003
Location: Toledo - Spain
|
Awesome stuff RKG.
It's nice to see how Jim payed as much attention to the draft combine numbers correlated to real ratings instead of using it just as cosmetics.
__________________
|
|
|
|
|
09-23-2007, 08:45 AM
|
#23 |
|
Mascot
Join Date: Jun 2007
|
Great work. I like how you picked what I believe to be the toughest position to assess. I've got a draft starting today, so could you hurry it up with the other positions!
 Anyone have a percentage idea/guess as to how often good combine scores are a bluff (player busts or doesn't develop as expected in correlated skillsets)? From my perspective, it seems very low, like in the less than 5% range. |
|
|
|
|
09-23-2007, 08:57 AM
|
#24 |
|
High School Varsity
Join Date: Feb 2007
Location: Budapest
|
Where does the overall draft grade fit in?
|
|
|
|
|
09-23-2007, 11:15 AM
|
#25 |
|
Pro Starter
Join Date: Sep 2005
Location: TX
|
cool stuff. now to go look at the qb's in my upcoming online draft
|
|
|
|
|
09-23-2007, 11:51 AM
|
#26 |
|
High School Varsity
Join Date: Dec 2003
Location: Downriver, MI
|
Great stuff RKG !
|
|
|
|
|
09-23-2007, 11:54 AM
|
#27 |
|
College Starter
Join Date: Oct 2002
|
Thanks for the comments guys.
I'm halfway done with running backs (which is actually a lot tougher than I thought) and hope to get my results posted tonight. Unfortunately, I've got some real-life stuff to get done today, but I've only got a few more things to look at for RB's anyway. |
|
|
|
|
09-23-2007, 04:59 PM
|
#28 | |
|
Coordinator
Join Date: Oct 2000
Location: Maassluis, Zuid-Holland, Netherlands
|
Quote:
__________________
* 2005 Golden Scribe winner for best FOF Dynasty about IHOF's Maassluis Merchantmen * Former GM of GEFL's Houston Oilers and WOOF's Curacao Cocktail |
|
|
|
|
|
09-23-2007, 09:01 PM
|
#29 |
|
Pro Starter
Join Date: Nov 2000
Location: Cary, NC, USA
|
FWIW, in 2k4 (and, for that matter, all versions of FOF that I've worked with, including TCY), all players had spots for all ratings. In FOF, generally these were specific to their own position, with some exceptions - quarterbacks, for instance, had ratings for all running back ratings. TCY actually had it much more spread out - a ton of defensive players would have offensive ratings and vice versa. The other ratings would be empty.
Edit: one exception - only QBs had QB ratings. There's a much cleaner way than position changing to test the draft bit - but I don't have 2k4 installed for the draft file generator. If someone wants to put together a sample one, I can swap the positions straight-up in the file, and test that way. Last edited by Celeval : 09-23-2007 at 09:02 PM. |
|
|
|
|
09-23-2007, 09:04 PM
|
#30 |
|
Pro Starter
Join Date: Nov 2000
Location: Cary, NC, USA
|
Dola... while the player data is encrypted in the 2k7 files, there are still some things to be gleaned - the encrypted player records are all the same size; as are the ratings. The footprint of those locations are identical for all players. I would hazard a pretty good guess that there are still values stored for all ratings for all players (again, except QB ratings for QBs).
|
|
|
|
|
09-23-2007, 10:30 PM
|
#31 |
|
College Starter
Join Date: Oct 2002
|
Bump for Running Backs in the top post.
Running Back correlations were very hard to figure out as speed to the outside/blitz pickup figure heavily into at least two (if not three) of the ratings. However, I think I did enough tests to feel satisfied with the correlations that I posted. Ben, if you could update the chart? (me read HTML no good too much )Receivers are next on the list, and I think those are pretty easy to figure out. It should be done and updated by tomorrow, but might have to wait until Tuesday night. Last edited by RedKingGold : 09-23-2007 at 10:30 PM. |
|
|
|
|
09-23-2007, 10:34 PM
|
#32 | ||
|
College Starter
Join Date: Oct 2002
|
Quote:
As per the Solecismic help file: Quote:
So, to my recollect, the "player grade" (grade on the left) is what all the scouts in the league think the player's overall value to be based on ratings and combine scores. The adjusted rating simply adjusts the overall rating by order of importance in the draft (so kickers aren't given first round grades) |
||
|
|
|
|
09-23-2007, 10:35 PM
|
#33 | ||
|
College Starter
Join Date: Oct 2002
|
Quote:
Quote:
Thanks and thanks. This is all I was trying to prove with my simple switching RB-SS example above. |
||
|
|
|
|
09-23-2007, 11:49 PM
|
#34 |
|
High School Varsity
Join Date: Dec 2003
Location: Downriver, MI
|
Question ..
I'm having a little bit of trouble turning your test results into expected ratings. I can follow what your saying, but what doesn't add up - is if I do it backwards. For example : For QBs - the Solecismic Test (ST) is graded : 50% Avoid Int 30% Short Passes 10% Read Defense 10% 2 Minute Drill OK, I can understand - you are saying you're perceived importance to those ratings are the listed values. But how does that work when I am trying to evaluate what a player will be rated in a specific rating ? What I mean - Of the ratings influenced by the ST - none are influenced by any other combine score. So, let's take "Read Defense" (RD). What does that tell me about that expected rating ? 1. His expected RD is 100% effected by ST (since no other score effects this rating). 2. His expected RD is 10% effected by ST (the listed value). - If so, what effects the other 90% ? Is it random ? Anyway - just trying to better grasp this, I appreciate your testing & work. Good stuff .PS - Also, am I right in assuming on your RB write up that all the "Blitz Frequency" comments refer to "Blitz Pickup" ? Last edited by DougW : 09-23-2007 at 11:54 PM. |
|
|
|
|
09-24-2007, 12:22 AM
|
#35 |
|
High School Varsity
Join Date: Dec 2003
Location: Downriver, MI
|
NM - Duh; I get it. The game just generates players and ratings - and not everything is made up completely of the combine scores (as in RL) - the bars "may" tell the rest of the tale, or may be masked.
The study simply shows how to help to possibly unmask certain players. |
|
|
|
|
09-24-2007, 03:44 AM
|
#36 | |
|
Morgado's Favorite Forum Fascist
Join Date: Oct 2000
Location: Greensboro, NC
|
Quote:
Done, but I'm not using HTML, just WYSIWYG. 
__________________
The media don't understand the kinds of problems and pressures 54 million come wit'! |
|
|
|
|
|
09-24-2007, 05:39 PM
|
#37 |
|
High School JV
Join Date: Nov 2006
|
I think it would be interesting to try this in reverse, take guys with maxxed out ratings and then zero them out one by one and see the skill effects, I wonder if it's the same as starting at a zero and working up.
|
|
|
|
|
09-26-2007, 08:15 PM
|
#38 |
|
College Starter
Join Date: Oct 2002
|
Don't worry guys, I haven't abandoned or forgotten about this. The past two days have been absolutely crazy (even got my car towed today, WOO-HOO!), but I'm about to start cracking on receivers (which should be pretty easy) and hope to have that posted by tomorrow night.
|
|
|
|
|
09-29-2007, 02:49 PM
|
#39 |
|
Pro Starter
Join Date: Sep 2005
Location: TX
|
thanks redking
|
|
|
|
|
10-07-2007, 10:35 PM
|
#40 | |
|
High School Varsity
Join Date: Dec 2003
Location: Downriver, MI
|
Quote:
I sure hope you're still planning on finishing this thread. Very informative stuff . |
|
|
|
|
|
11-12-2007, 06:04 PM
|
#41 | |
|
Mascot
Join Date: Nov 2006
|
Quote:
RedKingGold, I was wondering if you were still planning on getting back to this. Also, if your still pressed for time (aren't we all  ), is there anything that I could do to help out?Thanks, Nogram |
|
|
|
|
|
11-22-2007, 08:37 AM
|
#42 | |
|
n00b
Join Date: Mar 2007
|
Quote:
RKG - update? I'm loving this stuff but I'm not savy enough with stats to put it together, sure looking forward to the next installments.... Thanks for the work!!!!!!!
__________________
APFL - Detroit Lions TFL - Indianapolis Colts I guess I bleed BLUE!! |
|
|
|
|
|
11-22-2007, 10:06 AM
|
#43 |
|
College Starter
Join Date: Oct 2002
|
I'm still planning on coming back to this, most likely between Christmas and New Years. From November 1st til December 17th is my busy time, cutting drastically into my FOF/free time. For now, however, this is in my catalog of FOF projects (behind helping Rodney with Extractor -- it's coming, I swear!
) |
|
|
|
|
11-25-2007, 07:16 PM
|
#44 |
|
Pro Starter
Join Date: Sep 2005
Location: TX
|
goodie
|
|
|
|
|
12-13-2007, 01:44 PM
|
#45 |
|
Mascot
Join Date: Feb 2007
Location: London
|
This is one of the most awesome threads ever.
Bumping for greatness and so hopefully RKG won't forget about it Last edited by PiemasterUK : 12-13-2007 at 01:44 PM. |
|
|
|
|
01-16-2008, 11:04 PM
|
#46 |
|
College Starter
Join Date: Oct 2002
|
Bumpity for good news.
1. I used a new testing methodology which allows me more trials in less time. Using it for receivers; I was able to cut the time needed for previous testing down by 2/3. 2. I re-tested quarterbacks a little bit b/c of a flaw I found in testing. Please re-read quarterbacks to see the change (hint: Solecombine score) 3. Receivers have been tested and added. Offensive lineman are next (and will hopefully not take too long) SD, please do your thing with the WSYSIWHSGHSG Chart. Last edited by RedKingGold : 01-16-2008 at 11:05 PM. |
|
|
|
|
01-17-2008, 01:35 AM
|
#47 |
|
College Benchwarmer
Join Date: Sep 2003
Location: Back in Norway
|
Nice timing.
|
|
|
|
|
01-17-2008, 08:59 AM
|
#48 |
|
Pro Starter
Join Date: Oct 2000
Location: Cary, NC
|
Excellent...
__________________
-- Greg -- Author of various FOF utilities |
|
|
|
|
01-17-2008, 10:49 AM
|
#49 |
|
n00b
Join Date: Jan 2007
|
Great research! I had been looking at the intelligence thing all wrong. I had thought that high Intelligence would reflect high Read defense, hole recog, Route running and so on. I wouldn't have suspected that high intelligence takes away from these ratings.
|
|
|
|
 |
| Currently Active Users Viewing This Thread: 1 (0 members and 1 guests) | |
| Thread Tools | |
|
|
