Value Grouping - by rookBoston

Discussion in 'Patriots Draft Talk' started by Box_O_Rocks, Mar 4, 2007.

Thread Status:
Not open for further replies.
  1. Box_O_Rocks

    Box_O_Rocks Supporter Supporter

    It's time to once again bring forth this edifice to man and womankind's efforts to comprehend the mystical Belipioli drafting process. It's a bit tattered by time and new draft data, but an entertaining read and a symbol of rook's greatness as an amateur draft warrior. MM was the custodian, but by now several of us have it secreted away that it not die in DOS attacks, but remain safe for future generations. Enjoy!

    Value Grouping

    Value Grouping - by rookBoston
    Submitted By: Mark Morse
    April 20, 2004

    Value Grouping Theory

    This was a message by rookBoston who posts on the PatsFans Draft Forum Messageboard. I was so impressed with this post that I thought I would put it out here in the Draft section. I cannot take credit for this since it is someone else’s work, but I like the thinking behind it and I think you will find it interesting as well.
    I wanted to post something about "Value Grouping", which is the term I use to rationalize BB's draft day strategy. I posted something about this immediately after the 2003 draft.
    That draft was incredible, but it left me totally confused. I couldn't see any logic in the trading up and down. The strategy seemed haphazard, and yet it produced very reliable results. It could not have been blind luck that they filled all their positions of need at ideal value spots in the draft. How did they do it?
    I don’t know if this is something that is widely discussed in the Nation, or if it's just a pet theory of my own. I haven't read anything about it, or if I have, I've forgotten that I did. I think BB has probably explained the concept behind his draft strategy to the media, although Tom Curran's recent article makes no mention of this type of targeting. I’m not sure how explicit BB’s been in describing how it works.
    I thought, I’d post my understanding of how the Pats approach the draft, because so far as I know it is clearly not an intuitive or typical strategy.
    Here’s how it works:
    BB, SP and the scouting team identify a group of players who would provide roughly the same value for the team. It is absolutely critical that the staff “believes” in all the players in the value group, without exception. All of them have to be guys that the staff would feel good about drafting and would want to have on the team.

    Once a player makes it into a value group, that means he is being targeted for the draft. Before the draft, the staff identify what their key needs are, and decide how they plan on filling those needs, using the value groupings they’ve defined. Generally speaking, it is possible to say, “this group of players should go in the early 1st round, this group of players should go in the late 2nd, early 3rd.”

    On draft day, as the players in a targeted value group start coming off the board, the Pats lie in wait, patiently. The first players coming off the board for a given value group set the baseline value for that group. Once a value group is down to one or two players remaining, they trade into position to take the last guy in that group.

    The downside is that by definition, you are taking the guy in each Value Group which the other teams think is the worst of the bunch. But if you’ve setup the board correctly, it shouldn’t matter. It is your own evaluation that counts, not the rest of the teams.
    The upside is, you get good players for rock bottom prices. Most important, you eliminate (or at least dramatically reduce) the risk that you come away with absolutely nothing. Examples will help explain how this is possible, in practice.
    Examples #1 - Wilson
    The clearest example of this strategy in action was the drafting of Eugene Wilson in 2003. Belichick had said that one of his offseason goals was to get younger and faster in the secondary. Clearly, that meant an early round selection at CB. No surprises.
    At #19, there were a bunch of very good CBs available: Andre Woolfolk, Charles Tillman, Eugene Wilson, Sammy Davis, Nnamdi Asomugha. It’s hard for me to say which of these players were in the Pats draft board, but I would bet that Woolfolk, Tillman, Wilson and Davis were all in the same value grouping.

    At 19, Belichick had his choice of any of them. It would not have been a reach to take Woolfolk at that point, according to the scouts. Since the draft, so far as I hear, Woolfolk has established himself as a solid young player in Tennessee. But, instead, Belichick traded down to #41 and watched as Woolfolk, Davis, Asomugha all came off the board in the first round. Then, at the top of the 2nd, as soon as Tillman comes off the board at #35, Belichick promptly trades back up to take Wilson at #36 (from #41).

    Note: Give Mel Kiper his credit, he was the only one who had Wilson rated as a 1st round pick.

    After Wilson, only one CB was taken in the draft over the next 46 selections. That proves there was a significant dropoff in talent after Wilson. The question is whether the dropoff in talent from Woolfolk to Wilson is at all significant. The experience of the 2003 season suggests that all four CBs in that value group would have been solid contributors for the Patriots.

    You have to have a lot of faith in your scouts, if you’re going to apply this kind of draft philosophy. But, when it’s working, it can create a serious advantage, by delivering players who can help the team immediately, at spots in the draft where their value is maximized.
    Example #2 - Warren
    One of my biggest question marks after the 2003 draft was, “Why bother trading up *one draft spot* to take Warren”? After all, Warren was probably going to drop to #14 anyway. Most mock drafts had him going at the top of the 2nd round. BB could have (in all likelihood) saved himself an additional Day Two selection. For a staff which hoards draft picks like water in the desert, it seemed completely out of character. But, if you consider the situation from the perspective of “value grouping” the trade-up makes much more sense. In fact, the example is illuminating.

    In this case, I think the value group included a bunch of interior defensive linemen: Dewayne Robertson, Ty Warren, Jonathan Sullivan, Jimmy Kennedy... and possibly Kevin Williams (although he wouldn’t have been a great fit in the 3-4). The Pats interest in defensive lineman was obvious, and their specific interest in Robertson and Sullivan was also well publicized. (Possibly, the value group contained other players, like Terrell Suggs. But Suggs fits a very different draft profile, so I think it’s unlikely.)
    Of course, the Jets and Saints had read the reports that the Pats were interested in those guys, and traded up to make sure to get the ones they wanted. Belichick played it cool.
    Jimmy Kennedy dropped all the way to #12 overall. On draft day, I was shocked that Belichick didn’t trade up for Kennedy after he dropped past #10, because the scouting reports described him as an ideal NT for a 3-4. In retrospect, I believe the reason he didn’t need to trade up for Kennedy was because Ty Warren was still on the board. So, that draft group still had two players available. Until one of them came off the board, there was no need to panic. BB still expected to get a guy he wanted just by sitting still.
    Of course, Kennedy went off the board at #12 with the Pats waiting at #14. According to the draft philosophy, in this situation you ABSOLUTELY MUST trade up to take Warren at #13. If someone steals Warren away from you at #13, not only have you missed out on Warren, but you’ve missed out on the entire value group. The risk is way, way too high. If Kennedy had not been selected at #12, BB would gladly have stayed at #14, because he would have been certain to get one of the two. But with Kennedy gone, trading up became a priority.

    It’s interesting to note that William Joseph was clearly not included in this value group, even though popular scouting reports had him in the competition for a top-20 selection. The Pats passed on him (ignored him, really) and he lasted until #25.

    It’s also interesting that Warren made it into the value group at all. Robertson, Sullivan and Kennedy were all highly touted candidates, but Warren was widely considered a 2nd rounder until just prior to the draft. If the Pats scouting on Warren had been less positive, if (say) he had been grouped with Tyler Brayton and William Joseph instead of the elite group, then according to the theory, the Pats would have been obligated to trade up for Kennedy at #7 (easily within striking distance) immediately after Sullivan was selected at #6.

    After Warren, only two additional DTs were selected on Day One: Joseph at #25 and Anthony Adams at #57. Again, it seems clear that the drop-off in talent after this value group was pretty considerable.

  2. Box_O_Rocks

    Box_O_Rocks Supporter Supporter


    What it means for this year’s (2004) draft:

    Personally, I think this year’s "value group" at RB includes Jackson, Jones AND Perry. If that’s true, according to the draft philosophy, we should expect BB to wait until two of the three are off the board before he drafts his “franchise RB”. I’m not sure whether Greg Jones should be rated in that grouping or not. I’m leaning to “No”, because of his limited history of producing in the passing game.

    So, if Jackson goes to Denver at #17 and both Jones and Perry are still on the board at #21, BB may trade back a bit. The hook is that Dallas will probably pick a RB at #22, so if we trade back, it would only be one or two spots back. That may not be worth it.

    Alternatively, if Dallas trades into the early teens for Jackson and then Denver takes Dansby, Jones and Perry may both fall, together, all the way to the end of the 1st. Maybe as far as the Eagles at #28. If so, the Pats would grab the other RB at #29 in an easy trade up from #32.

    At WR, the elite value grouping is Fitzgerald, Roy Williams and Mike Williams. This group is probably out of range. After that, there is a drop-off from that threesome to the value group with Woods, Clayton, Jenkins, Evans and Reggie Williams. For simplicity, it may be easiest to group all eight of them into one single huge group. Maybe drop Evans out of the group, because he doesn’t have the size we want, and maybe drop Reggie because of the ego. But I like the rest of the group. I think this group will start being drafted in the early 1st, but should still have one of two guys in the 30s.

    At DB, Sean Taylor stands on his own. I don’t think you can wrap a value group around him, because he doesn’t have anyone like him in the entire draft class. After him, I think we can group Strait, Gamble, Carroll into a group. Depending on if you like them, the group may also include Shawntae Spencer, Dunta Robinson and Will Poole. Adding Deangelo Hall is pointless, because he will almost certainly be the first off the board. The group could also include S prospects, like Sean Jones, Sanders and Ware, depending on how you want to setup your board. I think this value group will start losing players in the mid-1st, but some of them should last into the early 2nd.
    My Value Groups in the 1st round for this draft are:
    Group 1: ( Fitzgerald, S. Taylor, Roy Williams )
    Group 2: ( Jackson, K. Jones, Perry )
    Group 3: ( Fitzgerald, Williams, Williams, Clayton, Woods, Jenkins )
    Group 4: ( Strait, Gamble, Carroll, Sanders, Jones )
    Group 5: ( Grove, Snee, Smiley, Carey )

    If the Group 1 still has players available into the #10-15 range, I think BB will move up to cash in. This group is top-5 talent and includes players who would be productive for us immediately. That trade would cost us players from two of the remaining value groups, but I think it would be worth it.

    Otherwise, I think the Pats have enough draft value to get one from each of the four remaining groups. The only way we could get a player from all five groupings would involve trading Law.

    Based on the groupings above, a fair draft prediction would be-

    Group 2: RB Perry at #29 (675),
    Group 3: WR Jenkins at #37 (540),
    Group 4: OG Snee at #45 (460),
    Group 5: S Sanders at #50 (410).

    Draft value: 2085
    By the book, our current picks are worth?
    #21 (875) #32 (600) #56 (350) #63 (275)
    Draft value: 2100
    - so in concept this is not completely unrealistic. It’s really a matter of whether we can find trading partners. Almost certainly, it will cost us later round selections, to attract trading partners.
    This is just a theory based on observation and speculation. It helps explain a number of strange things, like trading up for Warren, and trading down then up for Wilson.

    The Klecko pick followed the same pattern. Dan was taken at #117. The implication is that NT Ian Scott (taken #116) and NT Nick Eason (taken #114) triggered the Klecko pick.

    Bethel Johnson was picked at #45, in a trade up from #50. Not coincidentally, Taylor Jacobs was selected at #44. Jacobs was a college sprinter and a high character guy. Fits the Pats profile to a T, and sounds very similar in skills to BJ. Of course, the implication is that the Pats wanted one of the two of them, and waited till Jacobs was selected before moving up for Johnson.

    Tom Curran’s recent article on how BB stacks his draft board makes little mention of this type of strategy. That discussion focused on vertical (by position) and horizontal (across position) stacking. The closest BB comes to describing this process in the interview is when he says: “Or sometimes you look and say, 'This is the last tackle on the board for a long time. We have linebackers rated higher, but there are more of them.' So you need to take the tackle. That's just draft strategy.”

    Curran's article makes me thing that Pats FO may not think about the draft in the same formal terms that I’m using. But in the end, the value groupings that I’m describing are simply the product of the vertical and horizontal stacking that Belichick described. When you have a bunch of CBs prioritized vertically and they bunch up closely together after you do the horizontal stacking, you come away with what I’m describing as a value group. Even if you don’t use the term, or even think of it in that way. The key, then, is to hit the vertical need at a point where the horizontal value is optimized
  3. Mike the Brit

    Mike the Brit Minuteman Target Supporter

    Disable Jersey

    It has to be said that rookBoston himself believes that the theory has been completely contradicted by the Patriots' subsequent drafts. I have to agree. Still, it was a wonderful idea. At the time, it prompted me to try and think through rational draft strategy. The post below was the fruit of my reflections. I post it again in case it might be of interest.

    Rational Draft Strategy

    I threatened to produce an alternative to the theory of value grouping so here’s a first go at thinking about draft strategy from the ground up. The idea is to do it with maximum economic rationality. But be warned! I took economics a long time ago and it didn’t have much to offer on problems like this.

    1. Introduction
    You can think of the draft as basically a Dutch auction (an auction where prices diminish until someone bids). What makes it distinctive from most of the situations analysed by economists then is two things. First, the “goods” or “assets” being bid for are not homogenous. On the contrary, the participants in the auction try to create value by leveraging their assessments of the differential values of the “assets” on offer. Secondly, since it’s an auction, the participants have to bid without the price being known. (Note: this isn’t quite true but let’s leave it there for the moment; keep the complexities for later).

    The participants in the auction have picks, it is true, not money. But that difference is less significant than one might think because trading is allowed. Indeed, there is a fairly well agreed table of value for the trading of picks. So each team can be seen as starting the auction with so many “value points” which are like an agreed common currency.

    2. A Simple Starting Point

    Let’s imagine that there’s no trading and every team has to simply pick at its allocated slot. In that case the following strategy suffices. Put all of those players you’re interested in drafting into an order of desirability and, when it comes to your turn, take the highest available player.

    (This strategy might not be optimal for the following reason. Imagine that you want to come up with a draft that includes an OT. But in Round 1 you have a CB rated above the best available OT. Yet there is no OT of any quality available after Round 1 although there are decent CBs available. It might be that taking the CB in Round 1 won’t be your best strategy. But I think that this is a pretty nit-picking objection and the simple strategy is as good as you could wish for. The chaining of choices (externalities, in economists’ language) makes everything very much more complicated and this is already quite complex.)

    3. Why the Simple Strategy Isn’t Enough

    While there is no trading the question of how good value a player is at any position simply doesn’t arise: you’ve got the pick, use it or lose it, so it doesn’t matter whether your best player is “worth” a second-round pick (or whatever). Once there’s trading, however, the whole business changes. So here’s another strategy.

    Allocate a value to everyone that you’re interested in drafting such that the numbers reflect your point of indifference. (If player X is worth 500 and players Y and Z 250 each then that means that you don’t mind whether you end up with just X or both of Y and Z – you get the idea, I hope.)

    Once you’ve done that then your problem is to maximise the value of your picks subject to your budget constraint (the value points you started with).

    Now, if you don’t trade, you’ll just do, in effect, what you did under the very simple strategy: pick the player with the very highest value at any point. But, once there is trading, another question may enter your cunning little mind: OK, you may say, I could get Wilson with my number 20 (or whatever) pick and, from what I know, he’ll turn out much better than some more celebrated guys. But he’d be even better value if I got him at 52.

    So as well as how much the player is worth to you, you need to have an idea of how well he relates to the pick. If you could get a player you’ve rated at 800 for a 600 pick (number 32) it might seem a good deal. But what if you could get him at 41 (pick value 500)? Or trade down to 61 and 62 (pick value = 590) and get two players worth 500 and 450? Good business.

    So you try to trade in such a way as to increase the differential value of the picks you make (worth of the players you obtain net of the cost of pick). Of course, your problem is, in trading up, you don’t know who WOULD have been available had you stood pat, and, in trading down, you don’t know who will be available when it gets round to the new picks. If you only knew in advance who would pick whom and when then the whole business would be easy. You could hold off till the last moment. So, perhaps, instead of paying 600 for your 800-rated pick you could trade down to 41 and get him for 500. Clearly, intelligence like that is like gold dust. But you don’t have it, so what do you do?

    4. Evaluating the Worth of Picks

    The rational strategy, it seems to me, is to form a hypothesis about where each player you’re interested in will be taken. Be careful, of course, not to base it on what’s public (mock drafts, the press, etc.); other teams get paid too (as BB says) and you can’t be sure that somebody else won’t have spotted your sleeper. Still, if you’re not risk-averse, you can do well by not jumping in too soon. In fact, if you’re economically rational, you should hold off until one above your best guess of where a player will be taken.

    5. More?

    I think that this is the outline of a rational strategy and this is already a long post, but before I quit I want to make one more point which ties us back to the theory of value grouping.

    I said at the beginning that you don’t know prices. But that’s not strictly true. At each round you know that every player will cost (say) “less than 600 pick value points”. Your guess about what the price will be for each player will be based (presumably) on what you can discover about other teams’ scouting/needs/prejudices, etc. But, as the draft goes on, picks made may give you some indication. In particular, you can monitor how fast comparable players at the position of the player you’re interested in are going. So, if all the cornerbacks that you think that other teams will rate above the cornerback you yourself are interested in have gone (note: NOT all the cornerbacks that you yourself rate above the one you’re interested it) then you know that he is in the firing line for the next team that wants a cornerback. If you give a list of comparable/slightly superior-rated players for all of your targets, then you can modify your expectations of when your targets will be taken in the light of the data as it emerges in the course of the draft. That data then generates “trigger picks” of the sort that we saw with Warren and Wilson.

    All clear then? No, I didn’t think so.
  4. Fencer

    Fencer Veteran Starter w/Big Long Term Deal

    #12 Jersey

    The nice thing about that story is that it neatly explains why, when the Pats do trade up, they commonly overpay by trade-value-chart standards.

    In essence, they've giving back SOME of the value they get by bargain shopping.
  5. Box_O_Rocks

    Box_O_Rocks Supporter Supporter

    I thought we were trading Mike for a Swede and some magic beans? I wish we'd get it done, he's given me a headache again!
  6. rookBoston

    rookBoston 2nd Team Getting Their First Start

    #12 Jersey

    Thanks for digging that up BOR. As Mike says, BB's behavior in the drafts since 2003 do not follow the pattern predicted by Value Grouping. Value Grouping will result in a team trading up and down on draft day, picking specific, targeted players at specific points of value.

    2003 was a special year. To start, it wasn't a very good draft class, overall, so BB/SP were picking diamonds out of the trash. We had a lot of draft picks, thanks to the Bledsoe and Glenn trades. The team was slowfooted and aging, and had a lot of needs. The franchise was using the draft to retool the roster. Come 2007, the facts have changed.

    Still, I think the general philosophy is sound. There are a number of tactical factors that need to align to make the strategy operational:

    1) Need to have a lot of draft picks, and a lot of tradable picks. Recently our extra picks have been Compensatory, which in an inhibitor.

    2) Need to have trading partners. BB tried to trade up to #22 for Mark Clayton in 2005. He tried to trade up to #18 for Will Smith in 2004. In both cases, Baltimore and New Orleans listened to the offer, but chose to make the picks themselves. And, it takes trading partners to trade down, too. Even before drafting Vince Wilfork, the Pats ran the clock down very low before sending the card up. Same with Moroney. I'm sure they were listening to offers, but there needs to be equivalent value in exchange.

    3) Need to have a lot of needs. If you have 9 draft picks but only roster space for 5 rookies, you can afford to waste some draft value taking a guy who you like just a little better, instead of waiting for the Value Group to run down to zero. Instead of trading back, then trading up, just pick the guy you like best! That's a luxury that the team has earned itself by building such deep talent.

    This year, we have enough picks. But, even with all the concern about LB, DB and WR, we dont have a lot of needs. Our Day Two picks will be fighting just to make the Practice Squad.

    As for finding trading partners, who wants to be the GM on the other end of a trade with a football genius? How does the Kyle Boller trade (for what became Wilson and Wilfork) look now to Ravens fans? How does the Bledsoe trade with Buffalo look now? How does the Tebucky Jones trade with the Saints look now? It's still early, but people will be second guessing the Deon Branch trade in Seattle, too. Good trades are win-win, but those trades all look damn lopsided to me. And if you're a GM, bad trades end careers.

    What do I expect this year? I think BB will pick a player at 24 or a slight trade up for value. I'm not sure whether he'll pick a player at 28 or not. I'd say, even odds-- depends on who's available. I actually think there will be very good value in the 40s and 50s based on the mocks I've seen recently. No question BB will listen to trade offers.
  7. Seneschal2

    Seneschal2 In the Starting Line-Up

    I do think it's your own theory. If you eliminated the words "value grouping", it may be less confusing. Understanding the horizontal/ vertical rankings as explained by Belichick says it all. I wouldn't look much further.

    You lost me with this. They simply scout the players, find the ones who can play in their system, and rank them. "Provide the same value" is confusing.

    Well, they have their own draft board. They also have a mock draft board -- if this is what you mean.

    What do you mean by "targeted value group"? They sometimes target specific players -- but a whole group?

    I suppose I don't understand this at all because you lost me with the previous paragraph.

    They were ranked, like all other positions and most likely each had a different value attached based on the ranking system.

    But you don't even know if he fits the Pats system. Doesn't matter what other scouts say.

    Absolutely not. I had Wilson as a potential first-rounder, and because Kiper isn't my main source, there must have been other 'experts' who had Wilson ranked in round one (or I wouldn't have him ranked there). Oh, I just remembered an expert - Rick Gosselin. No one is more respected than him, and he mocked Wilson at #21 to Cleveland.

    Pure speculation.

    No, one of the more accurate mockers around had him going to the Pats at #14 - Gosselin.

    Here's the way that unfolded. As you stated, Warren was the lone remaining quality DL available. But what you haven't noted is this; The Pats have a 'team needs' board, which is separate from their main draft board. By anticipating the needs of other teams, they know (or guess) which position teams may be seeking to fill. SD, at #15, was in need of one of the DL. The Pats knowing this, trading up themselves from #14 to #13 to prevent SD from doing so. With none of their targeted DL left on the board, SD ended up trading out of that #15 slot.

    As I stated, he was mocked at #14, which is a long way from round two.

    Finally I understand your definition of "value grouping". Wish I did earlier. But I'm still wondering what the major diference is from the vertical/horizontal stacking (which I know very well) -- and why the "value grouping" label?
  8. psychoPat

    psychoPat Role Player Supporter

    Very enjoyable deja vu! Feels like attending a high-school reunion.

    When rook's article first appeared in this draft forum, naturally the cognoscenti here nodded sagely.
    To draw to it the wider attention that the theory deserved, i took it upon myself
    to open a thread on the main board
    - rather, my ancestor, plunkett2vataha, did -
    to serve as a pointer to the draft forum thread.

    I pat myself on the back for having thereby contributed in small measure
    to the lasting acclaim that Value Grouping subsequently earned.
  9. Fencer

    Fencer Veteran Starter w/Big Long Term Deal

    #12 Jersey

    I see one controversial claim in this -- that the Pats' draft board clusters into groups of similarly valuable players.

    Offhand, I don't know how we'd test that one way or the other ...
Thread Status:
Not open for further replies.

Share This Page