NASCAR FORUM

Comments

  • gje627

    Bristol Motor Speedway — “The World’s Fastest Half Mile”, “Thunder Valley”, “The Last Great Colosseum”, “The Bullring”

    Schedule — all times are Eastern Time (ET)

    Practice 1: Friday, August 16, 11:05-11:55 AM (NBCSN)
    Practice 2: Friday, August 16, 1:05-1:55 PM (NBCSN)
    Qualifying: Friday, August 16, 5:35 PM (NBCSN)
    Technical Inspection: Saturday, August 17, 2:00 PM (No TV)
    Race: Saturday, August 17, 7:46 PM (NBCSN); Lineup Lock 7:30 PM

    Inspection Protocol

    Impound Race: Cars will be inspected prior to race.

    Fail 1st Time: Qualifying time disallowed; car/driver start at the rear. Driver scored from the rear and not from starting position based on qualifying. Fail 2nd Time: + Pit crew member ejected for race. Fail 3rd Time: ++ Driver will pass through pit row on Lap 1 (more than likely Lap 2). At Bristol this is a significant penalty since it is a short track.

    Note: If multiple cars fail (i.e. go to the back), the starting positions of cars going to the back is set by owner points and not driver points. For most cars this is the same; however for teams that employ different drivers for different races this may be relevant (see Rick Ware Racing). The difference between owner points and driver points is that owner points are simply the points attached to a car number and driver points are attached to a driver. Pretty sure everyone here knows this already, but on the off-chance someone does not … this is the explanation.

    Race Details

    — 39 Drivers Entered; 500 Laps, 266.5 Miles
    — Stage 1: Lap 1 – 125; Stage 2: Lap 126 – 250; Stage 3: Lap 251 – 500
    — Expected Fuel Window – 170-180 Laps.

    Note: fuel window, fuel cell capacity is an estimate as of 8/12/2019.

    Fantasy Relevant Track Details/Rankings 2005 to 2019 (29 Races), 24 Tracks (Charlotte Motor Speedway and Charlotte ROVAL considered 2 Different Tracks)

    — 0.533 mile Short-Track Oval. 24 to 30 degree banking in turns, 6 to 10 degree banking in straights. Straights 650 long, 40 feet wide.
    — Winner Average Margin of Victory: 0.912 seconds; Low – 0.064 seconds (2007 – Race #1); High – 4.652 seconds (2005 – Race #1); Finishes under caution 2014, Race #1.
    — Rank Most Green Flag Passes: 10 of 24 Tracks (15.78 Green Flag Passes per 100 Miles)
    — Rank Most Quality Passes (Top 15 Passes): 3 of 24 Tracks (6.80 Quality Passes per 100 Miles)
    — Rank Most Lead Changes: 4 of 24 Tracks (6.36 Lead Changes per 100 Miles)
    — Rank Most Cautions: 2 of 24 Tracks (3.97 Cautions per 100 Miles)
    — Rank Highest % Drivers Finishing on Lead Lap: 23 of 24 Tracks (39.07% per Drivers/Race)
    — Rank Highest % DNF Drivers – Mechanical: 20 of 24 Tracks (6.72% per Drivers/Race)
    — Rank Highest % DNF Drivers – Crash: 5 of 24 Tracks (8.35% per Drivers/Race)
    — Rank Highest % DNF Drivers – Overall: 10 of 24 Tracks (15.07% per Drivers/Race)

    Note: Data and Rankings are Cup series only. Green Flag Passes, Quality Passes standardized to 100 miles per driver to account for differences in miles per race and number of drivers in field. Lead Changes and Cautions standardized to 100 miles for entire field. Lead Lap and all DNF categories standardized to drivers per race. All statistics updated through most recent race (Michigan International Speedway, 8/11/2019).

    Comparable Tracks (alphabetical order)

    Darlington Raceway, Dover International Speedway, Homestead-Miami Speedway, Las Vegas Motor Speedway.

    Note: For fantasy and analytical purposes Bristol really is a unique track. For analytical purposes, Dover International Speedway is the most comparable, but this is still somewhat of a stretch.

    Last Six (6) Race Winners

    — 2016-23: Kevin Harvick (start 24)
    — 2017-08: Jimmie Johnson (start 11)
    — 2017-24: Kyle Busch (start 18)
    — 2018-08: Kyle Busch (start 1)
    — 2018-24: Kurt Busch (start 9)
    — 2019-08: Kyle Busch (start 17)

    Last Six (6) DK and FD Fantasy Points Leaders

    2016-23 Kevin Harvick (DK FPts 133.50) — Kevin Harvick (FD FPts 117.3)
    2017-08 Kyle Larson (DK FPts 110.50) — Jimmie Johnson (FD FPts 106.1)
    2017-24 Kyle Busch (DK FPts 145.00) — Kyle Busch (FD FPts 117.1)
    2018-08 Kyle Larson (DK FPts 150.50) — Kyle Larson (FD FPts 112)
    2018-24 Clint Bowyer (DK FPts 96.00) — Joey Logano (FD FPts 104)
    2019-08 Joey Logano (DK FPts 117.00) — Kyle Busch (FD FPts 108.1)

    Data and Analytics Links – Archived, 8/19/2019, 8:12 AM MT

    Pre-Race

    DK and FD Base Points Grid
    DK and FD Projections — Base and Dominator
    Start Position Analysis — DK Only
    DK and FD Minimum Projected FPts Needed to Cash and Win — GPP Only
    Practice through Qualifying, Including 10-Lap Averages
    Bristol Qualifying Position and Owner Points

    Post-Race

    DK and FD Driver FPts and Top 500 Optimal Lineups

    Race Data, News, and Information

    Stevie’s NASCAR Package
    Jayski
    Motor Racing Network
    Racing Reference

    Race Day Breaking News and Weather

    Stephen Young Twitter – stevietpfl
    Bob Pockrass Twitter
    NASCAR Weather

    Note: As needed information above will be updated.

    Any errors and/or omissions above are mine and mine only. If someone identifies an error, please respond by posting a comment and I will correct.

  • gje627

    @Toddalan2286 said...

    If we are using practice as a predictive stat and we can’t predict crashes then I would consider removing those as data manipulation.

    Sorry, I hate to Debbie-Downer, but the biggest part of data analytics is cleaning, standardizing, and normalizing data when necessary.

    Regarding taking a sample of 1 race; technically that is not a sample, it is the population.

    Therefore, even if it did have predictive value, it would only have predictive value for the exact same race you are running the data from. Thus, again even if it were predictive, it would not have predictive value for any other race except for the exact race you are looking at. Thus, if the Spring 2019 race were run again, under exactly the same circumstances, with the exact same drivers, with the exact same practice speeds, etc.; it could possibly have some value for the Spring 2019 race; but not for any other race.

    Really, not trying to be an a**hole, but I think it’s important to only use good numbers.

  • gje627

    @Toddalan2286 said...

    We will agree to disagree. You will run your correlation with more data points and probably get a result of somewhere between -.325 and -.450 give or take and that won’t tell us much. My correlation is stating for that particular race practice was a moderate indicator for fast laps.

    Saying practice doesn’t matter is misleading people. Practice gives us an idea of who can be competitive, it’s a data point we have given to us prior to the event. It may not predict exactly how many fast laps or where they finish but it’s important to use it in different ways. Below is a stat I’d personally like to keep but I’ll tell it to make my point.

    Since Texas 12 of the last 13 (might be 13 of the last 14) non restrictor plate races have had at least 1 driver ranking 1 or 2 in first practice in the optimal lineup, the one exception being the coke 600. Again practice gives us more data to look for patterns and to evaluate the field.

    I give up ….

    Your data/correlations are NOT statistically valid and HIGHLY misleading.

    I don’t think this is a general disagreement. If you said 2 + 2 = 4, and I said 2 + 2 = 10 ….. ?

    Again, not trying to a jerk ….

    I’ll post my numbers in a second and the reason why fast laps and practice can’t be compared to produce a statistically valid correlation.

    Also, I did not say practice was completely useless. Rather, I said for analytical purposes there are problems.

  • gje627

    Okay, not worth anything … but here are my numbers.

    1) All races to date between 2005 and 2019. All drivers. Races n = 29; Drivers n = 1221.

    Spearman’s = +0.5111

    2) All races to date between 2005 and 2019. Only includes drivers that actually recorded a fastest lap in the race. Races n = 29; Drivers n = 874.

    Spearman’s = +0.3833

    3) ) All races to date between 2005 and 2019. All drivers except for drivers that not participate in final practice. Races n = 29; Drivers n = 1219.

    Spearman’s = +0.5139

    4) All races to date between 2005 and 2019. Only includes drivers that actually recorded a fastest lap in the race and participated in final practice. Races n = 29; Drivers n = 874.

    Spearman’s = +0.3833

    Per before, the fatal flaw here is that not all drivers record a fastest lap in any given race, where all drivers that participate in the final practice record a speed, or a speed rank relative to other drivers. In my sample between 2005 and 2019, 346 of 1221 drivers did not record a fast lap; meaning that for 28% of all drivers (nearly 1/3 of any given field) these results would have no meaning whatsoever. With that said, if you wanted to draw a “kind-a” conclusion, numbers 2 and 4 would be best; though I still wouldn’t make any concrete conclusion using those numbers either.

  • Toddalan2286

    Yes I understand you have to clean data and ensure it’s normal etc, and yes my correlation was for one race, I pointed that out. Do I know everything about data analytics and statistics?Not even close, am I open to learning? absolutely! But when someone immediately attacks my methods that I’m willing to share and have productive conversation about it takes the fun out of doing this each week.

    All I’m trying to say is in my opinion removing DNFs does us no good for DFS purposes. I also don’t fit my own purposes do correlations across multitude of races, each race has its own personality. This is my logic, If Kyle Busch has the best speed in practice and drives away from the field in the first 50 laps but then wrecks in traffic, why would we remove that from a correlation (for DFS purposes, not passing an exam in statistics)? I understand he didn’t run all the laps, i get it, so his total is skewed, and it will alter the correlation, but I WANT that for DFS value at the end of the race. If he was fastest and didn’t give me DFS value I want that to be reflected in MY metric.

    I’m not trying to be a statistician I’m trying to win money at DFS. And if that means I have to be creative by looking for patterns such as the stat I throughout there about practice then so be it. It has helped me… and may help others. I think it’s a good thing for people in here to hear these different perspectives and not be over analytical at all times. To me, that correlation serves a purpose, for that race practice meant something.

  • gje627

    @Toddalan2286 said...

    Yes I understand you have to clean data and ensure it’s normal etc, and yes my correlation was for one race, I pointed that out. Do I know everything about data analytics and statistics?Not even close, am I open to learning? absolutely! But when someone immediately attacks my methods that I’m willing to share and have productive conversation about it takes the fun out of doing this each week.

    All I’m trying to say is in my opinion removing DNFs does us no good for DFS purposes. I also don’t fit my own purposes do correlations across multitude of races, each race has its own personality. This is my logic, If Kyle Busch has the best speed in practice and drives away from the field in the first 50 laps but then wrecks in traffic, why would we remove that from a correlation (for DFS purposes, not passing an exam in statistics)? I understand he didn’t run all the laps, i get it, so his total is skewed, and it will alter the correlation, but I WANT that for DFS value at the end of the race. If he was fastest and didn’t give me DFS value I want that to be reflected in MY metric.

    I’m not trying to be a statistician I’m trying to win money at DFS. And if that means I have to be creative by looking for patterns such as the stat I throughout there about practice then so be it. It has helped me… and may help others. I think it’s a good thing for people in here to hear these different perspectives and not be over analytical at all times. To me, that correlation serves a purpose, for that race practice meant something.

    I agree and again my apologies if I was offensive … in any way, shape or form.

    In your Kyle Busch example, the best approach would be not to kick-him out of your sample altogether, but to only count the laps Kyle Busch runs in a race, and no more. By using per lap data instead of full race data it would be simple to normalize …. The problem here of course is of Kyle leads the first 50 of 50 laps and then crashes, it would normalize to leading the all 500 laps in a 500 lap race. Combined with the other normalized data this would create a race where there are 500 fast laps possible (actually less, because fast laps and not counted during caution … but laps led are … so this is hypothetical) but more than 500 fast laps for that portion of your sample involved in that race … but there are ways to get around this too ….

    EDIT: Ah-hah !!! You could normalize to count only the laps for all drivers where an actual fastest lap was recorded !!! Know we would need to figure out how to deal with the drivers who had a final practice speed but recorded no fastest lap during the race …. Uggh … my brain hurts !!!

    Since you seem to have a small sample size do you want all the data for Bristol for the 29 races between 2005 and current? That way you could play with it so more ….

    Just let me know and I can PM a link to you. :)

  • Toddalan2286

    I’ll absolutely take it. I just started building my database from 2016 and haven’t had time to add anything further back.

    Thanks for the heated debate! I respect what you’ve been telling me through these past few posts, and what you put together in such a short amount of time, so please don’t think that is lost. I also commend your commitment to the rules of data analysis!

  • gje627

    @Toddalan2286 said...

    I understand he didn’t run all the laps, i get it, so his total is skewed, and it will alter the correlation, but I WANT that for DFS value at the end of the race. If he was fastest and didn’t give me DFS value I want that to be reflected in MY metric.

    His data wouldn’t be skewed because for one race where he raced only 50 laps it would be 100% accurate. Now the data for all drivers in that would be likely skewed (without normalizing/standardizing). If the normalized/standardized data is still skewed (skewed meaning that the data distribution doesn’t represent a normal distribution) a log transformation is generally the desired methodology to use. Using completely raw data, not sure in this case it would work either.

    As far as wanting to use the data for DFS purposes, go for it !!!

    Remember, data is NEVER, EVER perfect. The important part is to identify and quantify a measure of potential error (where possible), and then determine how much risk (uncertainty) you are willing to accept in analysis. In brain tumor surgery, a low error rate may be necessary but for DFS purposes a higher error must be accepted. This is why all the data freaks don’t win every week. Also, there’s something to be said for Qualitative Research Methods, as opposed to strictly Quantitative Research Methods … For example, I’m a big race script guy, and I pretty much think I have a good handle on most races …. Back to our discussion last week about “touts”, as good as I think I am with race script …. by far, the best at this is in my opinion is stevietpfl. I don’t subscribe to his package anymore, but when I did I always wished he would do even more race script/qualitative analysis as nobody does it better … in my humble opinion anyways.

  • gje627

    @Toddalan2286 said...

    I’ll absolutely take it. I just started building my database from 2016 and haven’t had time to add anything further back.

    Thanks for the heated debate! I respect what you’ve been telling me through these past few posts, and what you put together in such a short amount of time, so please don’t think that is lost. I also commend your commitment to the rules of data analysis!

    :)

    I’ll PM you a link to my data in the next 30 minutes.

    Stay tuned !!!

  • Toddalan2286

    Some interesting pricing this weekend, some that caught my attention below:

    Jimmie Johnson: $7,000 (yeah he’s not had good results lately but $7k at Bristol… this is probably the lowest he’s ever been priced) look for him to be chalky.

    Matt DiBenedetto: $6,700 (thought for sure he’d get priced up here).

    Ricky Stenhouse Jr: $6,600 (same comment as MDB)

    Clint Bowyer: $8,200 (seems like he could provide good FPPK value if he can finish the race, good track for him)

    Paul Menard: $6,200 ( when priced below $7k this year I believe Paul has the best point ranking average out of sub $7k priced drivers, don’t think he’s been optimal yet this season but Priced this low at a track where experience matters he could be a great value. He’s $200 cheaper than Hemric.

  • gje627

    @gje627 said...

    Per previous message, I’m going to run Spearman’s on final practice with fast laps. I will not use 10-lap numbers because I only have a couple of years of this data, therefore my results would not be statistically valid (wish I had 10 lap averages for all years). I’m using final practice numbers because since 2007, the second Bristol has always had only 2 practice sessions (the first race is always 3 practice sessions).

    Hi Everyone ….

    Not to reply to my own post, but ….

    I was informed that per RG posting rules they don’t allow Dropbox links in the forums, so the link was removed.

    Once I read the message from RG staff and their explanation as to why they do not allow Dropbox links (potential malicious content being the major reason), I completely understood and 100% support RG’s policy.

    As punishment, I independently took it upon myself to bang my head against the wall until I became even more dizzy than normal !!! :(

    Sorry, everyone.

    EDIT: Ooops. Replied to my own post but it was the wrong one !!!. The post I intended to reply to was the one about DK and FD Base Points Grid. Nevertheless, this RG policy will also prohibit me from posting any links to other data or analysis …. Again sorry.

  • starstx

    Gje … I believe I recall last year you mentioned once that you do look or pay attention to practices, in particular, the last practice … I get that you are saying you are careful or do not like using practice for analytical purposes … I am just wondering if you do use it in some way or factor it in or adjust based on it or if it is more of just hey … a couple of these drivers have fast cars or faster than I expected, log it in the memory bank and kind of move on … just curious … it has seemed to me at least for the top guys that ran a lot of laps etc. Been fairly reliable just not something I could use for every driver because at some point the data falls off depending on whatever the particular driver was trying to do in practice … not everyone is running consecutive laps hard on the throttle and sometimes that is the tough part to decipher but usually looking at the speeds you can kind of tell which ones you may want to toss as not running full speed for the lap averages.

    On a seperate note … with the new package is any previous data really that reliable beyond the last spring race. This is the part I have had trouble with … if this new package has had a drastic effect on racing overall … is all the data previous to this not really very good data. Or is the effect more significant depend on the track or track type. At any rate … Bristol looks like a fun track for DFS … a ton of laps led that can come from a lot of places regardless of starting position. But lookinh at previous races it looks like some races a lot of the top 10 finished fairly close to where they started while in other races there have been a whole bunch of heavy PD movers in the teens and 20’s. Roster construction seems less cookie cutter and more spread out with a high variable of possibilities that could spread ownership out a little more than usual. My kind of race

  • gje627

    @starstx said...

    Gje … I believe I recall last year you mentioned once that you do look or pay attention to practices, in particular, the last practice … I get that you are saying you are careful or do not like using practice for analytical purposes …

    Hey Stars,

    To keep this post shorter, I only quoted a short section but I will reply briefly to all. The whole practice mess is my problem/fault. I do in fact, use practice data BUT I weight it a little different. Per the previous back-and-forth with Toddalan, it is really difficult to work with (and I still need to provide a more detailed explanation … hopefully to come soon, not tonight … messy day). Regarding one point, in my quick correlations I used the final practice, whether it be the 2nd or 3rd practice. The reality is, that all final practices are not the same. The big issue here is where the final practice occurs ….

    Does it occur before qualifying? Or after qualifying?

    The most important consideration … to me anyways … regarding the timing of final practice is whether or not drivers are practicing in qualifying trim, race trim, or a little of both. Notwithstanding this, a lot of drivers have different agendas when practicing. If you recall last year, and a little less this year, Kyle Larson was notorious for only running the top line during “race trim” practice sessions where all other drivers were working the low- to middle-lines. In this regard, Larson wasn’t looking for speed; he was looking to see how his car would drive on the high line and then have his crew chief/engineers work to set-up his car to work the high line. Other similar examples are drivers looking for/and adjusting the car for cornering entry or corner exit, and some even work primarily on getting onto pit row/road after a few laps …. In other words there are so many different agendas by 35 to 40 drivers during practice that it is impossible for the data to capture it all.

    Back to how I use the data …. I may be one of the few, if not only, players that relies mostly on simulations (Monte Carlo). Thus once I get the data reasonably (NOT perfectly) and weighted correctly the simulations really take care of building a reliable set of iterations with probabilities attached. From there, I just need to decide how much risk to take. Some drivers are more reliable with certain ranges of finishing positions and fantasy scores (off the top of head Ryan Newman comes to mind here) and some are less reliable/erratic (Kyle Larson, this year … Fer’ Sure !!!).

    ….

    Regarding this year’s package, certain race data is still reliable. Remember, the data we use is since 2005 (why 2005? … because that was the first year NASCAR started keeping “loop data”) … there have been multiple “generations” of cars, I believe 4, 5, 6 since 2005 (but maybe 3 as well). **Further, within each generation there are multiple different set-ups for different races (see this year … aero ducts, no aero ducts, wicker-bill, no wicker-bill …) and 550 HP or 750 HP. Thus, if we only used data where the cars were identical from race-to-race and/or track-to-track, then for a lot of tracks data-wise we would only have samples between 1 and (maybe) 4 races; thus data would be altogether useless for all races. I won’t even start on the different “rules” for different races at the same track, as this is a whole other bag of worms …..

    Did I answer your question Okay?

    I fear I may have rambled-on again … quickly drifting into my own strange world …. (play “Twilight Zone” theme music now …).

    EDIT: I’ve said this before … sadly my fingers are faster than my brain … as such I frequently seem to skip words when I type … sorry about my pea-brain (maybe I should get a doctor’s note before being allowed to post here.).

  • gje627

    @gje627 said...

    To keep this post shorter

    To reply to myself again ….

    HAHA !!!

    What a joke !!!

  • gje627

    … Last one tonight ….

    Per a previous post, I can’t post links using my Dropbox, but it appears I can using Google Drive. Thus, probably tomorrow, I will re-post the Base points grids using Google Drive, and also other “stuff” sometime later.

  • starstx

    Thanks gje … you answered quite well some of the things I have wondered about. I only started playing regularly last year so this helps me at least get a little better understanding and hopefully will help me work out a little better process in approaching these races. Every time I think I am making some progress I will have a complete miss and afterward see things I should have caught or taken into account better for that particular race. Whether that be laps led, PD or good lord a Daytona race. Nascar DFS is a different beast for sure with the way each track completely changes the Fpts potential landscape in different ways. Thanks for your input.

  • ajs1281

    @Toddalan2286 said...

    Some interesting pricing this weekend, some that caught my attention below:

    Jimmie Johnson: $7,000 (yeah he’s not had good results lately but $7k at Bristol… this is probably the lowest he’s ever been priced) look for him to be chalky.

    Matt DiBenedetto: $6,700 (thought for sure he’d get priced up here).

    Ricky Stenhouse Jr: $6,600 (same comment as MDB)

    Clint Bowyer: $8,200 (seems like he could provide good FPPK value if he can finish the race, good track for him)

    Paul Menard: $6,200 ( when priced below $7k this year I believe Paul has the best point ranking average out of sub $7k priced drivers, don’t think he’s been optimal yet this season but Priced this low at a track where experience matters he could be a great value. He’s $200 cheaper than Hemric.

    Yeah the pricing this week is really interesting and seems a little soft. They certainly aren’t punishing you for taking Kyle Busch this week as much as I had hoped with the quality of guys you can get in the 6-7k range. Should make for some interesting builds. Knowing you can take Kyle and Kyle and still get Jimmie Johnson as your average driver could to lead to some crazy high ownerships depending on qualifying and inspection

  • depalma13

    Earnhardt Jr and wife survive plane crash today.

  • gje627

    Okay ….

    The link I posted yesterday from my Dropbox account has now been posted through my NASCAR Google Drive account, per RG policy. Please find it in the header, under ‘Data and Analytics Links”.

    Under the same link I also posted (not blue … not live) a few other things I plan to post between tomorrow and Saturday morning. I will say while I only list one as TBD/TBA this has been a hellish week and since the race is on Saturday night, rather than Sunday, it will be difficult to get everything done. I will say that if I need to sacrifice anywhere it will likely be with anything related to FanDuel, as I don’t play there and get the feeling few, if any, people here play either.

    Finally, under the section entitled “Race Data, News, and Information” since I deleted the “Rotogrinders Free Tools” I added the link to Stevie’s NASCAR package.

    Besides the “former” plate tracks, Bristol is one of the more difficult races to get a firm grip on. As I mentioned a day or two ago, Stevie’s race breakdown is the probably the best there is, and his package includes a lot more than just that. If you play more than a couple of bucks, it’s worth it especially since you can purchase on a weekly basis.

    Back to the data/analytics ….

    I do have some other things I would like to post (see last placeholder) but no promises. If anyone here has a request for something specific analytics-wise and assuming I have the data (and the time) I will oblige.

  • starstx

    @Toddalan2286 said...

    Some interesting pricing this weekend, some that caught my attention below:

    Jimmie Johnson: $7,000 (yeah he’s not had good results lately but $7k at Bristol… this is probably the lowest he’s ever been priced) look for him to be chalky.

    Matt DiBenedetto: $6,700 (thought for sure he’d get priced up here).

    Ricky Stenhouse Jr: $6,600 (same comment as MDB)

    Clint Bowyer: $8,200 (seems like he could provide good FPPK value if he can finish the race, good track for him)

    Paul Menard: $6,200 ( when priced below $7k this year I believe Paul has the best point ranking average out of sub $7k priced drivers, don’t think he’s been optimal yet this season but Priced this low at a track where experience matters he could be a great value. He’s $200 cheaper than Hemric.

    I had my first real run through looking at this race and this looks pretty spot on. Give me Suarez again at 7200 this week. Disclaimer. I suck at Nascar DFS and you probably should disregard anything I post.

  • gje627

    If anyone’s interested ….

    Preliminary Projections posted in header (91% complete). Since we don’t have Qualifying or final Start Positions post-inspection, DK and FD fantasy points have not, and can not be calculated so only projecting Finishing Positions calculated.

    There are three different projections:

    1) Pre-Race
    2) After Practice 1 (P1)
    3) After Practice 2 (P2)

    In other words, projections updated as you go right in the Excel file.

    Still remaining are:

    4) Qualifying
    5) Start Positions (included regardless of whether or not start positions change after tech … combined Qualifying and Start Positions comprise 9% of my model, though each are weighted a little different)

    Finishing Position numbers are discrete; thus I also included my Driver Rating for each driver both Pre-Race and after each Practice Session. The Driver Rating numbers are the composite numbers of my actual projection system. I include this … again, since Finishing Position numbers are discrete, while Driver Rating is not; meaning when comparing 2 drivers … one projected to finish 8th and another 9th, you can more accurately see the difference between what the model expects of the 2 drivers. In this regard, a large separation between Driver Ratings suggests that my model believes that even though drivers are expected to finish one immediately after the other, the model indicates the higher projected finishing driver is projected to be substantially better. Likewise, where there is little separation between drivers, the model suggests that the finishing positions are virtually interchangeable.

    Driver Ratings can range between 0.000 and 100.000, where 100.00 is best.

    I have the practice data for both practice sessions. I would prefer to wait until after qualifying to post (and then I will update with final start positions if I have time after tech tomorrow).

    Also … I intended to mention this earlier … if anyone else here has data/analytics/projections/information they would like to post, I would be happy to include whatever in the header. Of course, you can also post as a post, but sometimes posts get lost in the shuffle.

    Let me know if you have questions.

    EDIT: Monte Carlo simulations run after starting field is set (post inspection). Primarily, this will give me low-end/high-end finish position projections at a specified probability (range of likely finish position, dominator points). Since tech pushes the envelope on the race itself, may not have time to add to share.

  • gje627

    Start Position Analysis link posted in header.

    DK only.

  • gje627

    Minimum projected DK and FD fantasy points needed to cash and win a GPP link posted in header.

    Large variance given dominator points at Bristol.

    Use in conjunction with Base Fantasy Points Grid; though the Grid will be less helpful here … again since dominator points are so important, especially on DK.

    EDIT: Two more posts tonight, one new … one edit … New post will be Practice through Qualifying. Revised post will be Projections after qualifying … and then I’m done for the night so if you have questions it will need to wait until tomorrow. Also, additional “non-blue” data/analytics will be posted tomorrow.

    Bet everyone is glad this is only a one-week cameo for me !!! :(

  • Moszman

    @gje627 said...

    Minimum projected DK and FD fantasy points needed to cash and win a GPP link posted in header.

    Large variance given dominator points at Bristol.

    Use in conjunction with Base Fantasy Points Grid; though the Grid will be less helpful here … again since dominator points are so important, especially on DK.

    EDIT: Two more posts tonight, one new … one edit … New post will be Practice through Qualifying. Revised post will be Projections after qualifying … and then I’m done for the night so if you have questions it will need to wait until tomorrow. Also, additional “non-blue” data/analytics will be posted tomorrow.

    Bet everyone is glad this is only a one-week cameo for me !!! :(

    Nope, have enjoyed .

  • gje627

    Practice through Qualifying speeds and ranks added to header. File includes 10-lap averages.

    Also included car numbers if anyone calculates their own revised starting positions based on tech. (if needed).

    I’m going to update the projections, then I will be done for tonight.

  • gje627

    @Moszman said...

    Nope, have enjoyed .

    Thank You. :)

  • X Unread Thread
  • X Thread with New Replies*
  • *Jumps to your first unread reply

Subforum Index

RotoGrinders.com is the home of the daily fantasy sports community. Our content, rankings, member blogs, promotions and forum discussion all cater to the players that like to create a new fantasy team every day of the week.

If you or someone you know has a gambling problem, crisis counseling and referral services can be accessed by calling 1-800-GAMBLER (1-800-426-2537) (IL). Gambling problem? Call 1-800-GAMBLER (NJ/WV/PA/MI), 1-800-9-WITH-IT (IN), 1-800-522-4700 (CO), 1-800-BETS OFF (IA), 1-888-532-3500 (VA) or call/text TN REDLINE 1-800-889-9789 (TN).